A Fourier-Mukai approach to spectral data for instantons

合集下载

EG Test 4 Passage 1 阅读译文

EG Test 4 Passage 1 阅读译文

'Sleep comes more easily than it returns.'—Victor Hugo,Les Misérables入睡容易,醒过来难——维克多.雨果《悲惨世界》A It is estimated that one in three adults in westernised countries regularly wakes up in the middle of the night and has difficulty getting back to sleep.Physicians often diagnose'insomnia'and prescribe sleeping pills,but these often have side effects such as negative interactions with food,drink or other drugs,and most are habit-forming. Cessation of the medication frequently causes unpleasant withdrawal symptoms,too,including panic attacks,mood-swings,and even heightened sleep disturbance.Is there a way to treat insomnia without such debilitating consequences?据估计,在西方国家中,有1/3的成年人经常会在半夜醒来,之后再难继续入睡。

医生经常会作出失眠症的诊断,然后开安眠药。

但是这些药通常是有副作用的,比如说和食物、饮料或其他药物发生不良反应,而且这些药大多数还容易上瘾。

如果停止用药,又会导致一些不良症状的反弹,比如产生恐慌、情绪波动,甚至是更严重的失眠。

THE FOURIER SPECTRAL METHOD

THE FOURIER SPECTRAL METHOD

a given non-zero.
u g H, Let
only
a denumerable be the
subset indices
of of
cothe
efficients
{Jk}kc ~
coefficients; it can be shown that the sequence
{Un}n=l,
where
H
is said to be separable if it admits a denumerable
2.
Simple Examples of Hilbert Bases In what follows the Hilbert space may be real or complex. We denote by B(H) iff T* the set M s ~ of T of all bounded such that is defined by linear operators of H to
expansions)
will give us, in general,
14
3.
Fourier Series in We consider now
L2(-~,w) the complex llilbert space L2(-~,~) with a scalar
p r o d u c t d e f i n e d by (f,g) = ~
be
defined for problesolution of Dirichlet
-u" = f
u(0) = u(~) -- 0.
*That is to say, T transforms bounded sets into relatively compact sets.

研究生涉海英语课后习题答案Unit4

研究生涉海英语课后习题答案Unit4

研究⽣涉海英语课后习题答案Unit4Unit 4 Climate V ariabilityText APart IUnderstanding and LearningBackground information1. Philosophical Transactions: the world's first science journal. In 1662, the newly formed “Royal Society of London for Improving Natural Knowledge” was grante d a charter to publish by King Charles II and on 6 March 1665, the first issue of Philosophical Transactions was published under the visionary editorship of Henry Oldenburg, who was also the Secretary of the Society. The first volumes of the world's first scientific journal were very different from today's journal, but in essence it served the same function,namely to inform the Fellows of the Society and other interested readers of the latest scientific discoveries. As such, Philosophical Transactions established the important principles of scientific priority and peer review, which have become the central foundations of scientific journals ever since. In 1886, the breadth and scope of scientific discovery had increased to such an extent that it became necessary to divide the journal into two, Philosophical TransactionsA and B, covering the physical sciences and the life sciences respectively.2. The Alfred Wegener Institute for Polar and Marine Research carries out research in the Arctic and Antarctic as well as in the high and mid latitude oceans.3. Tambora’s explosion: In 1815,Tambora in Indonesia exploded with the force of roughly 1, 000 megatons of TNT, the largest volcanic eruption in recorded history. Detailed Study of the TextI.Reality has a way of trumping art, and human-induced climate change isvery real indeed.(Para. 2)Meaning: The situation of climate change caused by human beings in reality is more critical than what is described in science fictions.trump:1. v. beat 胜过;压过The Socialists tried to trump this with their slogan.社会党党员设法⽤他们的⼝号把它压过去。

Conformal Field Theory Approach to the Kondo Effect

Conformal Field Theory Approach to the Kondo Effect

a r X i v :c o n d -m a t /9512099v 1 12 D e c 1995Conformal Field Theory Approach to the Kondo Effect ∗Ian AffleckCanadian Institute for Advanced Research and Physics Department,University of British Columbia,Vancouver,BC,V6T 1Z1,CanadaRecently,a new approach,based on boundary conformal field theory,has been applied to a variety of quantum impurity problems in condensed matter and particle physics.A particularly enlightening example is the multi-channel Kondo problem.In this review some earlier approaches to the Kondo problem are discussed,the needed material on boundary conformal field theory is developed and then this new method is applied to the multi-channel Kondo problem.OUTLINEI.Renormalization Group and Fermi Liquid Approaches to the Kondo Effect A)Introduction to The Kondo Effect B)Renormalization Group Approach C)Mapping to a One Dimensional Model D)Fermi Liquid Approach at Low TII.Conformal Field Theory (“Luttinger Liquid”)Techniques:Separation of Charge and Spin De-grees of Freedom,Current Algebra,“Gluing Conditions”,Finite-Size SpectrumIII.Conformal Field Theory Approach to the Kondo Effect:“Completing the Square”A)Leading Irrelevant Operator,Specific Heat,Susceptibility,Wilson Ratio,Resistivity at T >0IV.Introduction to the Multi-Channel Kondo Effect:Underscreening and Overscreening A)Large-k LimitB)Current Algebra Approach V.Boundary Conformal Field TheoryVI.Boundary Conformal Field Theory Results on the Multi-Channel Kondo Effect:A)Fusion and the Finite-Size Spectrum B)Impurity EntropyC)Boundary Green’s Functions:Two-Point Functions,T=0Resistivity D)Four-Point Boundary Green’s Functions,Spin-Density Green’s Function E)Boundary Operator Content and Leading Irrelevant Operator:Specific Heat,Susceptibility,Wilson Ratio,Resistivity at T >0I.RENORMALIZATION GROUP AND FERMI LIQUID APPROACHES TO THE KONDO EFFECTA.Introduction to the Kondo EffectMost mechanisms contributing to the resistivity of metals,ρ(T),give eitherρ(T)decreasing to0, as T→0(phonons or electron-electron interactions),orρ(T)→constant,as T→0(non-magnetricimpurities).However,metals containing magnetic impurities show aρ(T)which increases as T→0.This was explained by Kondo1in1964using a simple Hamiltonian:H= kαψ†α kψ kαǫ(k)+λ S· k k′ψ† k σ+...]2(1.2)THere D is the band-width,νthe density of states.This result stimulated an enormous amount oftheoretical work.As Nozi`e res put it,“Theorists‘diverged’on their own,leaving the experimentrealities way behind”.2What happens at low T,i.e.T∼T K=De−1ψ( 0,t) ,2where thefields are in the interaction picture.EFk F2D’Ek2DFIG.1.Reduction of the cut-offfrom D to D ′.As S(t )is independent of t ,we simply multiply powers of S using [S a ,S b ]=iǫabc S c ,S2=s (s +1).We must time-order S ’s which don’t commute.The first few diagrams are shown in Figure (2).In 2nd order in λ,we have:−λ22ψ(t )ψ†(t ′)σb2λ2dt dt ′ψ†σa2ψT ψ(t )ψ†(t ′) (θ(t −t ′)S a S b +θ(t ′−t )S b S a )=λ22ψ· Ssn(t −t ′) ψ(t )ψ†(t ′) ,(1.3)where sn (t −t ′)is the sign-function which arises from T -ordering spins.FIG.2.Feynman diagrams contributing to renormalization of the Kondo coupling constant to third order.We see that the integraldtǫ(t )G (t )=−idt(2π)3dωiω+δ+1ω−ǫk +iδsn(ǫk )(1.5)=d 3 k|ǫk |≈2νDD ′dǫD ′.(1.6)Thusδλ=νλ2lnDd ln D=−νλ2.(1.8)We see that lowering the band cut-offincreases λor,defining a length-dependent cut-off,l ∼v F /D ,dλ1−νλ0lnD 0νλ0,If λ0<0(ferromag-netic),λeff(D )→0.See Figure(3).effλFIG.3.RG flow of the Kondo coupling.The behaviour at temperature T is determined by λeff(T ):ρ(T )→0as T →0for the ferromag-netic case.What happens for the antiferromagnetic case?C.Mapping to a One-Dimensional ModelThe above discussion can be simplified if we map the model into a one dimensional one.Weassume a spherically symmetricǫ( k),ǫ(k)=k2√2ψ0,k′· S,(1.12)whereν=k2F/2π2v F is the density of states per spin.This can also be written in terms of radialco-ordinate.We eliminate all modes except for a band width2D:|k−k F|<D.Defining left andright movers(incoming and outgoing waves),ΨL,R(r)≡ ∧−∧dke±ikrψ0(k+k F),⇒ψL(0)=ψR(0),(1.13) we haveH0=v FdrψL−ψ†R i d2ψL(0)· S.(1.14)Here we have redefined a dimensionless Kondo coupling,λ→λν.Using the notationψL=ψL(x,τ)=ψL(z=τ+ix),ψR(x,τ)=ψR(z∗=τ−ix),(1.15) whereτis imaginary time and x=r,(and we set v F=1)we haveψL(z)ψ+L(0) =1z∗.(1.16) Alternatively,sinceψL(0,τ)=ψR(0,τ)ψL=ψL(z),ψR=ψR(z∗),(1.17) we may considerψR to be the continuation ofψL to the negative r-axis:ψR(x,τ)≡ψL(−x,τ).(1.18) Now we obtain a relativistic(1+1)dimensionalfield theory(a“chiral”one,containing left-moversonly)interacting with the impurity at x=0withH0=v FdxψL(1.19)and H INT as in Eq.(1.14).See Figure(4).LLL RFIG.4.Reflecting the left-movers to the negative axis.D.Fermi Liquid Approach at Low TWhat is the T →0behavior of the antiferromagetic Kondo model?The simplest assumption is λeff→∞.But what does that really mean?Consider the strong coupling limit of a lattice model,2for convenience,in spatial dimension D =1.(D doesn’t really matter since we can always reduce the model to D =1.)H =ti(ψ†i ψi +1+ψ†i +1ψi )+λ S ·ψ†0σl(n +1/2)λ=∞:k =πnNear the Fermi surface the energies are linearly spaced.Assuming particle-hole symmetry,the Fermi energy lies midway between levels or on a level.[See Figures(5)and(6).]The two situations switch with the phase shift.Wilson’s numerical RG scheme3involves calculating the low-lying spectrum numerically and looking for this shift.This indicates thatλrenormalizes to∞even if it is initially small.However,now we expect the screening to take place over a longer length scaleξ∼v FDe1/νλ.(1.24)In other words,the wave function of the screening electron has this scale.We get low energy Bloch states of free electrons only for|k−k F|<<1/ξ(so we must take l>>ξ).[See Figure(7).]The free electron theory with a phase shift corresponds to a universal stable low energyfixed point for the Kondo problem.This observation determines the T=0resistivity for an array of Kondo impurities at random locations of low density n i.It is the same as for non-magnetic s-wave scatterers with a π/2phase shift at the Fermi energy.δ=π/2gives the so-called unitary limit resistivity:ρu=3n ik-kFconditions.FIG.6.Free fermion energy levels with periodic boundary1k-kFFIG.7.Non-interacting Bloch states with a vanishing boundary condition occur for|k−k F|<<v F/T K.The low-T behaviour,so far,seems trivial.Much of the interesting behaviour comes from the leading irrelevant operator.The impurity spin has disappeared(screened)from the description ofthe low-T physics.However certain interactions between electrons are generated(at the impuritysite only)in the process of eliminating the impurity spin.We can determine these by simply writingthe lowest dimension operators allowed by symmetry.It is simplest to work in the1D formulation,with left-movers only.We write the interaction in terms ofψL,obeying the new boundary condition(but notψL+....(1.26)dxThe length and time dimensions are equivalent(we convert with v F),[H]=E⇒[ψ]=E1The interactions are localδH= iλi O i(x=0),[λi]+[O i]=1.Soλi has negative energy dimension if[O i]>1,implying that it is irrelevant.In RG theory one usually defines a dimensionless coupling constant by multiplying powers of the cut-offD,if[λi]=E−a,˜λi≡λi D a,˜λidecreases as we lower D:d˜λidx ψα(0)−id3v FlT+aT3v FT+ln2.(1.30)At low T,the impurity entropy decreases to0:S(T)=πlT K.(1.31)In general we may write:S(T)−πl2πv F +b2πv F +12πv F +1ln(T/T K)+... .(1.35)In general,we may write:χ−lTf(T/T K),(1.36)where f(T/T K)is another universal scaling function.See Figure(9).T KKTKln (T/T )4T+...Tχ 1 - 1imp1bFIG.9.Qualitative behaviour of the impurity susceptibility.The temperature dependent part of the low T resistivity for the dilute random array is 2nd orderin perturbation theory,ρ=ρu [1−dTWe start by considering a left-moving spinless fermion field with Hamiltonian density:H =1dxψL .(2.1)Define the current (=density)operator,J L (x −t )=:ψ+L ψL :(x,t )=lim ǫ→0[ψL (x )ψL (x +ǫ)− 0|ψL (x )ψL (x +ǫ)|0 ](2.2)(Henceforth we generally drop the subscripts “L”.)We will reformulate the theory in terms ofcurrents (key to bosonization).Consider:J (x )J (x +ǫ)asǫ→0=:ψ†(x )ψ(x )ψ†(x +ǫ)ψ(x +ǫ):+[:ψ†(x )ψ(x +ǫ):+:ψ(x )ψ†(x +ǫ):]G (ǫ)+G (ǫ)2G (ǫ)= 0|ψ(x )ψ†(x +ǫ)|0 =1ǫ2]=lim ǫ→01dxψ:H =1(x −y −iδ)2+1dx1x −y +iδ=2πid2∂φ2∂φ∂tφ(y )]=iδ(x −y )(2.7)We can again decompose it into the left and right-moving parts,(∂t 2−∂x 2)φ=(∂t +∂x )(∂t −∂x )φφ(x,t )=φL (x +t )+φR (x −t )(∂t −∂x )φL ≡∂−φL =0,∂+φR =0H =14(∂+φ)2=14(∂+φL )2(2.8)Consider the Hamiltonian density for a left-moving boson field:H =1dxδ(x −y )(2.9)Comparing to the Fermionic case,we see that:J L =√π∂+φ,(2.10)12since the commutation relations and Hamiltonian are the same.That means the operators are the same with appropriate boundary conditions.Let’s compare the spectra.For the Fermionic case,choose boundary condition:ψ(l)=−ψ(−l)(i.e.ψL(l)+ψR(l)=0),k=π2),n=0,±1,±2...(2.11)[See Figure(5).Note that we have shifted k by k F.]Consider the minimum energy state of chargeQ(relative to the ground state).See Figure(11).We have the single Fermion energy:E=v F k,(2.12) so:E(Q)=v F π2)=v Fπl ( 1k-kFFIG.12.A particle-hole excitation in which three electrons are raised four levels and then one electron is raised three levels.Now consider the bosonic spectrum.What are the boundary conditions?Try the periodic one,φ(l)=φ(−l)⇒k=πml (∞ 1n m·m),n m=occupation number:0,1,2,...(2.16)Where does the Q2term in Eq.(2.14)come from?We need more general boundary condition on the bosonfield.Letφbe an angular variable:φL(−l)=φL(l)+√πl·(x+t)+∞m=114πm(e−iπm2 ∂φ2 ∂φl[14πφL,(2.19) which gives the correct Green’s function and implies the same angular definition ofφL.For the Kondo effect we are also interested in the phase-shifted boundary condition:[See Figure(6).]ψL(l)=+ψL(−l),k=πl Q(Q−1)We have the degenerate ground state,Q=0or1,which correspond to an anti-periodic boundary condition onφ,φ(l)=φ(−l)+√2)E=π2(Q−1l(1dxψα,(α=1,2,summed).(2.22)Now we have charge and spin currents(or densities).We can write H in a manifestly SU(2)invariant way,quadratic in charge and spin currents:J=:ψα†ψα:, J=ψ†ασβα4:ψ†αψαψ†βψβ:+3idxψα+c-number,J2=:ψ†αψαψ†βψβ:+2iψα+d8πJ2+12[J↑+J↓,J↑−J↓]=0.(2.26) From[J, J]=0,we see that H is sum of commuting charge and spin parts.[J a(x),J b(y)]=2πψ†[σ2b]ψ·δ(x−y)+tr[σ2b]2πiddxδ(x−y).(2.27)We obtain the Kac-Moody algebra of central charge k=1.More generally the coefficient of the second term is multiplied by an integer k.Fourier transforming,Jn≡1lx J(x),[J an,J b m]=iǫabc J c n+m+1l18πJ2+14π(J2↑+J2↓)=14[(∂+(φ↑+φ↓2))2+(∂+(φ↑−φ↓2))2] =1Now we have introduced two commuting charge and spin free massless bosons.SU(2)symmetry is now concealed but boundary condition on φs must respect it.Consider the spectrum of fermion theory with boundary condition:ψ(l )=−ψ(−l ),E =πV22+Q ↓2(Q ↑−Q ↓)E =πv F4Q 2+(S z )2+∞ 1mn c m+∞1mn s m ](2.32)=E c +E sφc =√2√l (x +t )+...φs=π2S zl[1l[12),(±1,0).(2.36)Now Q =2S z +1(mod 2);i.e.we “glue”together charge and spin excitations in two different ways,either(even,integer)⊕(odd,half-integer)or (even,half-integer)⊕(odd,integer),(2.37)depending on the boundary conditions.Theπl·(integer).Likewise for all half-integer spin states,(s z )2=1dxψLα+λψ†αLσβα8πJ 2+1The Kondo interaction involves spinfields only,not chargefields:H=H s+H c.Henceforth we only consider the spin part.In Fourier transformed form,H s=π3∞ n=−∞ J−n· J n+λ∞ n=−∞ J n· S)[J a n,J b m]=iǫabc J c n+m+ndlnD =−λ2+···.That is a smallλ>0grows.What is the infrared stablefixed point?Considerλ=23l∞ n=−∞[( J−n+ S)·( J n+ S)−32δabδn,−m.(3.4)H is quadratic in the new currents, J n≡ J n+ S,which obey the same Kac-Moody algebra!Whatis the spectrum of H(λ=21−32-Integer.See Figure(13).This is equivalent to aπ2-integer)(even,16πJ(x)2+λ1 J(0)2δ(x).(3.9)This is the only dimension-2rotationally invariant operator in the spin sector.We have succeededin reducing two dimension-2operators to one.The other one is the charge-operatorλ2J(0)2δ(x),λ2=0because there is no interaction in the charge sector(with other regularization we expectλ1∼1D<<λ1).171/2 integer-s tower towerinteger-s FIG.13.At λ=2/3the 1/2-integer-spin conformal tower is mapped into the integer-spin conformal tower.Now we calculate the specific heat and susceptibility to 1st order in λ1.Susceptibility of left-moving free fermions:0-th orderM =12)−n (ǫ−h2π(forT <<D )1st order χ=13T 2[dx J(x )]2 J (0)2 +...(3.10)A simplifying trick is to replace:δH =λ1 J2(0)δ(x )−→λ16π+λ1l)H.(3.13)Equivalently in a thermal average,T →Tl≡T (λ1)(3.14)χ(λ1,T )=11+3πλ1/l χ(0,T (λ1))≈[1−3πλ12π−3λ1where in the last equality thefirst term represents the bulk part and the second one,of order∼13.(3.16) Each free left-moving boson makes an identical contribution.1st order inλ1C s(λ1,T)=∂3T3−π2λ1T(3.17)δC sl =2δC sδC/C =2=C8π2r1r2[e−ik F(r1+r2)(G LR(r1,r2)−G LR,0(r1,r2))+h.c.]=G03(r1)ΣG03(r2).(3.20) The self-energyΣdepends only on the frequency.It gets multiplied by the impurity concentationfor afinite density(in the dilute limit).We must calculate the1D Green’s function G LR(r1,r2,ω) perturbatively inλO(λ01):G LR(r1,r2)=−G0LL(r1,−r2)=−G0LL(r1+r2)=−G0LR(r1,r2),(3.21) where the(−)sign comes from the change in boundary conditions,G LR−G0LR=−2G0LR+O(λ1)(3.22) To calculate to higher orders it is convenient to write the interaction as:J2=−34(ψ†αddxψα)(3.23)To second order inλ1,we have the Feynman diagrams shown in Figure(14),giving:ΣR(ω)=−in2(3πλ1)2ω2−12πν[1−e2iδ(ω)]+ΣR inel(ω)δ=π2+...1τ(ω)=n i2(3πλ1)2ω2−1The leadingλ1dependence is O(λ21)in this case.The O(λ1)term inΣR is real.We calculate the conductivity from the Kubo formula.(There is no contribution from the scattering vertex for pure s-wave scattering.)σ(T)=2e2(2π)3 −∂n2n i[1+18(3πλ1)2(ǫ2k+(π2T2)]ρ(T)=1π(ev Fν)2[1−9T K.Numerical or Bethe ansatz methods are needed tofind the precise value ofλ1(D,λ)∝1d lnD =−νλ2+k2ψ0,(4.3)forλ>0(antiferromagnetic case)the minimum energy state has maximum spin for electrons at0i.e.spin=k/2.Coupling this spin-k/2to a spin-s,we don’t get a singlet if s=k/2,but ratheran effective spin of size|s−k/2|.[See Figure(15).]The impurity is underscreened(k/2<s)or overscreened(k/2>s).20FIG.15.Formation of an effective spin at strong Kondo coupling.k=3,s=1and s eff=1/2. Now let tλ2<<1See Figure(16).What is the sign ofλeff?The coupling of the electron spins is antiferromagnetic:λeff S e1,0· S e1,1,withλeff>0(as in the Hubbard model).But we must combine spinsSeff= S+ Sel,0.(4.4)For k2>s, S ef f||+ S el,0.So,ultimately,λeff<0in the underscreenedcase andλeff>0in the overscreened case.In thefirst(underscreened)case,the assumptionλ→∞was consistent since a ferromagneticλeff→0under renormalizaton and this impliesλ→∞,since λeff∼−tFIG.18.The overscreened case with s=1/2,k=2.rge-k LimitTheβ-function is:β=λ2−kdk λc=2λc−3k.(4.7) This implies that the leading irrelevant coupling constant at the non-trivial(infrared)fixed pointhas dimension2/k at large k,so that(λ−λc)scales asΛ2/k.Thus the leading irrelevant operatorhas dimension(1+2/k).This is not an integer!This implies that this critical point is not a Fermiliquid.B.Current Algebra ApproachWe can gain some insight into the nature of the non-trivial critical point using the current algebraapproach discussed in the previous section for the k=1case.It is now convenient to use a formof bosonization which separates spin,charge andflavour(i.e.channel)degrees of freedom.This representation is known as a conformal embedding.We introduce charge(J),spin( J)andflavour(J A)currents.A runs over the k2−1generators of SU(k).The corresponding elements of thealgebra are written T A.These are traceless Hermitean matrices normalized so that:tr T A T B=12 δb cδd a−12ψiβJ A≡ψ†iα(T A)j iψjα.(4.11) (All repeated indices are summed.)It can be seen using Eq.(4.9)that the free fermion Hamiltoniancan be written in terms of these currents as:H=12π(k+2) J2+1C V(G)+k,(4.14) where Dim(G)is the dimension of the group and C V(G)is the quadratic Casimir in the fundamental representation.For SU(k)this has the value:C V(SU(k))=k.(4.15) Thus the total value of the central charge,c,is:c TOT=1+3·kk+2=2k,(4.16)the correct value for2k species of free plicated“gluing conditions”must be imposed tocorrectly reproduce the free fermion spectra,with various boundary conditions.These were workedout in general by Altshuler,Bauer and Itzykson.27The SU(2)k sector consists of k+1conformaltowers,labelled by the spin of the lowest energy(“highest weight”)state:s=0,1/2,1,...k/2.32,33 We may now treat the Kondo interaction much as in the single channel case.It only involves thespin sector which now becomes:H s=12+k,(4.18) where the Hamiltonian reduces to its free form after a shift of the current operators by S whichpreserves the KM algebra.We note that at large k this special value ofλreduces to the one corresponding to the critical point:λc→2/k.While this observation is tantalizing,it leaves many open questions.We might expect that some rearranging of the(k+1)SU(2)k conformal towers takes place at the critical point but preciselywhat is it?Does it correspond to some sort of boundary condition?If so what?How can wecalculate thermodynamic quantities and Green’s functions?To answer these questions we need to understand some more technical aspects of CFT in the presence of boundaries.V.BOUNDARY CONFORMAL FIELD THEORY We will assume that the critical point corresponds to a conformally invariant boundary conditionon the free ing the general theory of conformally invariant boundary conditions developed by Cardy28we can completely solve for the critical properties of the model.Why assume that the critical point corresponds to such a boundary condition?It is convenient to work in the space-(imaginary)time picture.The impurity then sits at the boundary,r=0of the half-plane r>0 on which the Kondo effect is defined.If we consider calculating a two-point Green’s function when both points are taken very far from the boundary(with their separation heldfixed)then we expectto obtain bulk behaviour,unaffected by the boundary.[See Figure(19).]This,at long distances and times is the conformally invariant behaviour of the free fermion system.Very close to the boundary,we certainly do not expect the behaviour to be scale invariant(let alone conformallyinvariant)because various microscopic scales become important.The longest of these scales is presumably the Kondo scale,ξK≈v F/T L≈ae1/νλ.Beyond this distance,it is reasonable to expect scale-invariant behaviour.However,if the two points are far from each other compared to theirdistance from the boundary[Figure(20)]then the behaviour is still influenced by the boundary even when both points are far from it.We have a sort of boundary-dependent termination of the bulk conformally invariant behaviour.The dependence on the details of the boundary(such as the value ofξK)drops out.We may think of various types of boundaries as falling into universality classes,each corresponding to a type of conformally invariant behaviour.Rather remarkably,the above statements hold true whether we are dealing with a2-dimensional classical statistical system with some boundary condition imposed,or dealing with a(1+1)-dimensional quantum system with some dynamical degrees of freedom living on the boundary.In fact,we already saw an example of this in the single-channel Kondo problem.The dynamical impurity drops out of the description of the low-energy physics and is replaced by a simple,scale-invariant boundary condition,ψL=−ψR.FIG.19.The bulk limit.ξFIG.20.The boundary limit.Precisely what is meant by a conformally invariant boundary condition?Without boundaries,conformal transformations are analytic mappings of the complex plane:z ≡τ+ix,(5.1)into itself:z →w (z ).(5.2)(Henceforth,we set the Fermi velocity,v F =1.)We may Taylor expand an arbitrary conformaltransformation around the origin:w (z )=∞ 0a n z n ,(5.3)where the a n ’s are arbitrary complex coefficients.They label the various generators of the conformalgroup.It is the fact that there is an infinite number of generators (i.e.coefficients)which makesconformal invariance so powerful in (1+1)dimensions.Now suppose that we have a boundary atx =0,the real axis.At best,we might hope to have invariance under all transformations whichleave the boundary fixed.This implies the condition:w (τ)∗=w (τ).(5.4)We see that there is still an infinite number of generators,corresponding to the a n ’s of Eq.(5.3)except that now we must impose the conditions:a ∗n =a n .(5.5)We have reduced the (still ∞)number of generators by a factor of 1/2.The fact that there is still an∞number of generators,even in the presence of a boundary,means that this boundary conformalsymmetry remains extremely powerful.To exploit this symmetry,following Cardy,it is very convenient to consider a conformally invariantsystem defined on a cylinder of circumference βin the τ-direction and length l in the x direction,with conformally invariant boundary conditions A and B at the two ends.[See Figure (21).]Fromthe quantum mechanical point of view,this corresponds to a finite temperature,T =1/β.Thepartition function for this system is:Z AB =tr e −βH lAB ,(5.6)where we are careful to label the Hamiltonian by the boundary conditions as well as the length ofthe spatial interval,both of which help to determine the spectrum.Alternatively,we may make amodular transformation,τ↔x .Now the spatial interval,of length,β,is periodic.We write thecorresponding Hamiltonian as H βP .The system propagates for a time interval l between initial andfinal states A and B .Thus we may equally well write:Z AB =<A |e −lH βP |B >.(5.7)Equating these two expressions,Eq.(5.6)and (5.7)gives powerful constraints which allow us todetermine the conformally invariant boundary conditions.βBlA FIG.21.Cylinder of length l ,circumference βwith boundary conditions A andB at the two ends.To proceed,we make a further weak assumption about the boundary conditions of interest.We assume that the momentum density operator,T−¯T vanishes at the boundary.This amounts to a type of unitarity condition.In the free fermion theory this becomes:ψ†αi L ψLαi(t,0)−ψ†αiRψRαi(t,0)=0.(5.8)Note that this is consistent with both boundary conditions that occured in the one-channel Kondoproblem:ψL=±ψR.Since T(t,x)=T(t+x)and¯T(t,x)=¯T(t−x),it follows that¯T(t,x)=T(t,−x).(5.9) i.e.we may regard¯T as the analytic continuation of T to the negative axis.Thus,as in ourprevious discussion,instead of working with left and right movers on the half-line we may work withleft-movers only on the entire line.Basically,the energy momentum density,T is unaware of theboundary condition.Hence,in calculating the spectrum of the system with boundary conditions Aand B introduced above,we may regard the system as being defined periodically on a torus of length2l with left-movers only.The conformal towers of T are unaffected by the boundary conditions,A,B.However,which conformal towers occur does depend on these boundary conditions.We introducethe characters of the Virasoro algebra,for the various conformal towers:χa(e−πβ/l)≡ i e−βE a i(2l),(5.10) where E a i(2l)are the energies in the a th conformal tower for length2l.i.e.:E a i(2l)=π24l,(5.11)where the x a i’s correspond to the(left)scaling dimensions of the operators in the theory and c is theconformal anomaly.The spectrum of H l AB can only consist of some combination of these conformaltowers.i.e.:Z AB= a n a ABχa(e−πβ/l),(5.12)where the n a AB are some non-negative integers giving the multiplicity with which the various con-formal towers occur.Importantly,only these multiplicities depend on the boundary conditions,notthe characters,which are a property of the bulk left-moving system.Thus,a specification of allpossible multiplicities,n a AB amounts to a specification of all possible boundary conditions A.Theproblem of specifying conformally invariant boundary conditions has been reduced to determiningsets of integers,n a AB.For rational conformalfield theories,where the number of conformal towersisfinite,only afinite number of integers needs to be specified.Now let us focus on the boundary states,|A>.These must obey the operator condition:[T(x)−¯T(x)]|A>=0(∀x).(5.13) Fourier transforming with respect to x,this becomes:[L n−¯L n]|A>=0.(5.14) This implies that all boundary states,|A>must be linear combinations of the“Ishibashi states”:29|a>≡ m|a;m>⊗|a;0>.(5.17)(Note that while the states,|a;m>⊗S a0n a AB= b N a bc n b AA.(5.26)Here0labels the conformal tower of the identity operator.Importantly,the new boundary stateand multiplicities so obtained,obey Cardy’s equation.The right-hand side of Eq.(5.23)becomes:S a c<A|a0><a0|B>=<A|a0><a0|A>.(5.29)S a0This gives:b S a b n b AB=S ac S a0<A|a0><a0|A>=<A|a0><a0|B>,(5.30)proving that fusion does indeed give a new solution of Cardy’s equations.The multiplicities,n a BBare given by double fusion:n a BB= b,d N a bc N b dc n d AA.(5.31)[Recall that|B>is obtained from|A>by fusion with the primary operator c.]It can be checkedthat the Cardy equation with A=B is then obeyed.It is expected that,in general,we can generatea complete set of boundary states from an appropriate reference state by fusion with all possibleconformal towers.VI.BOUNDARY CONFORMAL FIELD THEORY RESULTS ON THE MULTI-CHANNEL KONDOEFFECTA.Fusion and the Finite-Size SpectrumWe are now in a position to bring to bear the full power of boundary conformalfield theory on the Kondo problem.By the arguments at the beginning of Sec.V,we expect that the infraredfixed points describing the low-T properties of the Kondo Hamiltonian correspond to conformallyinvariant boundary conditions on free fermions.We might also expect that we could determinethese boundary conditions and corresponding boundary states by fusion with appropriate operatorsbeginning from some convenient,trivial,reference state.We actually already saw a simple example of this in Sec.III in the single channel,s=1/2,Kondo problem.There we observed that the free fermion spectrum,with convenient boundary conditionscould be written:(0,even)⊕(1/2,odd).(6.1) Here0and1/2label the SU(2)1KM conformal towers in the spin sector,while“even”and“odd”label the conformal towers in the charge sector.We argued that,after screening of the impurityspin,the infraredfixed point was described by free fermions with aπ/2phase shift,correspondingto a spectrum:(1/2,even)⊕(0,odd).(6.2) The change in the spectrum corresponds to the interchange of SU(2)1conformal towers:0↔1/2.(6.3) This indeed corresponds to fusion,with the spin-1/2primaryfield of the WZW model.To see thisnote that the fusion rules for SU(2)1are simply[from Eq.(5.25)]:。

ENSEMBLE EMPIRICAL MODE DECOMPOSITION-A NOISE ASSISTED DATA ANALYSIS METHOD

ENSEMBLE EMPIRICAL MODE DECOMPOSITION-A NOISE ASSISTED DATA ANALYSIS METHOD

July 24,200813:6WSPC/244-AADA 00004Advances in Adaptive Data Analysis 1Vol.1,No.1(2008)1–41c World Scientific Publishing Company 3ENSEMBLE EMPIRICAL MODE DECOMPOSITION:A NOISE ASSISTED DATA ANALYSIS METHOD 5ZHAOHUA WU ∗and NORDEN E.HUANG †∗Center for Ocean–Land–Atmosphere Studies 74041Powder Mill Road,Suite 302Calverton,MD 20705,USA 9†Research Center for Adaptive Data Analysis National Central University 11300Jhongda Road,Chungli,Taiwan 32001A new Ensemble Empirical Mode Decomposition (EEMD)is presented.This new 13approach consists of sifting an ensemble of white noise-added signal and treats the mean as the final true result.Finite,not infinitesimal,amplitude white noise is necessary to 15force the ensemble to exhaust all possible solutions in the sifting process,thus mak-ing the different scale signals to collate in the proper intrinsic mode functions (IMF)17dictated by the dyadic filter banks.As the EMD is a time–space analysis method,the white noise is averaged out with sufficient number of trials;the only persistent part 19that survives the averaging process is the signal,which is then treated as the true and more physical meaningful answer.The effect of the added white noise is to provide a 21uniform reference frame in the time–frequency space;therefore,the added noise collates the portion of the signal of comparable scale in one IMF.With this ensemble mean,one 23can separate scales naturally without any a priori subjective criterion selection as in the intermittence test for the original EMD algorithm.This new approach utilizes the 25full advantage of the statistical characteristics of white noise to perturb the signal in its true solution neighborhood,and to cancel itself out after serving its purpose;therefore,it 27represents a substantial improvement over the original EMD and is a truly noise-assisted data analysis (NADA)method.29Keywords :1.Introduction 31Empirical Mode Decomposition (EMD)has been proposed recently 1,2as an adap-tive time–frequency data analysis method.It has proven to be quite versatile in 33a broad range of applications for extracting signals from data generated in noisy nonlinear and nonstationary processes (see,for example,Refs.3and 4).As useful 35as EMD proved to be,it still leaves some annoying difficulties unresolved.One of the major drawbacks of the original EMD is the frequent appearance 37of mode mixing,which is defined as a single Intrinsic Mode Function (IMF)either consisting of signals of widely disparate scales,or a signal of a similar scale residing39in different IMF components.Mode mixing is a consequence of signal intermittency.1July24,200813:6WSPC/244-AADA000042Z.Wu&N.E.HuangAs discussed by Huang et al.,1,2the intermittence could not only cause serious 1aliasing in the time–frequency distribution,but also make the physical meaningof individual IMF unclear.To alleviate this drawback,Huang et al.2proposed the 3intermittence test,which can indeed ameliorate some of the difficulties.However,the approach itself has its own problems:First,the intermittence test is based on 5a subjectively selected scale.With this subjective intervention,the EMD ceases tobe totally adaptive.Secondly,the subjective selection of scales works if there are 7clearly separable and definable timescales in the data.In case the scales are notclearly separable but mixed over a range continuously,as in the case of the majority 9of natural or man-made signals,the intermittence test algorithm with subjectivelydefined timescales often does not work very well.11To overcome the scale separation problem without introducing a subjective intermittence test,a new noise-assisted data analysis(NADA)method is proposed, 13the Ensemble EMD(EEMD),which defines the true IMF components as the meanof an ensemble of trials,each consisting of the signal plus a white noise offinite 15amplitude.With this ensemble approach,we can clearly separate the scale nat-urally without any a priori subjective criterion selection.This new approach is 17based on the insight gleaned from recent studies of the statistical properties ofwhite noise,5,6which showed that the EMD is effectively an adaptive dyadicfilter 19bank a when applied to white noise.More critically,the new approach is inspired bythe noise-added analyses initiated by Flandrin et al.7and Gledhill.8Their results 21demonstrated that noise could help data analysis in the EMD.The principle of the EEMD is simple:the added white noise would populate 23the whole time–frequency space uniformly with the constituting components ofdifferent scales.When signal is added to this uniformly distributed white back-25ground,the bits of signal of different scales are automatically projected onto properscales of reference established by the white noise in the background.Of course, 27each individual trial may produce very noisy results,for each of the noise-addeddecompositions consists of the signal and the added white noise.Since the noise in 29each trial is different in separate trials,it is canceled out in the ensemble mean ofenough trials.The ensemble mean is treated as the true answer,for,in the end, 31the only persistent part is the signal as more and more trials are added in theensemble.33The critical concept advanced here is based on the following observations:1.A collection of white noise cancels each other out in a time-space ensemble mean;35therefore,only the signal can survive and persist in thefinal noise-added signalensemble mean.37a A dyadicfilter bank is a collection of band-passfilters that have a constant band-pass shape(e.g.a Gaussian distribution)but with neighboringfilters covering half or double of the frequencyrange of any singlefilter in the bank.The frequency ranges of thefilters can be overlapped.Forexample,a simple dyadicfilter bank can includefilters covering frequency windows such as50to120Hz,100to240Hz,200to480Hz,etc.July24,200813:6WSPC/244-AADA00004Ensemble Empirical Mode Decomposition32.Finite,not infinitesimal,amplitude white noise is necessary to force the ensemble1to exhaust all possible solutions;thefinite magnitude noise makes the differentscale signals reside in the corresponding IMF,dictated by the dyadicfilter banks, 3and render the resulting ensemble mean more meaningful.3.The true and physically meaningful answer to the EMD is not the one without5noise;it is designated to be the ensemble mean of a large number of trialsconsisting of the noise-added signal.7This EEMD proposed here has utilized all these important statistical character-istics of noise.We will show that the EEMD utilizes the scale separation principle 9of the EMD,and enables the EMD method to be a truly dyadicfilter bank forany data.By addingfinite noise,the EEMD eliminates mode mixing in all cases 11automatically.Therefore,the EEMD represents a major improvement of the EMDmethod.13In the following sections,a systematic exploration of the relation between noise and signal in data will be presented.Studies of Flandrin et al.5and Wu and Huang6 15have revealed that the EMD serves as a dyadicfilter for various types of noise.Thisimplies that a signal of a similar scale in a noisy data set could possibly be contained 17in one IMF component.It will be shown that adding noise withfinite rather thaninfinitesimal amplitude to data indeed creates such a noisy data set;therefore, 19the added noise,havingfilled all the scale space uniformly,can help to eliminatethe annoying mode mixing problemfirst noticed by Huang et al.2Based on these 21results,we will propose formally the concepts of NADA and noise-assisted signalextraction(NASE),and will develop a method called the EEMD,which is based 23on the original EMD method,to make NADA and NASE possible.The paper is arranged as follows.Section2will summarize previous attempts of 25using noise as a tool in data analysis.Section3will introduce the EEMD method,illustrate more details of the drawbacks associated with mode mixing,present con-27cepts of NADA and of NASE,and introduce the EEMD in detail.Section4willdisplay the usefulness and capability of the EEMD through examples.Section5 29will further discuss the related issues to the EEMD,its drawbacks,and their corre-sponding solutions.A summary and discussion will be presented in thefinal section 31of the main text.Two appendices will discuss some related issues of EMD algorithmand a Matlab EMD/EEMD software for research community to use.332.A Brief Survey of Noise Assisted Data AnalysisThe word“noise”can be traced etymologically back to its Latin root of“nausea,”35meaning“seasickness.”Only in Middle English and Old French does it start to gainthe meaning of“noisy strife and quarrel,”indicating something not at all desirable.37Today,the definition of noise varies in different circumstances.In science and engi-neering,noise is defined as disturbance,especially a random and persistent kind 39that obscures or reduces the clarity of a signal.In natural phenomena,noise couldJuly24,200813:6WSPC/244-AADA000044Z.Wu&N.E.Huangbe induced by the process itself,such as local and intermittent instabilities,irresolv-1able subgrid phenomena,or some concurrent processes in the environment in whichthe investigations are conducted.It could also be generated by the sensors and 3recording systems when observations are made.When efforts are made to under-stand data,important differences must be considered between the clean signals that 5are the direct results of the underlying fundamental physical processes of our inter-est(“the truth”)and the noise induced by various other processes that somehow 7must be removed.In general,all data are amalgamations of signal and noise,i.e.x(t)=s(t)+n(t),(1) 9in which x(t)is the recorded data,and s(t)and n(t)are the true signal andnoise,respectively.Because noise is ubiquitous and represents a highly undesirable 11and dreaded part of any data,many data analysis methods were designed specifi-cally to remove the noise and extract the true signals in data,although often not 13successful.Since separating the signal and the noise in data is necessary,three important 15issues should be addressed:(1)The dependence of the results on the analysis meth-ods used and assumptions made on the data.(For example,a linear regression of 17data implicitly assumes the underlying physics of the data to be linear,while aspectrum analysis of data implies the process is stationary.)(2)The noise level to 19be tolerated in the extracted“signals,”for no analysis method is perfect,and inalmost all cases the extracted“signals”still contain some noise.(3)The portion 21of real signal obliterated or deformed through the analysis processing as part ofthe noise.(For example,Fourierfiltering can remove harmonics through low-pass 23filtering and thus deform the waveform of the fundamentals.)All these problems cause misinterpretation of data,and the latter two issues are 25specifically related to the existence and removal of noise.As noise is ubiquitous,steps must be taken to insure that any meaningful result from the analysis should 27not be contaminated by noise.To avoid possible illusion,the null hypothesis testagainst noise is often used with the known noise characteristics associated with the 29analysis method.6,9,7Although most data analysis techniques are designed specifi-cally to remove noise,there are,however,cases when noise is added in order to help 31data analysis,to assist the detection of weak signals,and to delineate the under-lying processes.The intention here is to provide a brief survey of the beneficial 33utilization of noise in data analysis.The earliest known utilization of noise in aiding data analysis was due to Press 35and Tukey10known as pre-whitening,where white noise was added toflatten thenarrow spectral peaks in order to get a better spectral estimation.Since then, 37pre-whitening has become a very common technique in data analysis.For exam-ple,Fuenzalida and Rosenbluth11added noise to process climate data;Link and 39Buckley,12and Zala et al.13used noise to improve acoustic signal;Strickland andIl Hahn14used wavelet and added noise to detect objects in general;and Trucco15 41used noise to help design specialfilters for detecting embedded objects on the oceanJuly24,200813:6WSPC/244-AADA00004Ensemble Empirical Mode Decomposition5floor experimentally.Some general problems associated with this approach can be 1found in the works by Priestley,16Kao et al.,17Politis,18and Douglas et al.19 Another category of popular use of noise in data analysis is more related to the 3analysis method than to help extracting the signal from the data.Adding noiseto data helps to understand the sensitivity of an analysis method to noise and 5the robustness of the results obtained.This approach is used widely;for example,Cichocki and Amari20added noise to various data to test the robustness of the 7independent component analysis(ICA)algorithm,and De Lathauwer et al.21usednoise to identify error in ICA.9Adding noise to the input to specifically designed nonlinear detectors could also be beneficial to detecting weak periodic or quasi-periodic signals based on a physical 11process called stochastic resonance.The study of stochastic resonance was pioneeredby Benzi and his colleagues in the early1980s.The details of the development of 13the theory of stochastic resonance and its applications can be found in a lengthyreview paper by Gammaitoni et al.22It should be noted here that most of the 15past applications(including those mentioned earlier)have not used the cancellationeffects associated with an ensemble of noise-added cases to improve their results.17Specific to analysis using EMD,Huang et al.23added infinitesimal magnitude noise to earthquake data in an attempt to prevent the low frequency mode from 19expanding into the quiescent region.But they failed to realize fully the implicationsof the added noise in the EMD method.The true advances related to the EMD 21method had to wait until the two pioneering works by Gledhill8and Flandrin et al.7 Flandrin et al.7used added noise to overcome one of the difficulties of the 23original EMD method.As the EMD is solely based on the existence of extrema(either in amplitude or in curvature),the method ceases to work if the data lacks 25the necessary extrema.An extreme example is in the decomposition of a Diracpulse(delta function),where there is only one extrema in the whole data set.To 27overcome the difficulty,Flandrin et al.7suggested adding noise with infinitesimalamplitude to the Dirac pulse so as to make the EMD algorithm operable.Since 29the decomposition results are sensitive to the added noise,Flandrin et al.7ran anensemble of5000decompositions,with different versions of noise,all of infinitesimal 31amplitude.Though they used the mean as thefinal decomposition of the Diracpulse,they defined the true answer as33E{d[n]+εr k[n]},(2)d[n]=lime→0+in which,[n]represents n th data point,d[n]is the Dirac function,r k[n]is a random 35number,εis the infinitesimal parameter,and E{}is the expected value.Flandrin’snovel use of the added noise has made the EMD algorithm operable for a data set 37that could not be previously analyzed.Another novel use of noise in data analysis is by Gledhill,8who used noise to 39test the robustness of the EMD algorithm.Although an ensemble of noise was used,he never used the cancellation principle to define the ensemble mean as the true 41answer.Based on his discovery(that noise could cause the EMD to produce slightlyJuly24,200813:6WSPC/244-AADA000046Z.Wu&N.E.Huangdifferent outcomes),he assumed that the result from the clean data without noise 1was the true answer and thus designated it as the reference.He then defined thediscrepancy,∆,as3∆=mj=1t(cr j(t)−cn j(t))21/2,(3)where cr j and cn j are the j th component of the IMF without and with noise added, 5and m is the total number of IMFs generated from the data.In his extensive study of the detailed distribution of the noise-caused“discrepancy,”he concluded that 7the EMD algorithm is reasonably stable for small perturbations.This conclusion is in slight conflict with his observations that the perturbed answer with infinitesimal 9noise showed a bimodal distribution of the discrepancy.Gledhill had also pushed the noise-added analysis in another direction:He had 11proposed to use an ensemble mean of noise-added analysis to form a“Composite Hilbert spectrum.”As the spectrum is non-negative,the added noise could not 13cancel out.He then proposed to keep a noise-only spectrum and subtract it from the full noise-added spectrum at the end.This non-cancellation of noise in the 15spectrum,however,forced Gledhill8to limit the noise used to be of small magnitude, so that he could be sure that there would not be too much interaction between the 17noise-added and the original clean signal,and that the contribution of the noise to thefinal energy density in the spectrum would be negligible.19Although noise of infinitesimal amplitude used by Gledhill8has improved the confidence limit of thefinal spectrum,Gledhill explored neither fully the cancella-21tion property of the noise nor the power offinite perturbation to explore all possible solutions.Furthermore,it is well known that whenever there is intermittence,the 23signal without noise can produce IMFs with mode mixing.There is no justification to assume that the result without added noise is the truth or the reference sig-25nal.These reservations notwithstanding,all these studies by Flandrin et al.7and Gledhill8had still greatly advanced the understanding of the effects of noise in the 27EMD method,though the crucial effects of noise had yet to be clearly articulated and fully explored.29In the following,the new noise-added EMD approach will be explained,in which the cancellation principle will be fully utilized,even withfinite amplitude noise.Also 31emphasized is thefinding that the true solution of the EMD method should be the ensemble mean rather than the clean data.This full presentation of the new method 33will be the subject of the next section.3.Ensemble Empirical Mode Decomposition353.1.The empirical mode decompositionThis section starts with a brief review of the original EMD method.The detailed 37method can be found in the works of Huang et al.1and Huang et al.2Different to almost all previous methods of data analysis,the EMD method is adaptive,with 39July24,200813:6WSPC/244-AADA00004Ensemble Empirical Mode Decomposition7 the basis of the decomposition based on and derived from the data.In the EMD 1approach,the data X(t)is decomposed in terms of IMFs,c j,i.e.x(t)=nj=1c j+r n,(4)3where r n is the residue of data x(t),after n number of IMFs are extracted.IMFs are simple oscillatory functions with varying amplitude and frequency,and hence 5have the following properties:1.Throughout the whole length of a single IMF,the number of extrema and the 7number of zero-crossings must either be equal or differ at most by one(althoughthese numbers could differ significantly for the original data set);92.At any data location,the mean value of the envelope defined by the local maximaand the envelope defined by the local minima is zero.11In practice,the EMD is implemented through a sifting process that uses only local extrema.From any data r j−1,say,the procedure is as follows:(1)identify all 13the local extrema(the combination of both maxima and minima)and connect all these local maxima(minima)with a cubic spline as the upper(lower)envelope; 15(2)obtain thefirst component h by taking the difference between the data and thelocal mean of the two envelopes;and(3)Treat h as the data and repeat steps1and 172as many times as is required until the envelopes are symmetric with respect to zero mean under certain criteria.Thefinal h is designated as c j.A complete sifting 19process stops when the residue,r n,becomes a monotonic function from which no more IMFs can be extracted.21Based on this simple description of EMD,Flandrin et al.5and Wu and Huang6 have shown that,if the data consisted of white noise which has scales populated 23uniformly through the whole timescale or time–frequency space,the EMD behaves as a dyadicfilter bank:the Fourier spectra of various IMFs collapse to a single 25shape along the axis of logarithm of period or frequency.Then the total number of IMFs of a data set is close to log2N with N the number of total data points. 27When the data is not pure noise,some scales could be missing;therefore,the total number of the IMFs might be fewer than log2N.Additionally,the intermittency 29of signals in certain scale would also cause mode mixing.3.2.Mode mixing problem31“Mode mixing”is defined as any IMF consisting of oscillations of dramatically dis-parate scales,mostly caused by intermittency of the driving mechanisms.When 33mode mixing occurs,an IMF can cease to have physical meaning by itself,suggest-ing falsely that there may be different physical processes represented in a mode. 35Even though thefinal time–frequency projection could rectify the mixed mode to some degree,the alias at each transition from one scale to another would irrecov-37erably damage the clean separation of scales.Such a drawback wasfirst illustratedJuly24,200813:6WSPC/244-AADA000048Z.Wu&N.E.Huangby Huang et al.2in which the modeled data was a mixture of intermittent high-1frequency oscillations riding on a continuous low-frequency sinusoidal signal.Analmost identical example used by Huang et al.2is presented here in detail as an 3illustration.The data and its sifting process are illustrated in Fig.1.The data has its funda-5mental part as a low-frequency sinusoidal wave with unit amplitude.At the threemiddle crests of the low-frequency wave,high-frequency intermittent oscillations 7with an amplitude of0.1are riding on the fundamental,as panel(a)of Fig.1shows.The sifting process starts with identifying the maxima(minima)in the 9data.In this case,15local maxima are identified,with thefirst and the last comingfrom the fundamental,and the other13caused mainly by intermittent oscillations 11(panel(b)).As a result,the upper envelope resembles neither the upper envelope ofthe fundamental(which is aflat line at one)nor the upper one of the intermittent 13oscillations(which is supposed to be the fundamental outside intermittent areas).Rather,the envelope is a mixture of the envelopes of the fundamental and of the 15(a)(b)(c)(d)Fig.1.The veryfirst step of the sifting process.Panel(a)is the input;panel(b)identifies localmaxima(gray dots);panel(c)plots the upper envelope(upper gray dashed line)and low envelope(lower gray dashed line)and their mean(bold gray line);and panel(d)is the difference betweenthe input and the mean of the envelopes.July24,200813:6WSPC/244-AADA00004Ensemble Empirical Mode Decomposition9Fig.2.The intrinsic mode functions of the input displayed in Fig.1(a).intermittent signals that lead to a severely distorted envelope mean(the thick grey 1line in panel(c)).Consequently,the initial guess of thefirst IMF(panel(d))is themixture of both the low frequency fundamental and the high-frequency intermittent 3waves,as shown in Fig.2.An annoying implication of such scale mixing is related to unstableness and lack 5of the uniqueness of decomposition using the EMD.With stoppage criterion givenand end-point approach prescribed in the EMD,the application of the EMD to 7any real data results in a unique set of IMFs,just as when the data is processedby other data decomposition methods.This uniqueness is here referred to as“the 9mathematical uniqueness,”and satisfaction to the mathematical uniqueness is theminimal requirement for any decomposition method.The issue that is emphasized 11here is what we refer to as“the physical uniqueness.”Since real data almost alwayscontains a certain amount of random noise or intermittences that are not known 13to us,an important issue,therefore,is whether the decomposition is sensitive tonoise.If the decomposition is insensitive to added noise of small butfinite ampli-15tude and bears little quantitative and no qualitative change,the decomposition isgenerally considered stable and satisfies the physical uniqueness;and otherwise, 17the decomposition is unstable and does not satisfy the physical uniqueness.Theresult from decomposition that does not satisfy the physical uniqueness may not be 19reliable and may not be suitable for physical interpretation.For many traditionaldata decomposition methods with prescribed base functions,the uniqueness of the 21July24,200813:6WSPC/244-AADA0000410Z.Wu&N.E.Huangsecond kind is automatically satisfied.Unfortunately,the EMD in general does not 1satisfy this requirement due to the fact that decomposition is solely based on thedistribution of extrema.3To alleviate this drawback,Huang et al.2proposed an intermittence test that subjectively extracts the oscillations with periods significantly smaller than a pre-5selected value during the sifting process.The method works quite well for thisexample.However,for complicated data with scales variable and continuously dis-7tributed,no single criterion of intermittence test can be selected.Furthermore,themost troublesome aspect of this subjectively pre-selected criterion is that it lacks 9physical justifications and renders the EMD nonadaptive.Additionally,mode mix-ing is also the main reason that renders the EMD algorithm unstable:Any small 11perturbation may result in a new set of IMFs as reported by Gledhill.8Obviously,the intermittence prevents EMD from extracting any signal with similar scales.13To solve these problems,the EEMD is proposed,which will be described in thefollowing sections.153.3.Ensemble empirical mode decompositionAs given in Eq.(1),all data are amalgamations of signal and noise.To improve the 17accuracy of measurements,the ensemble mean is a powerful approach,where dataare collected by separate observations,each of which contains different noise.To 19generalize this ensemble idea,noise is introduced to the single data set,x(t),as ifseparate observations were indeed being made as an analog to a physical experiment 21that could be repeated many times.The added white noise is treated as the possiblerandom noise that would be encountered in the measurement process.Under such 23conditions,the i th“artificial”observation will bex i(t)=x(t)+w i(t).(5) 25In the case of only one observation,one of the multiple-observation ensembles is mimicked by adding not arbitrary but different copies of white noise,w i(t),to 27that single observation as given in Eq.(5).Although adding noise may result insmaller signal-to-noise ratio,the added white noise will provide a uniform reference 29scale distribution to facilitate EMD;therefore,the low signal–noise ratio does notaffect the decomposition method but actually enhances it to avoid the mode mixing.31Based on this argument,an additional step is taken by arguing that adding whitenoise may help to extract the true signals in the data,a method that is termed 33EEMD,a truly NADA method.Before looking at the details of the new EEMD,a review of a few properties of 35the original EMD is presented:(1)the EMD is an adaptive data analysis method that is based on local charac-37teristics of the data,and hence,it catches nonlinear,nonstationary oscillationsmore effectively;39。

不规则波 理论

不规则波 理论

55第三章不規則波理論風浪使海面形成一種極為不規則(irregular) 的波形。

從風洞水槽或現地觀測中均可發現水面上的風浪,如照片3.1顯示,大波上面疊有小波,縱橫各方向的波重重疊疊,隨著時間和空間變化,同樣的波形不可能再次發生。

故風浪之波形本身構造複雜,是屬於時間及空間上的一種隨機性(random) 變動量。

風浪既是一種隨機現象,則須以統計的方法來描述其特性。

統計方法中波別分析法(individual wave analysis) 和波譜分析法(spectral analysis) 是目前被採用做為敘述海洋風浪之不規則性最普遍的方法。

利用這兩種方法從不規則波中定義出波高和週期,使其能適用於規則波的波浪理論,以達到各種工程設計應用的目的。

一般將這種統計稱為波浪的短期統計,此外另有波浪的長期統計,又分為波候統計和極值統計。

波候統計是對於長年監測到的資料做歸納、整理、和一般統計分析。

而極值統計是討論重現期的問題,將在第四章中再行敘述。

本章只就短期統計對水面波形之不規則性的問題來討論。

照片3.1實際海面風浪的照片3.1 不規則波的表示方法3.1.1 波別解析法不規則波理論56 單純來看,若視風浪的水面變位為一維的波形變化,如圖3.1所示。

對此不規則波形信號來定義個別波之波高與週期有三種方式。

第一種是零位上切 (zero up cross) 法,所謂上切零點是水位上昇曲線與平均水位線之交點,如圖3.1中小圓圈所示各點。

計算二相鄰上切零點間,水位變動之最高峰與最低谷點間之垂直高差即為波高,二相鄰上切零點的時間長度即為週期。

第二種是以水位下降曲線與平均水位線之交點,如圖3.1中小三角形所示各點,定義出個別波的方法,稱為零位下切 (zero down cross) 法。

另外第三種是無視平均水位的存在,兩相鄰波峰波谷的高差即為波高,兩相鄰波峰之間的時間即為週期,依此定義個別波的方法稱為峰至峰 (crest to crest) 法。

Questions and Answers for Fourier Optics_傅里叶光学问题解答

Questions and Answers for Fourier Optics_傅里叶光学问题解答
2014 Questions and Answers for Fourier Optics
By Li Pei
2014 Questions and Answers for Fourier Optics
1.What is the Fourier transform of a two-dimensional function g(x,y)? A: F {g ( x, y )} G ( f X , fY )
-2-
2014 Questions and Answers for Fourier Optics
By Li Pei
transfer function z 1 ( f X ) 2 ( fY ) 2 exp j 2 H ( f X , fY ) 0 The H ( f X , fY ) is nonzero only for f X 2 fY 2 1 f X 2 fY 2 otherwise 1


, showing the propagation
phenomenon as a linear spatial filter with a finite bandwidth characterized by a circular region of radius 1 in frequency plane.
-1-
2014 Questions and Answers for Fourier Optics
By Li Pei
angular spectrum of the transmitted field can be described by the convolution of the angular spectrum of the incident field (A i ) with the angular spectrum of the aperture (T). That is At ( , ) Ai ( , ) T ( , )

tpo54三篇阅读原文译文题目答案译文背景知识

tpo54三篇阅读原文译文题目答案译文背景知识

托福阅读tpo54全套解析阅读-1 (2)原文 (2)译文 (4)题目 (5)答案 (9)背景知识 (10)阅读-2 (10)原文 (10)译文 (12)题目 (13)答案 (18)背景知识 (20)阅读-3 (25)原文 (26)译文 (27)题目 (28)答案 (33)背景知识 (35)阅读-1原文The Commercialization of Lumber①In nineteenth-century America, practically everything that was built involved wood.Pine was especially attractive for building purposes.It is durable and strong, yet soft enough to be easily worked with even the simplest of hand tools.It also floats nicely on water, which allowed it to be transported to distant markets across the nation.The central and northern reaches of the Great Lakes states—Michigan, Wisconsin, and Minnesota—all contained extensive pine forests as well as many large rivers for floating logs into the Great Lakes, from where they were transported nationwide.②By 1860, the settlement of the American West along with timber shortages in the East converged with ever-widening impact on the pine forests of the Great Lakes states. Over the next 30 years, lumbering became a full-fledged enterprise in Michigan, Wisconsin, and Minnesota. Newly formed lumbering corporations bought up huge tracts of pineland and set about systematically cutting the trees. Both the colonists and the later industrialists saw timber as a commodity, but the latter group adopted a far more thorough and calculating approach to removing trees. In this sense, what happened between 1860 and 1890 represented a significant break with the past. No longer were farmers in search of extra income the main source for shingles, firewood, and other wood products. By the 1870s, farmers and city dwellers alike purchased forest products from large manufacturingcompanies located in the Great Lakes states rather than chopping wood themselves or buying it locally.③The commercialization of lumbering was in part the product of technological change. The early, thick saw blades tended to waste a large quantity of wood, with perhaps as much as a third of the log left behind on the floor as sawdust or scrap. In the 1870s, however, the British-invented band saw, with its thinner blade, became standard issue in the Great Lakes states' lumber factories.Meanwhile, the rise of steam-powered mills streamlined production by allowing for the more efficient, centralized, and continuous cutting of lumber. Steam helped to automate a variety of tasks, from cutting to the carrying away of waste. Mills also employed steam to heat log ponds, preventing them from freezing and making possible year-round lumber production.④For industrial lumbering to succeed, a way had to be found to neutralize the effects of the seasons on production. Traditionally, cutting took place in the winter, when snow and ice made it easier to drag logs on sleds or sleighs to the banks of streams. Once the streams and lakes thawed, workers rafted the logs to mills, where they were cut into lumber in the summer. If nature did not cooperate—if the winter proved dry and warm, if the spring thaw was delayed—production would suffer. To counter the effects of climate on lumber production, loggers experimented with a variety of techniques for transporting trees out of the woods. In the 1870s, loggers in the Great Lakes states began sprinkling water on sleigh roads, giving them an artificial ice coating to facilitate travel. The ice reduced the friction and allowed workers to move larger and heavier loads.⑤But all the sprinkling in the world would not save a logger from the threat of a warm winter. Without snow the sleigh roads turned to mud. In the 1870s, a set of snowless winters left lumber companies to ponder ways of liberating themselves from the seasons. Railroads were one possibility.At first, the remoteness of the pine forests discouraged common carriers from laying track.But increasing lumber prices in the late 1870s combined with periodic warm, dry winters compelled loggers to turn to iron rails. By 1887, 89 logging railroads crisscrossed Michigan, transforming logging from a winter activity into a year-round one.⑥Once the logs arrived at a river, the trip downstream to a mill could be a long and tortuous one.Logjams (buildups of logs that prevent logs from moving downstream) were common—at times stretching for 10 miles—and became even more frequent as pressure on the northern Midwest pinelands increased in the 1860s. To help keep the logs moving efficiently, barriers called booms (essentially a chain of floating logs) were constructed to control the direction of the timber. By the 1870s, lumber companies existed in all the major logging areas of the northern Midwest.译文木材的商业化①在19世纪的美国,几乎所有建筑材料都含有木材。

高英二第四课

高英二第四课

⾼英⼆第四课Lesson 4 Love Is a Fallacy by Max Shulmas Teaching PointsⅠ. Background Knowledge Ⅱ. Introduction to the Passage Ⅲ. Text analysisⅣ. Rhetorical DevicesⅤ. QuestionsTeaching ProcessWarming upQuestion 1:What is love?Question 2: What is logic?Question 3: Love is blind?Question 4: Love is reason?Introduction to the Passage1. Type of literature: a piece of narrative writing--protagonist/antagonists--climax--denouement2. The main theme3. Well chosen title and words4. Style--a very fast pace with a racy dialogue full of American colloquialism and slang--employing a variety of writing techniques to make the story vivid, dramatic and colorfulText AnalysisVocabulary1. Pay attention to words and expressions in the following aspects respectively:Spelling and PronunciationSynonymsOppositesSimilar words and expressionsSettled or habitual usage2. Word building knowledgeEffective Writing Skills1. Employing colorful lexical spectrum, from the ultra learned terms to the infra clipped vulgar forms2. Too much figurative language and ungrammatical inversion for specific purposes3. The using of short sentences, elliptical sentences and dashes to maintain the speed of narration Rhetorical Devices1. metaphor2. antithesis3. transferred epithet4. hyperbole5. metonymy6. litotes7. ellipsis8. synecdoche9. inversion10. simile11. mixed metaphor12. rhetorical questionsSpecial DifficultiesAnalyzing the logical fallaciesUsing inverted sentences to achieve emphasisEffectively using many figures of speechUnderstanding colloquial expressions and slangAllusions:--Frankenstein--PygmalionParaphrasing some sentencesIdentifying figures of speechQuestions1. Define and give an example of each of the logical fallacies discussed in this essay.2. Can you find any evidence to support the view that the writer is satirizing a bright but self-satisfied young man?3. Comment on the language used by Polly. What effect does her language create?4. Why does the writer refer to Pygmalion and Frankenstein? Are these allusions aptly chosen?5. In what sense is the conclusion ironic?Assignment:Write a composition of classification.Lesson 4 Love Is a Fallacyby Max ShulmanⅠ. Additional Information Related to the Text:1. Max SchulmanMax Schulman (1919-1988) was a 20th century American writer humorist best known for his television and short story character Dobie Gillis, as well as for best-selling novels.He first delved into the world of writing as a journalist student at the University of Minnesota. Max Schulman?s earliest published writing was for Ski-U-Mah, the college humor magazine of the University of Minnesota, in the 1930s. His writing often focused on young people, particularly in a collegiate setting. He wrote his first novel, Barefoot Boy with Cheek《⽆礼的⾚脚少年》a satire on college life, while still a student. Schulman?s works include the novels Rally Round the Flag, Boys!,《孩⼦们,团结在旗帜的周围吧》which was made into a film starring Paul Newman and Joanne Woodward; The Feather Merchant《⾐冠楚楚的商⼈》,The Zebra Derby, Sleep till Noon, and Potatoes Are Cheaper. He was also a co-writer, with Robert Paul Smith, of the long-running Broadway play, The Tender Trap, which was later adapted into a movie starring Frank Sinatra and Debbie Reynolds.Schulman?s college charater, Dobie Gillis, was the subject of a series of short stories complied under the title The Many Loves of Dobie Gillis, which became the basis for the 1953 movie The Affairs of Dobie Gillis. Shulman also wrote the series? theme song. The same year the series began. Schulman published a Dobie Gillis novel, I was a Teenage Dwarf (1959). After his success with Dobie Gillis, Shulman syndicated a humor column, “On Campus”, to over 350 collegiate newspapers at one point.A later novel, Anyone Got A Match? satirized both the television and tobacco industries, as well as the Soth and college football. His last major project was House Calls, which began as a 1978 movie based on one of his stories; it spun off the 1979-1982 television series of the same name. Schulman was the head writer.Also a screenwriter, Schulman was one of the collaborators on a 1954 non-fiction television program, Light’s Diamond Jubilee, timed to the 75th anniversary of the invention of the lihght bulb.2. Logical fallacy:逻辑谬误An argument in logic presents evidence in support of some thesis or conclusion.(逻辑论证,即提⽀持某些论题或结论的论据。

2024年高考英语试卷

2024年高考英语试卷

2024年高考英语真题试卷(新高考Ⅰ卷)第二部分一、阅读(共两节,满分50分)第一节(共15小题;每小题2.5分,满分37.5分)阅读下列短文,从每题所给的A、B、C、D四个选项中选出最佳选项。

HABITAT RESTORATIONTEAMHelp restore and protect Marin's natural areas from the Marin Headlands to Bolinas Ridge. We'll explore beautiful park sites while conducting invasive(侵入的)plant removal, winter planting, and seed collection. Habitat Restoration Team volunteers play a vital role in restoring sensitive resources and protecting endangered species across the ridges and valleys.GROUPSGroups of five or more require special arrangements and must be confirmed in advance. Please review the List of Available Projects and fill out the Group Project Request Form.AGE, SKILLS, WHAT TO BRINGV olunteers aged 10 and over are welcome. Read our Youth Policy Guidelines for youth under the age of 15.Bring your completed V olunteer Agreement Form. V olunteers under the age of18 must have the parent /guardian approval section signed.We'll be working rain or shine. Wear clothes that can get dirty. Bring layers for changing weather and a raincoat if necessary.Bring a personal water bottle, sunscreen, and lunch.No experience necessary. Training and tools will be provided. Fulfills(满足)community service requirements.UPCOMING EVENTS1.What is the aim of the Habitat Restoration Team?A.To discover mineral resources.B.To develop new wildlife parks.C.To protect the local ecosystemD.To conduct biological research.2.What is the lower age limit for joining the Habitat Restoration Team?A.5.B.10.C.15.D.18.3.What are the volunteers expected to do?A.Bring their own tools.B.Work even in bad weather.C.Wear a team uniform D.Do at least three projects."I am not crazy, "says Dr. William Farber, shortly after performing acupuncture (针灸) on a rabbit. "I am ahead of my time. "If he seems a little defensive, it might be because even some of his coworkers occasionally laugh at his unusual methods, But Farber is certain he'll have the last laugh. He's one of a small but growing number of American veterinarians(兽医)now practicing "holistic" medicine-combining traditional Western treatments with acupuncture, chiropractic(按摩疗法)and herbal medicine Farber, a graduate of Colorado State University, started out as a more conventional veterinarian. He became interested in alternative treatments 20 years ago when he suffered from terrible back pain. He tried muscle-relaxing drugs but found little relief. Then he tried acupuncture, an ancient Chinese practice, and was amazed that he improved after two or three treatments. What worked on a veterinarian seemed likely to work on his patients. So, after studying the techniques for a couple of years, he began offering them to pets Leigh Tindale's dog Charlie had a serious heart condition. After Charlie had a heart attack, Tindale says, she was prepared to put him to sleep, but Farber's treatments eased her dog's suffering so much that she was able to keep him alive for an additional five months And Priscilla Dewing reports that her horse, Nappy, "moves more easily and rides more comfortably" after a chiropractic adjustment.Farber is certain that the holistic approach will grow more popular with time, and if the past is any indication, he may be right: Since 1982, membership in the American Holistic Veterinary Medical Association has grown from 30 to over 700. "Sometimes it surprises me that it works so well, "he says. "I will do anything to help an animal. That's my job. "4.What do some of Farber's coworkers think of him?A.He's odd.B.He's strict C.He's brave.D.He's rude5.Why did Farber decide to try acupuncture on pets?A.He was trained in it at university.B.He was inspired by another veterinarian.C.He benefited from it as a patient.D.He wanted to save money for pet owners.6.What does paragraph 3 mainly talk about?A.Steps of a chiropractic treatment.B.The complexity of veterinarians' work.C.Examples of rare animal diseases.D.The effectiveness of holistic medicine.7.Why does the author mention the American Holistic Veterinary Medical Association?A.To prove Farber's point B.To emphasize its importance.C.To praise veterinarians.D.To advocate animal protection.Is comprehension the same whether a person reads a text onscreen or on paper? And are listening to and viewing content as effective as reading the written word when covering the same material? The answers to both questions are often "no. " The reasons relate to a variety of factors, including reduced concentration, an entertainment mindset(心态)and a tendency to multitask while consuming digital content.When reading texts of several hundred words or more, learning is generally more successful when it's on paper than onscreen. A large amount of research confirms this finding. The benefits of print reading particularly shine through when experimenters move from posing simple tasks-like identifying the main idea in a reading passage-to ones that require mental abstraction-such as drawing inferences from a text.The differences between print and digital reading results are partly related to paper's physical properties. With paper, there is a literal laying on of hands, along with the visual geography of distinct pages. People often link their memory of what they've read to how far into the book it was or where it was on the page.But equally important is the mental aspect. Reading researchers have proposed a theory called "shallowing hypothesis(假说). " According to this theory, people approach digital texts with a mindset suited to social media, which are often not so serious, and devote less mental effort than when they are reading print Audio(音频)and video can feel more engaging than text, and so university teachers increasingly tum to these technologies -say, assigning an online talk instead of an article by the same person. However, psychologists have demonstrated that when adults read news stories, they remember more of the content than if they listen to or view identical piecesDigital texts, audio and video all have educational roles, especially when providing resources not available in print. However, for maximizing leaning where mental focus and reflection are called for, educators shouldn't assume all media are the same, even when they contain identical words.8.What does the underlined phrase "shine through" in paragraph 2 mean?A.Seem unlikely to last.B.Seem hard to explain.C.Become ready to use.D.Become easy to notice.9.What does the shallowing hypothesis assume?A.Readers treat digital texts lightly.B.Digital texts are simpler to understand.C.People select digital texts randomly.D.Digital texts are suitable for social media.10.Why are audio and video increasingly used by university teachers?A.They can hold students' attentionB.They are more convenient to prepare.C.They help develop advanced skills.D.They are more informative than text.11.What does the author imply in the last paragraph?A.Students should apply multiple learning techniques.B.Teachers should produce their own teaching material.C.Print texts cannot be entirely replaced in education.D.Education outside the classroom cannot be ignored.In the race to document the species on Earth before they go extinct, researchers and citizen scientists have collected billions of records. Today, most records of biodiversity are often in the form of photos, videos, and other digital records. Though they are useful for detecting shifts in the number and variety of species in an area, a new Stanford study has found that this type of record is not perfect."With the rise of technology it is easy for people to make observations of different species with the aid of a mobile application, "said Barnabas Daru, who is lead author of the study and assistant professor of biology in the Stanford School of Humanities and Sciences. "These observations now outnumber the primary data that comes from physical specimens(标本), and since we are increasingly using observational data to investigate how species are responding to global change, I wanted to know: Are they usable?"Using a global dataset of 1. 9 billion records of plants, insects, birds, and animals, Daru and his team tested how well these data represent actual global biodiversity patterns."We were particularly interested in exploring the aspects of sampling that tend to bias(使有偏差)data, like the greater likelihood of a citizen scientist to take a picture of af lowering plant instead of the grass rightnext to it, "said Daru.Their study revealed that the large number of observation-only records did not lead to better global coverage. Moreover, these data are biased and favor certain regions, time periods, and species. This makes sense because the people who get observational biodiversity data on mobile devices are often citizen scientists recording their encounters with species in areas nearby. These data are also biased toward certain species with attractive or eye-catching features.What can we do with the imperfect datasets of biodiversity?"Quite a lot, "Daru explained." Biodiversity apps can use our study results to inform users of oversampled areas and lead them to places -and even species -that are not well-sampled. To improve the quality of observational data, biodiversity apps can also encourage users to have an expert confirm the identification of their uploaded image. "12.What do we know about the records of species collected now?A.They are becoming outdated.B.They are mostly in electronic formC.They are limited in numberD.They are used for public exhibition.13.What does Daru's study focus on?A.Threatened species.B.Physical specimens.C.Observational data D.Mobile applications14.What has led to the biases according to the study?A.Mistakes in data analysis.B.Poor quality of uploaded picturesC.Improper way of sampling.D.Unreliable data collection devices.15.What is Daru's suggestion for biodiversity apps?A.Review data from certain areas.B.Hire experts to check the records.C.Confirm the identity of the users.D.Give guidance to citizen scientists.二、第二节(共5小题;每小题2.5分,满分12.5分)(2024·新高考Ⅰ卷)阅读下面短文,从短文后的选项中选出可以填入空白处的最佳选项。

2024全国高考真题英语汇编:阅读理解D篇

2024全国高考真题英语汇编:阅读理解D篇

2024全国高考真题英语汇编阅读理解D篇一、阅读理解(2024·浙江·高考真题)The Stanford marshmallow (棉花糖) test was originally conducted by psychologist Walter Mischel in the late 1960s. Children aged four to six at a nursery school were placed in a room. A single sugary treat, selected by the child, was placed on a table. Each child was told if they waited for 15 minutes before eating the treat, they would be given a second treat. Then they were left alone in the room. Follow-up studies with the children later in life showed a connection between an ability to wait long enough to obtain a second treat and various forms of success.As adults we face a version of the marshmallow test every day. We’re not tempted by sugary treats, but by our computers, phones, and tablets — all the devices that connect us to the global delivery system for various types of information that do to us what marshmallows do to preschoolers.We are tempted by sugary treats because our ancestors lived in a calorie-poor world, and our brains developed a response mechanism to these treats that reflected their value — a feeling of reward and satisfaction. But as we’ve reshaped the world around us, dramatically reducing the cost and effort involved in obtaining calories, we still have the same brains we had thousands of years ago, and this mismatch is at the heart of why so many of us struggle to resist tempting foods that we know we shouldn’t eat.A similar process is at work in our response to information. Our formative environment as a species was information-poor, so our brains developed a mechanism that prized new information. But global connectivity has greatly changed our information environment. We are now ceaselessly bombarded (轰炸) with new information. Therefore, just as we need to be more thoughtful about our caloric consumption, we also need to be more thoughtful about our information consumption, resisting the temptation of the mental “junk food” in order to manage our time most effectively.1.What did the children need to do to get a second treat in Mischel’s test?A.Take an examination alone.B.Share their treats with others.C.Delay eating for fifteen minutes.D.Show respect for the researchers.2.According to Paragraph 3, there is a mismatch between_______.A.the calorie-poor world and our good appetites B.the shortage of sugar and our nutritional needsC.the tempting foods and our efforts to keep fit D.the rich food supply and our unchanged brains 3.What does the author suggest readers do?A.Be selective information consumers.B.Absorb new information readily.C.Use diverse information sources.D.Protect the information environment.4.Which of the following is the best title for the text?A.Eat Less, Read More B.The Later, the BetterC.The Marshmallow Test for Grownups D.The Bitter Truth about Early Humans(2024·全国·高考真题)In the race to document the species on Earth before they go extinct, researchers and citizen scientists have collected billions of records. Today, most records of biodiversity are often in the form of photos, videos, and other digital records. Though they are useful for detecting shifts in the number and variety of species inan area, a new Stanford study has found that this type of record is not perfect.“With the rise of technology it is easy for people to make observations of different species with the aid of a mobile application,” said Barnabas Daru, who is lead author of the study and assistant professor of biology in the Stanford School of Humanities and Sciences. “These observations now outnumber the primary data that comes from physical specimens (标本), and since we are increasingly using observational data to investigate how species are responding to global change, I wanted to know: Are they usable?”Using a global dataset of 1.9 billion records of plants, insects, birds, and animals, Daru and his team tested how well these data represent actual global biodiversity patterns.“We were particularly interested in exploring the aspects of sampling that tend to bias (使有偏差) data, like the greater likelihood of a citizen scientist to take a picture of a flowering plant instead of the grass right next to it,” said Daru.Their study revealed that the large number of observation-only records did not lead to better global coverage. Moreover, these data are biased and favor certain regions, time periods, and species. This makes sense because the people who get observational biodiversity data on mobile devices are often citizen scientists recording their encounters with species in areas nearby. These data are also biased toward certain species with attractive or eye-catching features.What can we do with the imperfect datasets of biodiversity?“Quite a lot,” Daru explained. “Biodiversity apps can use our study results to inform users of oversampled areas and lead them to places — and even species — that are not well-sampled. To improve the quality of observational data, biodiversity apps can also encourage users to have an expert confirm the identification of their uploaded image.”5.What do we know about the records of species collected now?A.They are becoming outdated.B.They are mostly in electronic form.C.They are limited in number.D.They are used for public exhibition.6.What does Daru’s study focus on?A.Threatened species.B.Physical specimens.C.Observational data.D.Mobile applications.7.What has led to the biases according to the study?A.Mistakes in data analysis.B.Poor quality of uploaded pictures.C.Improper way of sampling.D.Unreliable data collection devices.8.What is Daru’s suggestion for biodiversity apps?A.Review data from certain areas.B.Hire experts to check the records.C.Confirm the identity of the users.D.Give guidance to citizen scientists.(2024·全国·高考真题)Given the astonishing potential of AI to transform our lives, we all need to take action to deal with our AI-powered future, and this is where AI by Design: A Plan for Living with Artificial Intelligence comes in. This absorbing new book by Catriona Campbell is a practical roadmap addressing the challenges posed by the forthcoming AI revolution (变革).In the wrong hands, such a book could prove as complicated to process as the computer code (代码) thatpowers AI but, thankfully, Campbell has more than two decades’ professional experience translating the heady into the understandable. She writes from the practical angle of a business person rather than as an academic, making for a guide which is highly accessible and informative and which, by the close, will make you feel almost as smart as AI.As we soon come to learn from AI by Design, AI is already super-smart and will become more capable, moving from the current generation of “narrow-AI” to Artificial General Intelligence. From there, Campbell says, will come Artificial Dominant Intelligence. This is why Campbell has set out to raise awareness of AI and its future now — several decades before these developments are expected to take place. She says it is essential that we keep control of artificial intelligence, or risk being sidelined and perhaps even worse.Campbell’s point is to wake up those responsible for AI-the technology companies and world leaders—so they are on the same page as all the experts currently developing it. She explains we are at a “tipping point” in history and must act now to prevent an extinction-level event for humanity. We need to consider how we want our future with AI to pan out. Such structured thinking, followed by global regulation, will enable us to achieve greatness rather than our downfall.AI will affect us all, and if you only read one book on the subject, this is it.9.What does the phrase “In the wrong hands” in paragraph 2 probably mean?A.If read by someone poorly educated.B.If reviewed by someone ill-intentioned.C.If written by someone less competent.D.If translated by someone unacademic.10.What is a feature of AI by Design according to the text?A.It is packed with complex codes.B.It adopts a down-to-earth writing style.C.It provides step-by-step instructions.D.It is intended for AI professionals.11.What does Campbell urge people to do regarding AI development?A.Observe existing regulations on it.B.Reconsider expert opinions about it.C.Make joint efforts to keep it under control.D.Learn from prior experience to slow it down.12.What is the author’s purpose in writing the text?A.To recommend a book on AI.B.To give a brief account of AI history.C.To clarify the definition of AI.D.To honor an outstanding AI expert.(2024·全国·高考真题)“I didn’t like the ending,” I said to my favorite college professor. It was my junior year of undergraduate, and I was doing an independent study on Victorian literature. I had just finished reading The Mill on the Floss by George Eliot, and I was heartbroken with the ending. Prof. Gracie, with all his patience, asked me to think about it beyond whether I liked it or not. He suggested I think about the difference between endings that I wanted for the characters and endings that were right for the characters, endings that satisfied the story even if they didn’t have a traditionally positive outcome. Of course, I would have preferred a different ending for Tom and Maggie Tulliver, but the ending they got did make the most sense for them.This was an aha moment for me, and I never thought about endings the same way again. From then on, if I wanted to read an ending guaranteed to be happy, I’d pick up a love romance. If I wanted an ending I couldn’t guess, I’d pick up a mystery (悬疑小说). One where I kind of knew what was going to happen, historical fiction. Choosingwhat to read became easier.But writing the end — that’s hard. It’s hard for writers because endings carry so much weight with readers. You have to balance creating an ending that's unpredictable, but doesn’t seem to come from nowhere, one that fits what’s right for the characters.That’s why this issue (期) of Writer’s Digest aims to help you figure out how to write the best ending for whatever kind of writing you’re doing. If it’s short stories, Peter Mountford breaks down six techniques you can try to see which one helps you stick the landing. Elizabeth Sims analyzes the final chapters of five great novels to see what key points they include and how you can adapt them for your work.This issue won’t tell you what your ending should be — that’s up to you and the story you’re telling — but it might provide what you need to get there.13.Why did the author go to Prof. Gracie?A.To discuss a novel.B.To submit a book report.C.To argue for a writer.D.To ask for a reading list.14.What did the author realize after seeing Gracie?A.Writing is a matter of personal preferences.B.Readers are often carried away by character.C.Each type of literature has its unique end.D.A story which begins well will end well.15.What is expected of a good ending?A.It satisfies readers’ taste.B.It fits with the story development.C.It is usually positive.D.It is open for imagination.16.Why does the author mention Peter Mountford and Elizabeth Sims?A.To give examples of great novelists.B.To stress the theme of this issue.C.To encourage writing for the magazine.D.To recommend their new books.(2024·北京·高考真题)Franz Boas’s description of Inuit (因纽特人) life in the 19th century illustrates the probable moral code of early humans. Here, norms (规范) were unwritten and rarely expressed clearly, but were well understood and taken to heart. Dishonest and violent behaviours were disapproved of; leadership, marriage and interactions with other groups were loosely governed by traditions. Conflict was often resolved in musical battles. Because arguing angrily leads to chaos, it was strongly discouraged. With life in the unforgiving Northern Canada being so demanding, the Inuit’s practical approach to morality made good sense.The similarity of moral virtues across cultures is striking, even though the relative ranking of the virtues may vary with a social group’s history and environment. Typically, cruelty and cheating are discouraged, while cooperation, humbleness and courage are praised. These universal norms far pre-date the concept of any moralising religion or written law. Instead, they are rooted in the similarity of basic human needs and our shared mechanisms for learning and problem solving. Our social instincts (本能) include the intense desire to belong. The approval of others is rewarding, while their disapproval is strongly disliked. These social emotions prepare our brains to shape our behaviour according to the norms and values of our family and our community. More generally, social instincts motivate us to learn how to behave in a socially complex world.The mechanism involves a repurposed reward system originally used to develop habits important for self-care. Our brains use the system to acquire behavioural patterns regarding safe routes home, efficient food gathering and dangers to avoid. Good habits save time, energy and sometimes your life. Good social habits do something similar in a social context. We learn to tell the truth, even when lying is self-serving; we help a grandparent even when it is inconvenient. We acquire what we call a sense of right and wrong.Social benefits are accompanied by social demands: we must get along, but not put up with too much. Hence self-discipline is advantageous. In humans, a greatly enlarged brain boosts self-control, just as it boosts problem-solving skills in the social as well as the physical world. These abilities are strengthened by our capacity for language, which allows social practices to develop in extremely unobvious ways.17.What can be inferred about the forming of the Inuit’s moral code?A.Living conditions were the drive.B.Unwritten rules were the target.C.Social tradition was the basis.D.Honesty was the key.18.What can we learn from this passage?A.Inconveniences are the cause of telling lies.B.Basic human needs lead to universal norms.C.Language capacity is limited by self-control.D.Written laws have great influence on virtues. 19.Which would be the best title for this passage?A.Virtues: Bridges Across Cultures B.The Values of Self-disciplineC.Brains: Walls Against Chaos D.The Roots of Morality参考答案1.C 2.D 3.A 4.C【导语】这是一篇说明文。

人类基因组概况ppt课件

人类基因组概况ppt课件
A+T含量 G+C含量 不能确定的碱基 重复序列(不含异染色质) 编码序列(基因)数目 功能未知基因比例 外显子最多的基因 SNP数量 SNP密度
2.91Gbp
54% 38% 9% 35% 26588 42% Titin(234) 约300万个 1/12500 bp
最长的染色体 最短的染色体 基因最多的染色体 基因最少的染色体 基因密度最大的染色体 基因密度最小的染色体 重复序列含量最高的染色体
It is essentially immoral not to get it (the human genome sequence) done as fast as possible.
James Watson
人类基因组计划的完成,使得我们今天有可能来探 讨基因组的概,但我们仍然无法来谈论细节。
重复序列含量最低的染色体
编码外显子序列的比例 基因的平均长度
2(240 Mbp) Y(19 Mbp) 1(2453) Y(104) 19(23/Mb) 13,Y(5/Mb) 19(57%)
2,8,10,13,18(36%)
1.1~1.4% 27 Kb
女 平均 男
染色体上距着丝粒越远,重组率越高
4. Francis S. Collins, Eric D. Green, Alan E. Guttmacher, Mark S. Guyer :A Vision for the Future of Genomics Research. A blueprint for the genomic era. Nature Apr 24 2003: 835.
而 Celera 的测序样本来自5个人:分别属于西班牙裔、 亚洲裔、非洲裔、美洲裔和高加索裔(2男3女),是从21个志 愿者样本中挑选的。

英语专业-英美文学试卷及答案-期末

英语专业-英美文学试卷及答案-期末

英美文学试卷AI.Mark the following statements as true (T) or false (F).(10 x 1’=10’)1.( ) Chaucer is the first English short-story teller and the founder of English poetry as well as the founder of English realism.His masterpiece The Canterbury tales contains 26 stories.2.( ) English Renaissance is an age of essay and drama.3.( ) The rise of the modern novel is closely related to the rise of the middle class and an urbanlife.4.( ) The French Revolution and the American War of Independence were two big influencesthat brought about the English Romantic Movement.5.( ) Charlotte’s novels are all about lonely and neglected young women with a fierce longingfor life and love.Her novels are more or less based on her own experience and feelings and the life as she sees around.6.( ) The leading figures of the naturalism at the turn of 19th century are Thomas Hardy, John Galsworthy and Bernard Shaw.7.( ) Emily Dickinson is remembered as the “All American Writer”.8.( )The Civil War divides the American literature into romantic literature and realist literature.9.( ) Mark Twain is the first American writer to discover an American language and Americanconsciousness.10.( ) In the decade of the 1910s, American literature achieved a new diversity and reached itsgreatest heights.II.Fill in the blanks.(20 x 1’=20’)11.The most enduring shaping influence in American thought and American literature was ___________.12.The War of Independence lasted eight years till__________.13.Ralph Waldo Emerson's essay__________ has been regarded as "America's Declaration of Intellectual Independence". It called on American writers to write about America in a way peculiarly American.14.The American ___________ writers paid a great interest in the realities of life and described the integrity of human character reacting under various circumstances and pictured the pioneers of the Far West, the new immigrants and the struggles of the working class.The leading figures were ____________, ____________, ____________, ____________, etc.15.No period in American history is more eventful than that between the two world wars.The literary features of the time can be seen in the writings of those ________ writers as Ezra Pound, and the writers of the Lost Generation as ___________.16.Two features of English Renaissance are the curiosity for ___________ and the interest in the activities of _____________________.17.Shakespeare’s earliest great success in tragedy is ____________, a play of youth and love, with the famous balcony scene.18.There are three types of poets in 17th century English literature.They are Puritan poets, ___________ poets and ______________ poets.19.Pope’s An Essay on Criticism is a didactic poem written in ___________________.20.___________ has been regarded by some as “Father of the English Novel”for his contribution to the establishment of the form of the modern novel.21.“Beauty is truth, truth beauty”is an epigrammatic line by _______________.wrence’s most controversial novel is ___________, the best probably _________.III.Multiple choice.(20 x 1’=20’)23.Among the three major works by John Milton ________ is indeed the only generally acknowledged epic in English literature since Beowulf.A.Paradise RegainedB.Samson AgonistesC.LycidasD.Paradise Lost24. Francis Bacon’s essays are famous for their brevity, compactness and __________.plicityplexityC.powerfulnessdness25.As one of the greatest masters of English prose, _______ defined a good style as “proper words in proper places”.A.Henry FieldingB.Jonathan SwiftC.Samuel JohnsonD.Alexander Pope26.The Pilgrim’s Progress by John Bunyan is often said to be concerned with the search for _________.A.material wealthB.spiritual salvationC.universal truthD.self-fulfillment27.“It is a truth universally acknowledged that a single man in possession of a good fortune must be in want of a wife.”The quoted part is taken from _________.A.Jane EyreB.Wuthering HeightsC.Pride and PrejudiceD.Sense and Sensibility28.Which of the following poems is a landmark in English poetry?A.Lyrical Ballads by William Wordsworth and Samuel Taylor ColeridgeB.“I Wandered Lonely as a Cloud”by William WordsworthC.“Remorse”by Samuel Taylor ColeridgeD.Leaves of Grass by Walt Whitman29.The most distinguishing feature of Charles Dickens’works is his _________.A.simple vocabularyB.bitter and sharp criticismC.character-portrayalD.pictures of happiness30.“My Last Duchess”is a poem that best exemplifies Robert Browning’s ________.A.sensitive ear for the sounds of the English languageB.excellent choice of wordsC.mastering of the metrical devicese of the dramatic monologue31.________ is the most outstanding stream of consciousness novelist, with ______as hisencyclopedia-like masterpiece.A James Joyce, UlyssesB.E.M.Foster, A Passage to Indiawrence, Sons and loversD.Virginia Woolf, Mrs.Dalloway32.Which of the following comments on Charles Dickens is wrong?A.Dickens is one of the greatest critical realist writers of the Modern PeriodB.His serious intention is to expose and criticize all the poverty, injustice, hypocrisy andcorruptness he sees all around him.C.The later works show the development of Dickens towards a highly conscious artist of themodern type.D.A Tale of Two Cities is one of his late works.33._____was known as “the poets’poet”.A.William ShakespeareB.Edmund SpenserC.John DonneD.John Milton34.Which of the following poet belongs to the active Romantic poet?A.KeatsB.SoutheyC.WordsworthD.Coleridge35.______ is regarded today as the national epic of the Anglo-Saxons.A.BeowulfB.The Canterbury TalesC.Don JuanD.Paradise Lost36.___________ is the first modern American novel.A.Tom SawyerB.Huckleberry FinnC.The Sketch BookD.The Leatherstocking Tales37.Which of the following statements is NOT true of American Transcendentalism?A.It can be clearly defined as a part of American Romantic literary movement.B.It can be defined philosophically as “the recognition in man of the capacity of knowing truth intuitively”.C.Ralph Waldo Emerson was the chief advocate of this spiritual movement.D.It sprang from South America in the late l9th century.38.The theme of Washington Irving’s Rip Van Winkle is _________.A.the conflict of human psycheB.the fight against racial discriminationC.the familial conflictD.the nostalgia for the unrecoverable past39.The Nobel Prize Committee highly praised ________ for “his powerful style-forming mastery of the art”of creating modern diction.A.Ezra PoundB.Ernest HemingwayC.Robert FrostD.Theodore Dreiser40.Who exerts the single most important influence on literary naturalism?A.EmersonB.Jack LondonC.Theodore DreiserD.Darwin41.________ is NOT true in describing American naturalists.A.they were deeply influenced by DarwinismB.they were identified with French novelist and theorist Emile ZolaC.they chose their subjects for the lower ranks or societyD.they used more serious and more sympathetic tone in writing than realists42.Henry James’s fame generally rests upon his novels and stories with ________.A.international themeB.national themeC.European themeD.regional themeIV.Explain the following literary items.(4x 5’=20’)43.Spenserian Stanzake Poets45.Humanism46.BalladV.Questions.(3x 10’=30’)47.“Robinson Crusoe”is usually considered as Daniel Defoe’s masterpiece.Discuss why it became so successful when it was published?48.What is "Byronic hero"?49.Mark Twain and Henry James are two representatives of the realistic writers in American literature.How is Twain’s realism different form James’s realism?参考答案:I.Mark the following statements as true (T) or false (F).(本题共10空,每空1分,共10分)1-5: FFTTT 6-10: FFTTFII.Fill in the blanks.(本题共20小题, 每题1分, 共20分)11.(American) Puritanism12.178313.The American Scholar14.realistic; Mark Twain; Henry James; Jack London; Theodore Dreiser.15.Imagist; Hemingway.16.the classical literature; humanity.17.Romeo and Juliet18.Cavalier; Metaphysical19.heroic couplet20.Henry Fielding21.John Keatsdy Chatterley’s lover; The RainbowIV. Ex pla in the foll owi ng lite rar y ite ms.(本题4小题,每小题5分,共20分)43.Spenserian Stanza: it refers to a verse form created by Edmund Spenser for his poems.Each stanza has nine lines.Each of the first eight lines is in iambic pentameter, and the ninth line is an iambic hexameter line.The rhythm scheme is ababbcbccke Poets: it refers to those English romantic poets at the beginning of th e19th century, William Wordsworth, for example, who lived in the heart of the Lake District in the north-western part of England and enjoyed the experience of living close to nature, and these poets were the older generation of Romantic poets who had been deeply influenced by the French Revolution of 1789 and its effects.In their writings, they described the beautiful scenes and the country people of the area.45.Humanism refers to the literary culture in the Renaissance.Humanists emphasize the capacities of the human mind and the achievements of human culture.Humanism became the central theme of English Renaissance.Thomas More and William Shakespeare are the best representatives of the English humanists.46.Ballad: a story told in songs, usually in 4-line stanzas, with the second and fourth rhymed. V.Questions.(本题3小题,每小题10分,共30分)47.A: Robinson Crusoe is supposedly based on the real adventure of an Alexander Selkirk who once stayed alone on the uninhabited island for five year4s.Actually, the story is an imagination.B: In Robinson Crusoe, Defoe traces the growth of Robinson from a naïve and artless youth into a shrewd and hardened man, tempered by numerous trials in his eventful life.C.In the novel, Robinson is a real hero and he is an embodiment of the rising middle-class virtues in the mid-eighteenth century England.Robinson is a true empire-builder, a colonizer and a foreign trader, who has the courage and will to face hardships and who has determination to preserve himself and improve his livelihood by struggling against nature.D.Robinson Crusoe is an adventure story very much in the spirit of the time.Because of the above reasons, when it was published, people all liked that story, and it became an immediate success.48.Byronic hero is a proud, mysterious rebel figure of noble origin.With immense superiorityin his passions and powers, this Byronic hero would carry on his shoulders the burden of righting all the wrongs in a corrupt society, and would rise single-handedly against any kind of tyrannical rules wither in government, in religion, or in moral principles with unconquerable wills and inexhaustible energies.The conflict is usually one of rebellious individuals against outworn social systems and conventions.Such a hero appeared in many of his works, for example, "Don Juan".The figure is somewhat modeled on the life and personality of Byron himself, and makes Byron famous both at home and abroad.49.A.Mark Twain’s realism is tainted with local color, preferring to have his won region and people at the forefront of his stories.B.James’s realism is concerned with the “inner world”of man and the international theme.C.Twain’s language is simple and colloquial and he employs humor in his writing.D.James’s language is elaborate and refined with lengthy psychological analyses.。

学术英语习题及部分答案

学术英语习题及部分答案

Training and Practice for English for Academic PurposesPart I1.Discuss the following questions.What are basic principles the researchers must try to follow when they write their research papers? And would you please list some deadly sins a researcher must avoid when they want to publish a research paper? What are the main contents of a research paper?2. Translate the following Chinese introduction into English.提高起重机生产力和安全性的设备研究近些年来,就用研究人员对起重机(crane)的研究兴趣与日俱增。

起重机种类繁多,从樱桃采摘机(cherry pickers)到巨型塔式起重机(huge tower cranes) ,是建筑工地不可或缺的重要设备之一。

由于建筑用起重机工作环境多变(constantly changing working environment), 操作者(operator)责任重大(heavy reliance)。

过去几十年里,超重机技术日新月异,但是操作员与其他工种人员配合协作方面的技术发展缓慢。

起重机的发展步伐如此迅猛,我们似乎要问,在某些方面,是不是已经超出(outstrip)了人们安全使用的能力?本文旨在探讨如何通过新型设备的引进提高起重机生产力以及提出相关安全性的举措,进而为新型起重机的应用和案例提供新的思路。

In recent years, researchers have become more interested in crane research.The variety of cranes, from cherry pickers to giant tower cranes, is one of the most important equipment on construction sites.As a result of the changing working environment of the construction crane, operator is responsible for heavy reliance.Over the past few decades, the technology of overweight machines has been changing rapidly, but the operators have been slow to cooperate with other workers in collaboration.The pace of development of cranes is so rapid that we seem to be asking whether in some respects, the outstrip has exceeded the ability of people to safely use it.This paper aims to explore how to improve crane productivity and raise related security measures through the introduction of new equipment, so as to provide new ideas for the application and case of new cranes.3. You are writing a research paper entitled “The Effects of Radiation from the Sun on Life o n Earth”. In your introduction you need to review, in general terms, how the sun supports life on the earth. Prepare an Introduction section for your paper based on the information below.⏹Distance from the earth: 92,976,000 miles⏹The Sun’s energy comes from nuclear fusion of hydrogen to helium.⏹Intense radiation, including lethal ultraviolet radiation, arrives at the earth’s outer atmosphere.⏹Ozone in the stratosphere protects life on earth from excessive ultraviolet radiation.⏹The seasons of the earth’s climate results from (1) the 23.30tilt of the earth’s axis of rotation from the normal to the plane of the earth’s orbit around the Sun, (2) the large coverage area of water on the earth (about 75% of the earth’s surface), an d (3) the rotation of the earth with associated generation of jet-stream patterns.⏹Radiation passing through the earth’s atmosphere loses most short-wave radiation, butsome arriving at the surface is converted into infrared radiation which is then trapped by water vapor and other tri-atomic molecules in the troposphere and stratosphere, causing global warming.Life on earth is maintained from photosynthesis and conversion of carbon dioxide to oxygen by plants.4.Translate the following parts of sentences in Introduction into proper English.(1)过去对……的研究工作说明……The previous work on … has indicated that…(2)A在1932年做了关于……的早期研究。

(2024年高考真题含解析)2024年普通高等学校招生全国统一考试英语试卷 新课标Ⅰ卷(含解析)

(2024年高考真题含解析)2024年普通高等学校招生全国统一考试英语试卷 新课标Ⅰ卷(含解析)

2024年普通高等学校招生全国统一考试新课标Ⅰ卷英语试卷姓名________________ 准考证号________________全卷共12页,满分150分,考试时间120分钟。

养成良好的答题习惯,是决定成败的决定性因素之一。

做题前,要认真阅读题目要求、题干和选项,并对答案内容作出合理预测;答题时,切忌跟着感觉走,最好按照题目序号来做,不会的或存在疑问的,要做好标记,要善于发现,找到题目的题眼所在,规范答题,书写工整;答题完毕时,要认真检查,查漏补缺,纠正错误。

考生注意:1. 答题前,请务必将自己的姓名、准考证号用黑色字迹的签字笔或钢笔分别填写在试题卷和答题纸规定的位置上。

2. 答题时,请按照答题纸上“注意事项”的要求,在答题纸相应的位置上规范作答,在本试题卷上的作答一律无效。

第一部分听力(共两节,满分30分)做题时,先将答案标在试卷上。

录音内容结束后,你将有两分钟的时间将试卷上的答案转涂到答题纸上。

第一节(共5小题;每小题1.5分,满分7.5分)听下面5段对话。

每段对话后有一个小题,从题中所给的A、B、C三个选项中选出最佳选项。

听完每段对话后,你都有10秒钟的时间来回答有关小题和阅读下一小题。

每段对话仅读一遍。

例:How much is the shirt?A. £19.15.B. £9.18.C. £9.15.答案是C。

1.What is Kate doing?A.Boarding a flight. B.Arranging a trip. C.Seeing a friend off.2.What are the speakers talking about?A.A pop star. B.An old song. C.A radio program.3.What will the speakers do today?A.Go to an art show. B.Meet the man's aunt. C.Eat out with Mark.4.What does the man want to do?A.Cancel an order. B.Ask for a receipt. C.Reschedule a delivery.5.When will the next train to Bedford leave?A.At 9:45. B.At 10:15. C.At 11:00.第二节(共15小题;每小题1.5分,满分22.5分)听下面5段对话或独白。

Constraints on the decay of dark matter to dark energy from weak lensing bispectrum tomogra

Constraints on the decay of dark matter to dark energy from weak lensing bispectrum tomogra

2
B.M. Sch¨ afer, G.A. Caldera-Cabral and R. Maartens
model ΛCDM ΛΓ CDM φCDM φΓ CDM Ωm 0.25 0.25 0.25 0.25 σ8 0.8 0.8 0.8 0.8 ns 1 1 1 1 w0 -1 -1 −2 3 −2 3 wa 0 0 1 −3 1 −3 Γ 0 0
1 3 1 3
a very powerful observational probe for investigating the influence of dark energy on structure formation and the geometry of the universe (Schneider et al. 1992; Mellier 1999; Bartelmann & Schneider 2001; Refregier 2003), even in the nonlinear regime of structure formation (Jain & Seljak 1997; Bernardeau et al. 1997; Benabed & Bernardeau 2001). Lensing data is best used in tomographic measurements for constraining dark energy equation of state properties (Hu 1999, 2002; Heavens 2003; Jain & Taylor 2003), where one either measures the power spectrum or the bispectrum of a weak lensing quantity (Kilbinger & Schneider 2005; Schneider & Bartelmann 1997; Bernstein & Jain 2004; Dodelson & Zhang 2005). Supplementing the recent paper by La Vacca & Colombo (2008), who derived lensing bounds on interacting models from weak lensing power spectra, we focus on bispectrum tomography, and we use a more general, albeit phenomenological cosmological model. Bispectra have the advantage that the perturbative treatment is easier to carry out and that they are sensitive on the transition from linear to nonlinear dynamics in structure formation. After introducing the cosmological model and the peculiarities of gravitational lensing in the decaying CDM models in Sect. 2, we compute the weak lensing bispectrum and tomographic measurements in Sect. 3. Fisher-constraints on cosmological parameters are derived in Sect. 4 and the main results are summarised in Sect. 5. The parameter accuracies are forecast for the weak lensing survey proposed for the Dark UNiverse Explorer1 (DUNE). For the fiducial model, we take a spatially flat ΛCDM cosmology with w = −1, adiabatic initial conditions and stable CDM (Γ = 0). Specific parameter choices are H0 = 100h km/ s/Mpc with h = 0.72, Ωm = 0.25, σ8 = 0.8 and ns = 1.

Big-Bang Cosmology with Photon Creation

Big-Bang Cosmology with Photon Creation

a rXiv:as tr o-ph/978215v225Aug1997Big-Bang Cosmology with Photon Creation U.F.Wichoski ∗Department of Physics,Brown University,Providence,RI 02912,USA J.A.S.Lima †Departamento de F´ısica Te´o rica e Experimental,Universidade Federal do Rio Grande do Norte,59072-970,Natal -RN,Brazil The temperature evolution law is determined for an expanding FRW type Universe with a mixture of matter and radiation where “adiabatic”creation of photons has taken place.Taking into account this photon creation we discuss the physical conditions for having a hot big bang Universe.We also compare our results to the ones obtained from the standard FRW model.BROWN-HET-1089August 1997.astro-ph/9708215Typeset in REVT E XI.INTRODUCTIONIt is widely believed that matter and radiation need to be created in order to overcome some conceptual problems of the standard hot big-bang cosmology[1].The most popular approach accounting for the phenomenon of creation is based on the idea that the early Universe underwent an inflationary phase during which the temperature decreased nearly1028orders of magnitude.At the end of this supercooling process,the energy density of the inflatonfield was completely or almost completely converted into radiation,and the resulting Universe could have been reheated in less than one expansion Hubble time[2].However,there are theories where the gravitational particle creation phenomenon is conceived with no appealing for inflation and,consequently,allow the creation process to occur continuously in the course of the evolution.Probably,the best example is the adiabatic vacuum mechanism invented long ago by Parker and collaborators using the Bugoliubov mode-mixing technique in the context of quantumfield theory in curved spacetimes[3,4].However,this approach is plagued with several conceptual and mathematical difficulties.In particular,there is not a well-defined prescription of how the created matter and/or radiation should be incorporated in the Einsteinfield equations(EFE)[5].More recently,a new phenomenological macroscopic approach to gravitational creation of matter and radiation has attracted considerable attention[6]-[16].In this framework,the creation event of the inflationary scenario is also replaced by a continuous creation process.The crucial ingredients of this formulation are a balance equation for the number density of the created particles and a negative pressure term in the stress tensor so that the back-reaction problem present in Parker’s mechanism is naturally avoided.Another advantage of this formulation is that the laws of non-equilibrium thermodynamics were used since the very beginning,thereby leading to definite relations among the classical thermodynamic quantities.In particular,the creation pressure depends on the creation rate in a well defined form,and potentially may alter significantly several predictions of the standard big-bang pleting such an approach,a spectrum for blackbody radiation when photon creation takes place has also been proposed in the literature[16,17].This spectrum is preserved during a free expansion(for instance,after decoupling between matter and radiation),and more important still,it is compatible with the present spectral shape of the cosmic background radiation(CBR).On the other hand,in the photon-conserving Friedmann-Robertson-Walker(FRW)Universes,the temperature of the matter content follows the radiation temperature law when there is any thermal contact between these components. This state of affairs define what is called a hot big-bang ually,the condition that the Universe underwenta very hot phase in its beginning is expressed by requiring that there are many photons for each proton or neutron in the Universe today.This fact allows one to establish the cosmic eras,and is closely related to the high value of the radiation specific entropy(per baryon)in the present Universe.In this letter,by taking into account the photon creation process described by the thermodynamic formulation of irreversible processes,we analyze the temperature evolution law for the matter-energy content in the framework of a FRW metric.Our aim here is to discuss under which conditions the basic concept of hot big bang Universe remains valid when a continuous photon creation phenomenon is considered.II.CBR SPECTRUM AND THE TEMPERATURE LA W WITH PHOTON CREATIONLet us consider a spectrum of photons whose number and energy densities are,respectively,n r∼T3andρr∼T4 and let N r(t)be the instantaneous comoving total number of photons,where T is the temperature.Since N r=n r R3, where R(t)is the scale factor of a FRW cosmology,one may writeN r(t)−1N or)1N or )4c3ν3N or)1kT]−1.(3)In the absence of creation(N r(t)=N or),the standard Planckian spectrum is recovered[18].The derivation of the above spectrum depends only on the new temperature law and satisfies the equilibrium relationsn r(T)= ∞0ρT(ν)dνN or)1ρr(T)= ∞0ρT(ν)dν=aT4,(5) where b=0.24415¯h3c3,are the blackbody radiation constants.A gravitational photon creation process satisfying the above equilibrium relations has been termed“adiabatic”creation[8,16].The temperature law(1)implies that the exponential factor appearing in the spectrum given by Eq.(3)is time independent.As a consequence,the spectrum is not destroyed as the Universe evolves,at least not after the transition from an opaque to a transparent Universe.Note also that the above distribution cannot be distinguished from the blackbody spectrum at the present epoch when T=T o and N r(t o)=N or.In what follows we study under which conditions the temperature law(1)may be applied before decoupling,that is,during the time when matter and radiation were in thermal contact.Let us now consider a mixture of a non-relativistic gas in thermal contact with the blackbody radiation described by Eq.(3).For completeness we set up the basic equations including“adiabatic”creation of both components.For this system,the total pressure(p)and energy density(ρ)are given by(c=1)ρ=nm+(γ−1)−1nkT+aT4=ρm+ρr,(6)p=nkT+13n r Hψr,(8) andp mc=−ρm+p mn +3˙Rn,(10)and˙n rR =ψrdR(ρR3)=−3pR2.(12) Before proceeding further,it is worth noticing that the analysis of the temperature evolution law may be separated in several cases:(i)the standard model(ψr=ψm=0);(ii)photon creation(ψr=0,ψm=0);(iii)matter creation (ψr=0,ψm=0);and(iv)radiation and matter creation(ψr=0,ψm=0).In this letter we are primarily interested in the case of photon creation whose spectrum is defined by Eq.(3).Thus,henceforth we restrict our attention to the case(ii),for whichψm=p mc=0.Inserting Eq.(6)and Eq.(7)into Eq.(12)it follows that1dR nmR3+(γ−1)−1nkT R3+aT4R3 =−3nkT−aT4+4n r H.(13) Now,by considering that the number of massive particles is conserved,one obtains from Eq.(10)dT dT13nk,(16)andβ=ψrH,whereΓ=ψrfrom a kinetic theoretical approach or from a quantumfield theory.In any case,a reasonable upper limit to this rate isΓ=H,since for this value the photon creation rate exactly compensates for the dilution of particles due to expansion(see Eq.(11)).A new cosmological scenario will be thus obtained only if0<β≤1.In particular,for a radiation dominated model(p=1TdTT dT3n r H.(19)This equation can be rewritten asdT3N r −dR3T R=const,(21) which is the same temperature law for a freely propagating blackbody spectrum with photon creation(see Eq.(1)). Therefore,as long as the quantityσr is large,the radiative component will continue to overpower the material component.Hence,while there is any significant thermal contact between them,the matter temperature will follow Eq.(21)as well.This means that the condition for a hot big bang cosmology is not modified when“adiabatic”photon creation occurs.Note that the above result holds regardless of the value of theβparameter and assures the validity of the above equation during a considerable part of the evolution of the Universe.The important point here is that the cosmic eras are still viable using the above generalized temperature law.However,unlike in the standard model, the radiation specific entropy does not remain constant when photon creation takes place.Since n r∝T3the usual expressionσr=0.37n rN(22)remains valid nonetheless n r does not vary proportionally to R−3.As a consequence,the variation rate ofσr is directly proportional to the variation rate of N r(t),which in turn depends on the magnitude of theβparameter.So, ifβapproaches to zero,N r(t)andσr assume their constant values,and the standard model results are recovered.III.A SPECIFIC MODELWe consider the simplest photon creation model for which the parameterβis constant.This scenario can be defined by taking into account the“interaction”rateΓ=αH,whereαis a positive constant smaller than unity.As one may check from Eq.(18),in this case the temperature scaling law assumes the simple formT∝R−(1−α),(23) or still,in terms of the redshiftT=T o(1+z)1−α,(24) so that for z>0the Universe is cooler than the standard model.Considering the balance equation written as˙n r+3(1−α)n r H=0,we obtain,n r∝R−3(1−α),(25) which could be obtained directly from the number density-temperature relation(see Eq.(4)).As a consistency check, by substituting N r=n r R3in the expression above,it is easily seen that the temperature law also follows directly from the generalized expression Eq.(21).Replacing Eq.(25)into Eq.(22)one may see thatRσr=σor(,(27)(1+z)3αwhereσor∼108−9is the now observed radiation specific entropy.The above equations are a concrete example that the usual physical conditions defining a hot big bang cosmology may be weakened.In particular,specific entropy so large(and constant)as108is not required,providing that the product T R varies in the course of the evolution.In other words,108is only the present value of an increasing time-dependent quantity.Therefore,instead of predicting the currently observed value,the important question in thisframework is how a reasonable“initial value”,sayσr≃102,could be explained from thefirst principles.In particular, for the toy model presented here,it follows from equations(24)and(27),that a value ofσr≃102in the beginning of the nucleosynthesis epoch is possible only ifα≃0.18.More details on the nucleosynthesis of light elements,using a properly modified nucleosynthesis code which considers“adiabatic”creation of neutrinos and effectively massless species at nucleosynthesis epoch,will be discussed in a forthcoming communication[21].In conclusion,we have considered an evolutionary Universe where“adiabatic”photon creation has taken place.The conditions defining a big bang scenario with a blackbody spectrum endowed with photon creation and compatible with the present observed CBR distribution have been discussed.As we know,earlier approaches to matter creation processes,for instance,the steady state model,C-field theory,scale-covariant theory and others[22],fail the test of the CBR spectrum.As we have seen,this does not happen with the thermodynamic approach considered here. Naturally,in order to have a viable alternative to the photon-conserving FRW model,other cosmological properties need to be investigated.In particular,it would be important to use the Sachs-Wolf effect to test a big bang model with“adiabatic”photon creation.ACKNOWLEDGMENTSIt is a pleasure to thank Robert Brandenberger and Jackson Maia for their valuable comments.This work was partially supported by the US Department of Energy under grant DE-F602-91ER40688,Task A,(at Brown),and by Conselho Nacional de Desenvolvimento Cient´ıfico e Tecnol´o gico-CNPq(Brazilian Research Agency),(JASL).[11]W.Zimdhal and D.Pav´o n,Mon.Not.R.Astr.Soc.266,872(1994).[12]J.Triginer,W.Zimdhal and D.Pav´o n,Class.Quantum Grav.13,403(1996).[13]J.Gabriel,G.Le Denmat,Phys.Lett.A200,11(1995).[14]J.A.S.Lima,A.S.M.Germano and L.R.W.Abramo,Phys.Rev.D53,4287(1996).[15]L.R.W.Abramo and J.A.S.Lima,Class.Quantum Grav.13,2953(1996).[16]J.A.S.Lima,Phys.Rev.D54,2571(1996).[17]J.A.S.Lima,Cosmologies with Photon Creation and the3K Relic Radiation Spectrum,GRG(1997),in press.[18]The“adiabatic”creation of neutrinos may also be easily described by the same formalism.By choosing units such that¯h=k=c=1,the spectral distribution for a massless gas with g internal degrees of freedom can be written asρT(ω)=(N r(t)3gN or )1T−ε −1,where the numberεis+1for photons and−1for fermions.As expected,at early times,the thermal radiation energy density due to the relativistic particles at temperature T is given by the usual expression ρ=g∗π2。

Mining Decision Trees from Data Streams in a Mobile Environment £

Mining Decision Trees from Data Streams in a Mobile Environment £

Mining Decision Trees from Data Streams in a Mobile EnvironmentHillol Kargupta and Byung-Hoon ParkDepartment of Computer Science and Electrical Engineering1000Hillltop Circle,University of Maryland Baltimore CountyBaltimore,MD21250,USAhillol,park1@AbstractThis paper presents a novel Fourier analysis-based tech-nique to aggregate,communicate,and visualize decision trees in a mobile environment.Fourier representation of a decision tree has several useful properties that are partic-ularly useful for mining continuous data streams from small mobile computing devices.This paper presents algorithms to compute the Fourier spectrum of a decision tree and the vice versa.It offers a framework to aggregate decision trees in their Fourier representations.It also describes a touch-pad/ticker-based approach to visualize decision trees using their Fourier spectrum and an implementation for PDAs.1.IntroductionAnalyzing and monitoring time-critical data streams us-ing mobile devices in a ubiquitous manner is important for many applications infinance,defense,process control and other domains.These applications demand the ability to quickly analyze large amount of data.Decision trees(e.g., CART[4],ID3[21],and C4.5[22])are fast,and scalable. So decision tree-based data mining is a natural candidate for monitoring data streams from ubiquitous devices like PDAs,palmtops,and wearable computers.However,there are several problems.Mining time-critical data streams usually requires on-line learning that often produces a series of models[6,8,15, 23]like decision trees.From a data mining perspective it is important that these models are properly aggregated.This is because different data blocks observed at different time frames may generate different models that may actually be-long to a single model which can be generated when all the data blocks are combined and mined together.Even for de-cision trees[5,24]that are capable of incrementally modi-fying themselves based on new data,in many applications Patent application pending (e.g.,multiple data streams observed at different distributed locations)we end up with an ensemble of trees.Apart from better understanding of the model,communication of large number of trees over a wireless network also poses a major problem.We need on-line data mining algorithms that can easily aggregate and evolve models in an efficient represen-tation.Visualization of decision trees[1]in a small display is also a challenging task.Presenting a decision tree with even a moderate number of features in a small display screen is not easy.Since the number of nodes in a decision tree may grow exponentially with respect to the number of features defining the domain,drawing even a small tree in the dis-play area of a palmtop device or a cell phone is a difficult thing to do.Reading an email in a cell phone is sometimes annoying;so imagine browsing over a large number of tree-diagrams in a small screen.It simply does not work.We need an alternate approach.We need to represent trees in such a way that they can be easily and intuitively presented to the user using a small mobile device.This paper takes a small step toward that possibility.It considers manipulation and visualization of decision trees for mining data streams from small computing devices.It points out that Fourier basis offers an interesting representa-tion of decision trees that can facilitate quick aggregation of a large number of decision trees[9,18]and their visualiza-tion in a small screen using a novel“decision tree-ticker”. The efficient representation of decision trees in Fourier rep-resentation also allows quicker communication of tree en-sembles over low-bandwidth wireless networks.Although we present the material in the context of mobile devices,the approach is also useful for desktop applications.Section2explains the relation between Fourier repre-sentation and decision trees.It also presents an algorithm to compute the Fourier spectrum of a decision tree.Section 3considers aggregation of multiple trees in Fourier repre-sentation.Section4presents a decision tree visualization technique using a touch-pad and a ticker.Section5de-scribes an application of this technology for mining stockdata streams.Finally,Section6concludes this paper.2.Decision Trees as Numeric FunctionsThis paper adopts an algebraic perspective of decision trees.Note that a decision tree is a function that maps the domain members to a range of class labels.Sometimes,it is a symbolic function where features take symbolic(non-numeric)values.However,a symbolic function can be eas-ily converted to a numeric function by simply replacing the symbols with numeric values in a consistent manner.See Figure1for an example.A numeric function-representation of a decision tree may be quite useful.For example,we may be able to aggregate a collection of trees(often produced by ensemble learning techniques)by simply performing ba-sic arithmetic operations(e.g.adding two decision trees, weighted average)in their numeric ter in this paper we will see that a numeric representation is also suitable for visualizing and aggregating decision trees.Once the tree is converted to a numeric discrete function, we can also apply any appropriate analytical transformation that we want.Fourier transformation is one such possibility and it is an interesting one.Fourier basis offers an additively decomposable representation of a function.In other words, the Fourier representation of a function is a weighted linear combination of the Fourier basis functions.The weights are called Fourier coefficients.The coefficients completely de-fine the representation.Each coefficient is associated with a Fourier basis function that depends on a certain subset of features defining the domain of the data set to be mined. The following section presents a brief review of Fourier rep-resentation.2.1.A Brief Review of the Fourier BasisFourier bases are orthogonal functions that can be used to represent any function.Consider the function space over the set of all-bit Boolean feature vectors.The Fourier ba-sis set that spans this space is comprised of Fourier ba-sis functions;for the time being let us consider only dis-crete Boolean Fourier basis.Each Fourier basis function is defined as.Where and are bi-nary strings of length.In other words,and;denotes the inner product of and.can either be equal to1 or-1.The string is called a partition.The order of a partition is the number of1-s in.A Fourier basis func-tion depends on some only when.Therefore,a partition can also be viewed as a representation of a cer-tain subset of-s;every unique partition corresponds to a unique subset of-s.If a partition has exactly number of1-s then we say the partition is of order since the corre-sponding Fourier function depends on only those number of features corresponding to the1-s in the partition.A function,that maps an-dimensional space of binary strings to a real-valued range,can be written using the Fourier basis functions:;where is the Fourier coefficient corresponding to the partition;.The Fourier coefficient can be viewed as the relative contribution of the partition to the function value of.Therefore,the absolute value of can be used as the“significance”of the corresponding parti-tion.If the magnitude of some is very small compared to other coefficients then we may consider the-th partition to be insignificant and neglect its contribution.2.2.Fourier Analysis with Non-binary FeaturesThe Fourier basis can be easily extended to the non-Boolean case.Consider a domain defined by possibly non-Boolean features where the-th feature can take dis-tinct values.Let.The generalized Fourier basis function over an-ary feature space is defined as,(1) When we can write,.Where and are-ary strings of length.The Fourier coefficients for non-Boolean domain can be defined as follows:(2)where is the complex conjugate of.Since a decision tree is a function defined over a discrete space(inherently discrete or some discretization of a contin-uous space)we can compute its Fourier transformation.We shall discuss the techniques to compute the Fourier spec-trum of a decision tree later.It turns out that the Fourier representation of a decision tree with bounded depth has some very interesting properties[12,14].These observa-tions are discussed in the following section.2.3.Fourier Spectrum of a Decision Tree:Whybother?For almost all practical applications decision trees have bounded depths.The Fourier spectrum of a bounded depth decision tree has some interesting properties.1.The Fourier representation of a bounded depth(say)Boolean decision tree only has a polynomial number of non-zero coefficients;all coefficients corresponding to partitions involving more than feature variables are zero.The proof is relatively straight forward.Figure andthat simply2.If the order of a partition be its number of defining fea-tures then the magnitude of the Fourier coefficients de-cay exponentially with the order of the corresponding partition;in other words low order coefficients are ex-ponentially more significant than the higher order co-efficients.This was proved in[14]for Boolean deci-sion trees.Its counterpart for trees with non-boolean features can be found elsewhere[18].These observations suggest that the spectrum of the de-cision tree can be approximated by computing only a small number of low-order1coefficients.So Fourier basis offers an efficient numeric representation of a decision tree in the form of an algebraic function that can be easily stored,com-municated,and manipulated.2.4.From Fourier Coefficients to a Decision TreeFourier coefficients have important physical meanings. Recall that every coefficient is associated with a Fourier ba-sis function.Any given basis function depends on a unique subset of features defining the domain.For an-bit Boolean domain there are unique feature subsets.There are also different Fourier basis functions and each of them is asso-ciated with a unique subset.The magnitude of a coefficient 1Order of a coefficient is the number of features defining the corre-sponding partition.Low-order coefficients are the ones for which the or-ders of the partitions are relatively small represents the“strength”of the contribution of the corre-sponding subset of features to the overall function value. So if the magnitude of a coefficient is relatively large thenthe corresponding features together have strong influenceon the function value.For example,consider a linear func-tion of the form.All terms in this func-tion are linear;so the features together(in a multiplicative sense)do not make any contribution to the function value.This linearity is also reflected in its Fourier spectrum.Itis easy to show that all Fourier coefficients of this function corresponding to basis functions that depend on more thanone variable are zero.This connection between the struc-ture of the function and its spectrum is a general property of the Fourier basis.The magnitudes of the Fourier coef-ficients expose the underlying structure of the function by identifying the dependencies and correlation among differ-ent features.However,Fourier spectrum of a decision tree tells usmore than the interactions among the features.This coef-ficients can also tell us about the distribution of class labelsat any node of the decision tree.Recall that any node in the tree is associated with some feature.A downward linkfrom the node is labeled with an attribute value of this -th feature.As a result,a path from the root node to a suc-cessor node represents the subset of domain that satisfiesthe different feature values labeled along the path.Thesesubsets are essentially similarity-based equivalence classes. In this paper we shall call them schemata(schema in sin-gular form).If is a schema in a Boolean domain,then,where denotes a wild-card that matchesany value of the corresponding feature.For example,the path in Figure1(Bottom)represents the schema,since all members of the data subset at the final node of this path take feature values and for and respectively.The distribution of class labels in a schema is an impor-tant aspect since that identifies the utility of the schema as a decision rule.For example,if all the members of some schema has a class label value of1then can be used as an effective decision rule.On the other hand,if the propor-tion of label values1and0is almost equal then cannot be used as an effective decision rule.In decision tree learning algorithms like ID3and C4.5this“skewedness”in the dis-tribution for a given schema is measured by computing the information-gain defined in the following.where is the set of all possible values for at-tribute,and is the set of those members of that have a value for attribute;if and are the propor-tions of the positive and negative instances in some,then.The gain computation clearly depends on the proportion of class labels in a schema which can be determined directly from the Fourier spectrum of the tree.For example consider a problem with Boolean class labels.The total num-ber of members of schema with class labels1,(3)where,if*ifififLet be the total number of features that are set to spe-cific values in order to define the schema and be the total number of features defining the entire domain.So the total number of members covered by the schema is .The total number of members in with class label ,.Clearly,we can compute this distribu-tion information for any given schema using the spectrum of the tree.In other words,we can construct the tree using the Fourier spectrum of the information gain.The following section presents a fast technique for computing the Fourier spectrum of a decision tree.puting the Fourier Transform of a TreeThe Fourier spectrum of a decision tree can be easily computed by traversing the leaf nodes in a systematic fash-ion.The proposed algorithm is extremely fast and the over-head of computing these coefficients is minimal compared to the load for learning the tree.Let be the complete in-stance space.Let us also assume that there are leaf nodes in the decision tree and the-th leaf node covers a subset of ,denoted by.Therefore,.The-th coefficient of the spectrum can be computed as follows:Now note that both and take a constant value for every member in.Since the path to a leaf node represents a schema,with some abuse of symbols we can write,Where is a schema defined by a path to respectively. Further details about this algorithm can be found elsewhere [18].For each path from the root to a leaf node(schema h), all non-zero Fourier coefficient are detected by enumerating all possible value for each attribute in h.The running time of this algorithm is.3.Aggregating Multiple TreesThe Fourier spectrum of a decision tree-ensemble classi-fier can also be computed using the algorithm described in the previous section.Wefirst have to compute the Fourier spectrum of every tree and then aggregate them using the chosen scheme for constructing the ensemble.Let be an ensemble of different decision trees where the out-put is a weighted linear combination of the outputs of these trees.where and are decision tree and its weight re-spectively.is set of non-zero Fourier coefficients that are detected by decision tree and is a Fourier coefficient in.Now we can write,(4)where and.Therefore Fourier spectrum of(an ensemble clas-sifier)is simply the weighted spectrum of each tree.The spectrum of the ensemble can be directly useful for at least two immediate purposes:1.visualization of the ensemble through its Fourier spec-trum,2.minimizing the communication overhead,3.understanding the ensemble classifier through its spec-trum,and4.construction of simpler and smaller ensemble(possi-bly a single concise tree)from the aggregated spec-trum.The following section considers the visualization aspect. 4.Visualization of the Fourier RepresentationFourier representation offers a unique way to visualize decision trees.This section presents a ticker and touchpad-based approach to visualize decision trees that are espe-cially suitable for smaller displays like the ones that we getin palmtop devices.However,the approach is also suitablefor desktop applications.4.1.Fourier Spectrum-based Touch-pad and TickerFourier coefficients of a function provide us a lot of use-ful information.As we noted earlier,individual coefficients tell us about strongly mutually interacting features thatsignificantly contribute toward the overall function value.Moreover,different subsets of these coefficients can also provide us the distribution information(i.e.“information-gain”value used in ID3/C4.5)at any node.This section de-velops a novel approach for visualizing decision trees thatexploits both of these properties of the Fourier spectrum.The approach is based on two primary components:1.A touch-pad that presents a2D density plot of all thesignificant coefficients and2.a ticker that allows continuous monitoring ofinformation-gain produced by a set of classifiers con-structed by a certain group of features(possibly se-lected using the touch-pad).This is particularly im-portant for time-critical applications.Each of them is described in details next.4.2.The Touch-padThe touch-pad is a rectangular region that presentsa graphical interface to interact with the distribution of Fourier coefficients.Let us consider any arbitrary set ofcoefficients.These coefficients can be viewed as a graph where every node in is associated with a unique coefficient;is the set of edges where every edge betweennode and is associated with the“distance”() between their associated partitions.The distance metric() can be chosen in different ways.For example,if the domain is binary,partitions are also going to be binary strings.So we may choose hamming distance as the metric for defin-ing the graph.For discrete non-binary partitions we may chose any of the widely known distance metrics.The choice of distance metric does not fundamentally change the pro-posed approach.Any reasonable distance metric that cap-tures the distance between a pair of integer strings(i.e.par-titions)should workfine with the proposed visualization ap-proach.Once the graph is defined,our next task is to project it to a two dimensional space in such a way that “neighbors”in the original high dimensional space do re-main“neighbors”in the projected space.In other words, we would like to have near isometric projections where for all pairs and and constant value of;and are the projections of and respectively.This work makes use of existing off-the-shelf algorithms to con-struct this projection.It uses the widely known self organiz-ing feature map(SOM)[10,11,13]for this purpose.The SOM takes the Fourier coefficients and maps them to a two dimensional grid where coefficients with similar partitions are located in neighboring positions.Although,this par-ticular approach uses the SOM for the2-D projection,any other technique(possibly greedy algorithms)should work fine as long as the relative distance between the nodes are reasonably preserved.The touch-pad is also color coded in order to properly convey the distribution of significant and insignificant coefficients.Figure2shows a screen shot of the implementations of the touch-pad in a HP Jornada palmtop device.The touch-pad is also interactive.Since the display area is usually small our implementation of the touch-pad of-fers variable degrees of resolutions.The user can interac-tively zoom-in or zoom-out of a specific region.The user can alsofind out the subset of features defining a certain part of the touch-pad by interactively choosing a region from the screen.By marking a certain region of interest,the user will get a better understanding about interesting patterns among features residing and theirfiner and more detailed visual-ization.This interactive feature of the touch-pad also gives the user an opportunity to explore the dependencies among selected subset of features with the aid of a dynamically changing“ticker”.The following section describes this in detail.4.3.The TickerThe ticker is provided to allow continuous monitoring of specific classifiers produced by the decision trees.This is particularly important for time-critical applications.The ticker shows the“strengths”(measures like information-gain used frequently in the decision tree literature)for a subset of“good”decision rules produced by the trees.Fig-ure2shows screen shot of the ticker depicting the strengths of a set of classifiers at a particular time.As new data arrive the strengths are modified.It is inherently designed to complement the touch-pad-based approach to visualize the spectrum of the decision tree.Recall that the information-gain(i.e.difference in the entropies)at a particular node of a decision tree is computed by using the distribution of the domain members that it cov-ers.As we noted earlier,the information-gain can also be computed from the spectrum of the decision er’s se-lection of a certain region from the touch-pad identifies the current interest of the user to a certain subset of features. The user can invoke the ticker and instruct it to monitor all the decision rules that can be constructed using those fea-tures.The user can also interactively select the decisionFigure2.(Top)Fourier spectrum-based de-cision tree mining interface for HP Jornada.(Bottom)An enlarged view of the interface.The interface has three windows.The left-most window is the touch-pad.The bottomwindow shows the ticker.The right window is a small area to interactively construct and compute the accuracy of the classifiers.rules that are comprised of different features.The ticker in turn collects the relevant Fourier coefficients and computes the information-gain associated with all the decision rules that can be constructed using those features and shows only the good ones.5.Experiments with Financial Data StreamsThis section presents the experimental performance of the proposed aggregation approach for combining multiple decision trees.It presents results using two ensemble learn-ing techniques,namely Bagging and Arcing.The ensemble learning literature considers different ways to compute the output of the ensemble.Averaging the outputs of the individual models with uniform weight is probably the simplest possibility.Perrone and Cooper [19][17]refer to this method as Basic Ensemble Method (BEM)or naive Bagging.Breiman proposed an Arcing method Arc-fx[2][3]for mining from large data set and stream data.It is fundamentally based on the idea of Arc-ing–adaptive re-sampling by giving higher weights to those instances that are usually mis-classified.We consider both of these ensemble learning techniques to create an ensem-ble of decision trees from the data stream.These trees are however combined using the proposed Fourier spectrum-based approach which the regular ensemble learning tech-niques do not offer.We also performed experiments us-ing an AdaBoost-based approach suggested elsewhere[7]. However,we choose not to report that since its performance appears to be considerably inferior to those of Bagging and Arcing for the data set we use.Not all the models generated from different data blocks should always be aggregated together.Sometimes we may want to use a pruning algorithm[16,20]for selecting the right subset of models.Sometimes a“windowing scheme”[7,3]can be used where only most recent classifiers are used for learning and classification.Our experiments were performed using a semi-synthetic data stream with174Boolean attributes.The objective is to continuously evolve a decision tree-based predictive model for a Boolean attribute.The data-stream generator is es-sentially a C4.5decision tree learned from three years of S&P100and Nasdaq100stock quote data.The original data are pre-processed and transformed to discrete data by assigning discrete values for“increase”and“decrease”in stock quote between consecutive days(i.e.local gradient). Decision trees predict whether the Yahoo stock is likely to increase or decrease based on the attribute values of the stocks.We assume a non-stationary sampling strategy in order to generate the data.Every leaf in the decision tree-based data generator is associated with a certain probability value. Data samples are generated by choosing the leaves accord-ing to the assigned probability distribution.This distribu-tion is changed several times during a single experiment. We also add white noise to the generator.Test data set is comprised of instances.We implemented the naive Bagging and Arcing tech-niques and performed various tests over the validation data set as described below.Decision tree models are gener-ated from every data block collected from the stream and their spectrums are combined using the BEM and Arc-ing.We studied the accuracy of each model with var-ious sizes()of data block at each update.We use.We also studied the accuracy of each model generated using the“windowing”technique with various window sizes,.All the re-sults are measured over iterations where every iteration corresponds to a unique discrete time step.Figure3plots the classification accuracies of Bagging and Arcing with various data block sizes.Bagging con-verges rapidly with all different block sizes before or around 50iterations,while Arcing shows gradual increase through-out all iteration steps.Although we stopped at300itera-A c c u r a c y (%)UpdatesAccuracy of Bagging with Various Block Sizes vs UpdatesA c c u r a c y (%)UpdatesAccuracy of Arcing with Various Block Sizes vs UpdatesFigure 3.Accuracy vs.iterations with vari-ous data block sizes:(Top)Bagging (Bottom)Arcing.tions,the accuracy does not seem to reach its asymptotic value.Figure 4plots the classification accuracies of Bag-ging and Arcing with various window size.With relatively large window size (80and 100),we observed small decrease in accuracies in both cases.Figure 5plots the classification accuracies of Bagging and Arcing using different signifi-cant fractions of the Fourier spectrum.It shows that only a small fraction of all the coefficients in the spectrum is suffi-cient for accurate classification.Further details about these and additional experiments studying the complexity of the Fourier representations of decision trees can be found else-where [18].6.ConclusionThe emerging domain of wireless computing is alluding the possibility of making data mining ubiquitous.However,this new breed of applications is likely to have different ex-pectations and they will have to work with different system resource requirements.This will demand dramatic changes in the current desktop technology for data mining.This pa-per considered a small but important aspect of this issue.This paper presented a novel Fourier analysis-based ap-proach to enhance interaction with decision trees in a mo-bile environment.It observed that a decision tree is a func-5055606570758085050100150200250A c c u r a c y (%)UpdatesAccuracy of Bagging with Different Window Sizes vs UpdatesW=50W=80W=1005055606570758085050100150200250A c c u r a c y (%)UpdatesAccuracy of Arcing with Different Window Sizes vs UpdatesW=50W=80W=100Figure 4.Accuracy vs.iterations with vari-ous window sizes:(Top)Bagging (Bottom)Arcing.tion and its numeric functional representation in Fourier ba-sis has several utilities;the representation is efficient and easy to compute.It is also suitable for aggregation of multiple trees frequently generated by ensemble-based data stream mining techniques like boosting and bagging.This approach also offers a new way to visualize decision trees that is completely different from the traditional tree-based presentation used in most data mining software.The ap-proach presented here is particularly suitable for portable applications with small display areas.The touch-pad and ticker-based approach is very intuitive and can be easily implemented on touch sensitive screen often used in small wireless devices.We are currently extending the ticker in-terface,introducing different additional application func-tionalities,and exploring related techniques to mine real-valued financial data online.AcknowledgmentsThe authors acknowledge supports from the United States National Science Foundation CAREER award IIS-0093353and NASA (NRA)NAS2-37143.。

傅立叶光学导论英文版教学设计

傅立叶光学导论英文版教学设计

Fourier Optics Introduction Teaching Design - EnglishVersionIntroductionFourier optics is a branch of optics that deals with the mathematical analysis of light propagation. It is an essential subject for students pursuing a degree in optics, physics, or engineering. This teaching design ms to provide an overview of Fourier optics for undergraduate and graduate students.Course ObjectivesThe course objectives are as follows:•Introduce students to the foundations of Fourier optics•Teach students how to apply Fourier transforms and diffraction theory in optics•Demonstrate how Fourier optics is used in real-life applications such as holography and image processingCourse OutlineWeek 1: Introduction to Fourier Optics•Review of Maxwell’s equations•Overview of Fourier transforms and their applications in optics•Introduction to the fundamental principles of diffraction theoryWeek 2: Diffraction Theory•Detled discussion of diffraction phenomena and their mathematical models•Analysis of diffraction patterns and image formation using the transfer function concept•Introduction to coherence theory and its role in optical imagingWeek 3: Fourier Transform Optics•Laplace transform and its applications in optical system analysis•Fourier transform properties and their applications in optical signal processing•Introduction to imaging systems based on Fourier transform conceptsWeek 4: Holography and Applications•Principles of holography and holographic recording•Fourier optics analysis of holographic image formation•Applications of holography in imaging, non-destructive testing, and data storageWeek 5: Image Processing•Fourier analysis of image processing operations•Image filtering, restoration, and enhancement using Fourier optics techniques•Overview of modern image processing techniques based on Fourier transform conceptsWeek 6: Review and Assessment•Course review and discussion of open research questions in Fourier optics•Final examCourse Materials•Textbook: Goodman, J. W. Introduction to Fourier Optics.Roberts & Company Publishers, 2005.•Lecture notes and course handouts•MATLAB and Mathematica software for signal analysis and simulationTeaching MethodologyThe course will be taught using a combination of lectures, problem-solving sessions, and laboratory experiments. Students will be encouraged to participate in discussions and ask questions about course content and concepts. The course will be assessed through problem assignments, a mid-term exam, and a final exam.ConclusionFourier optics is a complex but essential subject in optics and physics. This teaching design provides a comprehensive overview of Fourier optics for undergraduate and graduate students. The course materials and teaching methodology are designed to help students develop a deep understanding of Fourier optics and its applications in real-life situations.。

  1. 1、下载文档前请自行甄别文档内容的完整性,平台不提供额外的编辑、内容补充、找答案等附加服务。
  2. 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
  3. 3、如文档侵犯您的权益,请联系客服反馈,我们会尽快为您处理(人工客服工作时间:9:00-18:30)。

a rX iv:mat h /654v2[mat h.AG ]16Dec22A Fourier-Mukai approach to spectral data for instantons Marcos Jardim University of Massachusetts at Amherst Department of Mathematics and Statistics Lederle Graduate Research Tower Amherst,MA 01003-9305USA Anthony Maciocia Department of Mathematics and Statistics University of Edinburgh James Clark Maxwell Building King’s Buildings Mayfield Road Edinburgh,EH93JZ,UK February 1,2008Abstract We study SU (r )instantons on elliptic surfaces with a section and show that they are in one-one correspondence with spectral data con-sisting of a curve in the dual elliptic surface and a line bundle on that curve.We use relative Fourier-Mukai transforms to analyse their prop-erties and,in the case of the K3and abelian surface,we show that the moduli space of instantons has a natural Lagrangian fibration structure with respect to the canonical complex symplectic structures.1IntroductionThe mathematical study of gauge theory was born some25years ago,and one of thefirst main results is just as old,namely the ADHM construc-tion of instantons over the Euclidean4-space.Since then,several different types of Euclidean instantons have been studied from various points of view: monopoles,calorons,Higgs pairs,doubly-periodic instantons,to name a few. The common feature in the study of such objects is the so-called Nahm trans-form.It relates instantons on R4which are invariant under a subgroup of translationsΛ⊂R4to instantons on the dual(R4)∗which are invariant under the dual subgroup of translationsΛ∗(see[15,Section7]for a de-tailed exposition).The Nahm transform has many interesting properties, but perhaps its greatest virtue is that it often converts the difficult problem of solving nonlinear PDE’s into a simpler problem involving only ODE’s or just vector spaces and linear maps between them.More recently,the so-called Fourier-Mukai transforms have generalised the Nahm transform,bringing it to the realms of algebraic geometry and derived categories.In this paper,we shall study a particular version of such transforms.It is defined for torsion-free sheaves E over relatively minimal elliptic surfaces Xπ→B with a section,transforming them into torsion sheaves on J Xˆπ→B,the relative Jacobian of X,which are supported over the spectral curves studied by Friedman-Morgan-Witten[6,7]for physical reasons.These constructions are briefly reviewed in Sections2and3,and we observe that there is a correspondence between instantons on a4-torus, spectral curves with line bundles on it,and instantons on the dual4-torus.The bulk of the paper is contained in Section5,where we compare the µ-stability of E with the concept of stability introduced by Simpson[16]for its transform.In particular,we show that E isµ-stable and locally-free if and only if its transform is stable in the sense of Simpson.As an application of these ideas,we shed new light into the moduli space of irreducible SU(r)-instantons over elliptic K3and abelian surfaces, showing in Sections6and7that it has the structure of a complex Lagrangian fibration.The case of rank2instantons on an elliptic surface is treated in Friedman’s book[5].We extend the results to higher rank and provide some more details about the structure of thefibration.The proofs are greatly simplified using the Fourier-Mukai technology and this enables us to go further in our description of the moduli spaces.Notation.The elliptic surface X will be polarised byℓwhich can be any polarisation(for the product of elliptic curves it is convenient to choose the sum of the elliptic curves).If F x denotes thefibre of Xπ→B passing through x∈X,thenˆπ−1(π(x))=ˆFπ(x)is the dual of F x.Let J X denote the dual elliptic surface parametrisingflat line bundles on thefibres ofπ.It alsofibres over B and has a canonical section given by the trivial bundles.We shall denote itsfibres byˆF b,for b∈B.Given a sheaf E on X,we write its Chern character as a triple:ch(E)=(rank(E),c1(E),ch2(E)),where ch2(E)=1∂A on E;we denote by E the associated holomorphic vector bundle.Given a holomorphic vector bundle E→X as above,we say that E is genericallyfibrewise semistable if its restriction to a generic ellipticfi-bre is semistable;E is said to befibrewise semistable if its restriction to every ellipticfibre is semistable.We shall also say that E is regular if h0(Fˆπ(p),E(p))≤1for all p∈J X.Observe that regularity is a generic condition.Now assume that E is a regular holomorphic vector bundle over X;we define the instanton spectral curve with respect to the projection Xπ→B in the following manner:S={p∈J X|h1(Fˆπ(p),E(p))=0}(1) It is not difficult to see that Sπ→B is a branched r-fold covering map.To define the second part of the instanton spectral data,recall that χ(E(p))=0for every p∈J X.We define a line bundle L on S by attaching the vector space H1(Fˆπ(p),E(p))to the point p∈S.Alternatively,consider the diagram:τ1X×B J X X×B Sτ2whereτ1is the inclusion map andτ2,σare the obvious projections.We define L as follows:L=R1σ∗(τ∗2E⊗τ∗1P)(3) Geometrically,it is not difficult to see that L can be regarded as a bundle of cokernels associated to a family of coupled Dirac operators parametrised by S.For the simple case X=T×P1,see[10].This follows from the natural identification between H1(Fˆπ(p),E(p))and the cokernel of the Dolbeault operatorLet us now recall the details of the definition of stability(due to Mumford and Takemoto)of torsion-free sheaves:Definition.A torsion-free sheaf E over a polarised surface(X,ℓ)is said to beµ-stable(respectively,µ-semistable)if for all proper subsheaves F of E,µ(F)<µ(E)(respectively,µ(F)≤µ(E)),whereµ(•)=c1(•)·ℓ/r(•)is called the slope.Recall that since we are working over surfaces,we can assume that the destabilizing subsheaf is locally-free and that the quotient E/F is torsion-free.Moreover,we also may choose F to beµ-stable.If E isµ-stable and locally-free,that is if E is a holomorphic vector bundle arising from an irreducible instanton connection via the Hitchin-Kobayashi correspondence,it is easy to see that E isΦ-WIT1.Hence Φ(E)=Φ1(E)is not a complex of sheaves,but simply a sheaf on J X. This also applies if E is not necessarily locally-free but is still torsion-free, as we shall see this in Section5.The following Proposition brings together the instanton spectral data described in the previous Section with the Fourier-Mukai transform.Proposition1.Let E be a vector bundle with an irreducible SU(r)instan-ton connection over an elliptic surface X.Let E be the associatedµ-stable, regular holomorphic vector bundle with c1(E)=0.ThenΦ1(E)is a torsion sheaf supported over the instanton spectral curve S⊂J X.Moreover,the restriction ofΦ1(E)to its support is naturally isomorphic to L.Proof.Let j:S→J X be the inclusion map.Thisfits into a commuting diagram of maps introduced in(2):X×B Sτ1X×B J XˆπJ XThenj∗L∼=R(j∗σ∗)(τ∗2E⊗τ∗1P)∼=Rˆπ∗Rτ1∗(τ∗2E⊗τ∗1P)∼=Rˆπ∗(P⊗Rτ1∗τ∗2 E).Butτ2=πτ1(see diagram2),so Rτ1∗τ∗2E∼=π∗E⊗O X×B S .Hence,we havea natural mapΦ1(E)→j∗L.Since thefibres ofΦ1(E)and j∗L are naturally isomorphic,this must be an isomorphism.Hence,j∗Φ1(E)=j∗j∗L=L as required.4Instantons on flat 4-toriLet T be a flat 4-dimensional torus,with a fixed complex structure.Denote by ˆTthe corresponding dual torus,which inherits a flat metric and a complex structure from T polarized by the associated K¨a hler class.We start by considering an irreducible SU (r )bundle E →T with c 1(E )=0and c 2(E )=k ,plus an instanton connection A .Equivalently,we can adopt an algebraic geometric point of view and look at the associated µ-stable holomorphic vector bundle E →T of rank r and the same Chern classes.Stability now implies the irreducibility hypothesis,which in turns implies that h 0(T ,E )=h 2(T ,E )=0,so that h 1(T ,E )=k .The Nahm transformed bundle ˆE →ˆT is then constructed as follows.Let P →T ׈T be the Poincar´e line bundle,and consider the natural projection maps:T p ←−T ׈T ˆp −→ˆT (6)Then ˆE =R 1ˆp ∗(p ∗E ⊗P ∨).The fibre of ˆE over ξ∈ˆT can be canoni-cally identified with ˆE ξ=H 1(T ,E ⊗L ξ),where L ξ→T denotes the flat holomorphic line bundle associated with ξ∈ˆT.Conversely,it can be shown that E =R 1p ∗(ˆp ∗ˆE ⊗P ),with fibres canon-ically identified with E z =H 1(ˆT,ˆE ⊗L z ),where L z →ˆT denotes the flat holomorphic bundle associated with z ∈T .Theorem 2.ˆE is a µ-stable holomorphic vector bundle of rank k ,flat deter-minant and c 2(ˆE)=r .Therefore,since it is invertible,the Nahm transform is a bijective correspondence M T (r,k )→M ˆT (k,r ).For the proof,we refer to [3].The algebro-geometric version is given in[12].In particular,in the k =1case,the Theorem tells us that there are no stable holomorphic bundles with this Chern character,thus precluding the existence of instantons with unit charge.For larger k ,this result is not really helpful as it only converts the problem of constructing instantons/stable bundles over T into the same problem over ˆT.A better understanding of the moduli space of SU (r )instantons on T with k >1requires a new tool,namely the Relative Fourier-Mukai transform introduced in the previous Section.Now assume that T =V ×W ,i.e.T is given by the product of two elliptic curves V and W .We shall denote by ˆV and ˆW the respective Jacobian curves,so that the dual torus is given by ˆT=ˆV ׈W .This case is particularly interesting because we have two fibration structures leading apparently to two distinct spectral curves,and we would like to understand the relation between them.Regarding T as an elliptic surface over V ,consider the diagram:V ×W V ×W ׈WπV V ׈W (7)where πV is the projection onto the first and second factors,while ˆπV is the projection onto the first and third factors.Let also P W →W ׈W be the Poincar´e line bundle.ThenL =R 1ˆπV ∗(π∗V E ⊗P W )(8)is a torsion sheaf on V ׈W ,supported over the instanton spectral curve S .On the other hand,regard ˆTas an elliptic surface over ˆW and consider the Nahm transformed bundle ˆE→ˆT .Based on the diagram:ˆV ׈W V ×V ׈V ׈W πW V ׈W (9)where πW is the projection onto the second and third factors,while ˆπW is the projection onto the first and third factors.Let also P V →V ׈V be the Poincar´e line bundle.ThenN =R 1ˆπW ∗(π∗W E ⊗P V )(10)is a torsion sheaf on V ׈W,supported over the dual instanton spectral curve R .Since the bundles E and ˆEare related via Nahm transform,it seems natural to ask how are the two sets of spectral data (S,L )and (R,N )are related to one another.Regarding T as an elliptic surface over V ,we have just seen that the spectral data (S,L )is encoded into the torsion sheaf Φ1(E )on V ׈W .On the other hand,regarding ˆTas an elliptic surface over ˆW let us denote the Fourier-Mukai transform of ˆE by Ψ(ˆE ).It is a torsion sheaf on V ׈W encoding the dual spectral data (R,N ).Let also ˆΨdenote the inverse of Ψ.We are finally in a position to prove our first main result:Theorem 3.The spectral pairs (S,L )and (R,N )are equivalent,in the sense that L and N can be canonically identified as sheaves on J X .Proof.If we define the functor F :D (T )→D (ˆT)as follows:F (E )=R ˆp ∗(p ∗E ⊗P ∨)where p and ˆp are the projection maps in (6),then ˆE=F 1(E ).The functor F is Mukai’s original Fourier transform introduced in [13].Using this notation,it is enough to show that ˆΨ◦Φ=F .To see this it suffices to show that ˆΨ(Φ(O x ))=ˆΨ(P x )=P x over ˆT .But P x ⊗P ∨y is supported at the intersection of translates of V and ˆW .Hence,P x is Ψ-WIT 0and ˆΨ(P x )is a flat line bundle.On the other hand,the properties of Ψand Φimply that translating x to y twists ˆΨ(P x )by P y −x and so by normalizing the Poincar´e bundles appropriately,we have ˆΨ(P x )=P x .Similarly,one could regard T as an elliptic surface over W and ˆT as an elliptic surface over ˆV .The analogue of Theorem 3would again hold,so that the corresponding spectral data would also coincide as torsion sheaves in ˆV ×W .More generally,one has the following commuting diagram of derived categories:D (V ×W )ΦF N N N N N N N N N N N D (V ׈W )ΨD (ˆV ׈W )(11)Of course,the functor Υis given by the Fourier-Mukai transform regard-ing T as an elliptic surfaceover W ,while the functor Ξis the Fourier-Mukaitransform regarding ˆT as an elliptic surface over ˆV .Remark .One concludes from Theorem 3that there is an equivalence be-tween the following three objects:instantons on a flat 4-torus (i.e.µ-stable bundles over T ),spectral data,and instantons on the dual 4-torus (i.e.µ-stable bundles over ˆT ).Such “circle of ideas”has been previously establishfor monopoles [9],doubly-periodic instantons [10]and periodic monopoles[2].Indeed,one expects that a similar scheme will hold for all translation invariant instantons on R 4.instantons over T NahmIn particular,a torsion sheaf which is supported on an irreducible curve and whose restriction has rank one is automatically p-stable with respect to any polarisation.Now let X→B be a(relatively minimal)elliptic surface with a section σ;let E be a torsion-free sheaf on X with ch(E)=(r,0,−k).Proposition4.Suppose that the restriction of E to some smoothfibre F s is semistable.Then E isΦ-WIT1and L=Φ1(E)has Chern character (0,kˆf+rˆσ,0),whereˆσis the class given by the zero section of J X.In particular,p(L,n)=n.Proof.First note that if E is torsion-free thenΦ0(E)=ˆπ∗(π∗E⊗P)is also torsion-free.Hence its support consists is either empty or consists of the whole relative jacobian surface J X.Moreover,if E is notΦ-WIT1then the support ofΦ0(E)is contained in the support ofΦ1(E)sinceχ(E|F)=0xfor all x∈X.However,Φ1(E)cannot be supported on the whole surface since it doesis semistable.ThusΦ0(E)=0 not contain the wholefibreˆF s where E|Fsand E isΦ-WIT1.NowΦ1(O X)is the trivial line bundle supported on the zero section of J X andΦ0(O x)is aflat line bundle supported on(a divisor in)ˆf.Hence, since ch(E)=r·ch(O X)+k·ch(O x),we conclude that:ch(Φ1(E))=r·ch(Φ1(O X))+k·ch(Φ0(O x))=(0,kˆf+rˆσ,0)as required.Note that the rank andfibre degrees of L are given by r(L)=c1(E)·f=0 and c1(L)·ˆf=r from the definition of the relative transform.Let S denote the support ofΦ1(E);the statement above does not im-ply that S is a divisor in J X,since it might contain some0-dimensional components(i.e.isolated points).Lemma5.If E is genericallyfibrewise semistable,then S=suppΦ1(E)has no0-dimensional components.Moreover,L=Φ1(E)is of pure dimension one.Proof.Suppose that q∈S is an isolated point.It cannot belong to afibre ˆFx such that E|F x is semistable,since the restriction to every nearbyfibre is also semistable and the restriction of E to thefibres varies holomorphically.is unstable.But S contains all suchfibres,since h1(Fˆπ(p),E(p))> So E|Fq0for all p∈ˆπ−1(π(q)).To obtain the second statement,we must show that L has no proper subsheaves supported on points.For a contradiction,suppose it does,and let F be a subsheaf of L with0-dimensional support.ThenˆΦ0(F)is a torsion subsheaf ofˆΦ0(L)=E,thus contradicting the hypothesis that E is torsion-free.Thus,we will call S the spectral curve associated to the genericallyfi-brewise semistable torsion-free sheaf E,since it generalizes our previous definition for regular locally-free sheaves.As before,let us denote by L the restriction ofΦ1(E)to its support;in general,it is a coherent sheaf on S. Note that deg(L)=0.Remark.If the restriction of E to some smoothfibre has no sections,then semicontinuity implies that E is genericallyfibrewise semistable,and hence Φ-WIT1.It is somewhat surprising that the spectral data can actually be defined in an interesting,meaningful way under such mild conditions.Summing up,we also conclude:Corollary6.If E is a torsion-free andΦ-WIT1,thenΦ(E)=Φ1(E)is a torsion sheaf of pure dimension one.Conversely,we also have:Lemma7.If L is aˆΦ-WIT0torsion sheaf of pure dimension one on J X with ch(L)=(0,kˆf+rˆσ,0),thenˆΦ0(L)is torsion-free.Proof.Suppose that the support of L decomposes asΣ+F,where F is the sum offibres.When this happens,we have a subsheaf K of L supported on F with degree0.Assume that F is the maximal such effective subdivisor of S.Let Q=L/K.Then Q isˆΦ-WIT0and K isˆΦ-WIT1.The resulting sequence after transforming withˆΦis then0→E→E∗∗→O Z→0(12) where E=ˆΦ0(L),for some zero dimensional subscheme Z.Remark.Notice that applyingΦto the sequence(12)shows that if E is torsion-free but not locally-free,then the support of L containsfibres.The following lemma due to Bridgeland[1]characterizesˆΦ-WIT0sheaves on J X:Lemma8.A sheaf L on J X isˆΦ-WIT0if and only ifHom(L,P x)=0∀x∈X5.1Suitable polarizationsNow letℓbe a polarisation of the elliptic surface X,and letˆℓbe the induced polarisation on J X.Ifℓis arbitrary,it is not difficult to see that aµ-unstable torsion-free sheaf on X can have a p-stable transform.Indeed,let X be an elliptic surface whosefibres are all smooth,and let E be the locally-free sheaf given by the following extension:0→O(−σ+d f)→E→O(σ−d f)→0Clearly,c1(E)=0and c2(E)=−σ2+2d.For d sufficiently large,c2(E)>0 andℓ•(−σ+d f)=−(ℓ•σ)+d(ℓ•f)>0,so E isµ-unstable with respect toℓ.On the other hand,notice that the restriction of E to eachfibre is an extension of Q∈Pic d(T)by its dual.Thus,the bundles obtained by the above extension are generically regular.This meansΦ1(E)is supported on a smooth,irreducible curve,so thatΦ1(E)is necessarily p-stable with respect toˆℓ.Therefore,we can only expect the Fourier-Mukai transform to preserve stability if we restrict the choice of polarisation on X in some convenient way.Definition.Let c be a positive integer,and consider the setΞ(c)={ξ∈Div(X)|−4c≤ξ2<0andξmod2=0}Let Wξbe the intersection of the hyperplaneξ⊥with the ample cone of X.A polarisationℓis said to be c-suitable ifℓ/∈Wξand sign(ℓ•ξ)=sign(f•ξ) for allξ∈Ξ(c).It is easy to see that suitable polarizations exist for every c.The following important result is due to Friedman,Morgan&Witten[7]:Theorem9.Let E be a torsion-free sheaf on X with ch(E)=(r,0,−k). If E isµ-semistable with respect to a k-suitable polarisationℓ,then E is genericallyfibrewise semistable.5.2Preservation of stabilityIt follows from Theorem9that if E is a torsion-free sheaf on X with ch(E)= (r,0,−k),which isµ-semistable with respect to a k-suitable polarisationℓ, then L=Φ1(E)is a torsion sheaf of pure dimension one.Let us now consider its stability(in the sense of Simpson).Proposition10.Let E be a torsion-free sheaf on X which isµ-semistable with respect to a k-suitable polarisationℓ.Then L=Φ1(E)is p-semistable with respect toˆℓ.Proof.Suppose that0−→M−→L−→N−→0is a destabilizing sequence for L.We can assume M is semistable and N has pure dimension1.Now p(M,n)=n+α,for someα>0.Then Hom(M,P x)=0for all x∈X,since p(P x,n)=n.Thus M isˆΦ-WIT0 by Lemma8.Arguing as in Proposition4,we have c1(ˆΦ0(M))=αf hence µ(ˆΦ0(M))>0.SoˆΦ0(M)will destabilize E=ˆΦ0(L)unless r(ˆΦ0(M))= r(E).For this to be the case we must have c1(M)·ˆf=c1(L)·ˆf.Then Nis supported on a sum of fibres.But since N must be ˆΦ-WIT 0we see thatit has non-negative degree on these fibres and so α=0,thus contradicting the assumption on M .If we examine the proof more carefully,we can also see that the as-sumption that E is µ-stable implies that L is p -stable unless its support decomposes as D +D ′,where D is the sum of fibres.When this happens,we have a subsheaf K of L supported on D with degree 0.Assume that D is the maximal such effective subdivisor of S .Let Q =L/K .Then Q is ˆΦ-WIT 0and K is ˆΦ-WIT 1.The resulting sequence after transforming with ˆΦis then E →E ∗∗→O Z for some zero dimensional subscheme Z .Con-versely,applying Φto this sequence shows that the support of L contains fibres.We have therefore established:Proposition 11.Suppose E is µ-semistable with respect to a k -suitable polarisation ℓwith c 1(E )=0.If E is µ-stable and locally-free then L is p -stable.Moreover,E is locally-free if and only if L is destabilized by a sheaf supported on a fibre.We can now consider the opposite question:if we assume L is p -semistable what can we say about its transform?Proposition 12.Suppose L is a p -stable sheaf on J X with Chern character (0,k ˆf +r ˆσ,0),where r,k >1.Then L is ˆΦ-WIT 0and ˆΦ0(L )is locally-free,µ-stable with respect to a k -suitable polarisation and such that ch (ˆΦ0(L ))=(r,0,−k ).Proof.The first statement follows from lemma 8,which requires us to show that Hom(L,P x )=0for all x .However,p (P x ,n )=n and any map L →P x would contradict the stability of L .Now suppose that ˆΦ0(L )is not µ-stable and let A be the destabilizing subsheaf,so that µ(A )≥µ(E )=0Moreover,we may assume that A is µ-stable,thus A is Φ-WIT 1.We may also assume that the quotient B =E/F is µ-stable.Then both A and B are Φ-WIT 1.Since their transforms must have zero rank,A and B both have zero fibre degree and so p (Φ1(A ),n )≥n and this contradicts the stability of L .The fact that ˆΦ0(L )is locally-free follows from the last part of Proposi-tion 11.A similar argument shows that if L is p -semistable and ˆΦ-WIT 0then ˆΦ0(L )must be µ-semistable.Note,however,that it the semistability of L does not imply that L is ˆΦ-WIT 0.Examining the proof above we seethat this happens precisely when L is destabilized by mapping to a sheaf supported on a fibre.In particular,such sheaves are S-equivalent to sheaves which are ˆΦ-WIT 0but whose transforms are not locally-free.We summarize the results of this section in the following theoremTheorem13.Suppose E is a torsion-free sheaf with rank r,c1(E)=0and c2(E)=k on a relatively minimal elliptic surface X over B.Letωbe a k-suitable polarisation on X.Then1.E is aµ-stable locally free sheaf if and only if its transform is a p-stable torsion sheaf supported on a divisor in|kˆf+rˆσ|,whereˆσis a section of J X andˆf is afibre class.2.E isµ-stable properly torsion-free if and only if its transform is a p-semistable torsion sheaf supported on a reducible divisor in|kˆf+rˆσ| which is destabilized only by a sheaf supported on afibre.3.E is properlyµ-semistable if and only if its transform is a p-semistabletorsion sheaf supported on a reducible divisor in|kˆf+rˆσ|.One can also give criteria for the Gieseker stability of E in terms of destabilizing properties of L,but these seem to be less useful.Remark.Recently,Hern´a ndez Ruip´e rez and Mu˜n oz Porras[8]and Yosh-ioka[17]have independently obtained stability results close to those pre-sented in this Section.Remark.It follows from thefirst item in Theorem13that the Fourier-(0,kˆf+ Mukai functorΦinduces a bijective map from M X(r,k)onto S JXrˆσ,0),the Simpson moduli space of p-stable torsion sheaves on J X with the given Chern classes.As we will observe in Section7below,this map is also a hyperk¨a hler isometry when X is an elliptic K3or abelian surface(i.e.when X is hyperk¨a hler).We can see that the geometry of the spectral data is easily linked to the sheaf theoretic properties of the original sheaf.This should make it very easy in practice to use the spectral data to analyse properties of the specimen sheaf.We shall see an example of this in the subsequent Sections where we use the spectral data to explore thefibration structure on the moduli spaces.6The Fibration StructureFrom now on we assume that X is either an elliptic K3surface with a section or a product of elliptic curves.We have seen that there is a natural map Πfrom the moduli space ofµ-stable torsion-free sheaves M TF X(r,k)to the set B of spectral curves.In the case of the K3surface the base B is just the linear system|kˆf+rσ|while for the abelian surface it can be expressed as the total space of the projectivized bundle P F′(O(kf+rˆσ)),where F′is the Mukai transform D(ˆV×W)→D(V׈W)(this can be factored as ΦˆΥ=ˆΨΞfrom(11)).Thefibre ofΠover S is given by suitable subspaces of Jac g(S)−1(S).The hypothesis on X implies that for M to be non-empty we must have both r and k at least2.Theorem14.The map M X(r,k)Π→B is a surjective map of varieties. Proof.We mustfirst show thatΠis a well defined map of varieties.To see this we use the argument given by Friedman and Morgan in[4,Chapter VII,Thm1.14].The key idea is to observe thatΠcoincides(locally in the ´e tale or complex topology)with the projection map from the universal sheaf corresponding to the relative Picard scheme of degree g−1line bundles on families of genus g curves.The universal sheaf only exists locally in these topologies but this is enough to show thatΠis holomorphic.Given a spectral curve S,observe that its structure sheaf O S isˆΦ-WIT0 since in the short exact sequence0→Λ−1→O→O S→0,Λ−1isˆΦ-WIT1 and O isˆΦ-WIT0.We aim to construct a g(S)−1degree line bundle L over the curve S, which is stable as a torsion sheaf on J X.That is,we need to choose g(S)−1 points on S and guarantee that the restriction of L to any proper component S i of S satisfies deg(L|S)>0.We do this by choosing a g(S i)points onieach component S i.This can always be done because if g(S i)>0then S i intersects S−S i at least twice since k and r are both at least2and so g(S−S i)<g(S)−g(S i).Without loss of generality,we assume that the set we have just chosen Z⊂S consists of distinct points away from the singularities of S.Therefore,Ext1(O Z,O S)= z∈Z Ext1(O z,O S)= z∈Z T z SWe pick a class in Ext1(O Z,O S)which is non-zero on each factor.This defines a torsion sheaf L on J X given by this extension class.Then L is stable and locally free on its support by the choice of Z.By Proposition12,L isˆΦ-WIT0and E=ˆΦ0(L)isµ-stable vector bundle such thatΠ(E)=S.From this proof we can also see that any E∈M can be written as an extension0−→ˆΦ0(O S)−→E−→ˆΦ0(O Z)−→0.We shall make use of this in the next Section.Note that such representation is not unique.Such afibration structure will exist for the moduli spaces over any elliptic surface with a section but this surjectivity result may not hold.For the rank 2case see[5,Thm.37]or[4,Thm.1.14].It is also the case that thefibres will not exist in the middle dimension which is the situation we wish to consider in the next Section.7The Hyperk¨a hler StructureWhen X is K3or a torus,it is well known that the moduli spaces of sheaves are hyperk¨a hler.To see this is,note that by a result of Mukai’s[14],for each complex structure I on X we obtain a complex symplectic structure ΩI on M X.Then by Yau’s proof of the Calabi conjecture to obtain the full hyperk¨a hler structure on M X.In fact,the complex symplectic structures arise in a very natural way: T[E]M×T[E]M∼=Ext1(E,E)×Ext1(E,E)∪−→Ext2(E,E)∼=C,where the E is an O X-module with respect to I.But we can express this even more simply by observing that Ext i(E,E)=Hom D(X)(E,E[i]).Then the cup product∪is just composition of maps in the derived category.Since Φis a functor,it must preserve these and so we see thatΦinduces a complex symplectomorphism:(0,kˆf+rˆσ,0)M X(r,k)−→S JXSince this happens for each complex structure,the moduli map induced by Φis actually a hyperk¨a hler isometry.We aim to show now that thefibration structure we have defined in the last Section on M has Lagrangianfibres with respect to this complex symplectic structure.Proposition15.If t1and t2are two tangent vectors to thefibre ofΠthen ΩI(t1,t2)=0.Proof.By continuity it suffices to prove this when t i are defined over a point E given by an extension:0−→ˆΦ(O S)−→E−→ˆΦ(O Z)−→0.where Z consists of discrete points and S is smooth.Deformations of E aris-ing from thefibre ofΠare determined by deformations of Z along S.Then Ext2(E,E)∼=Ext2(L,L)is generated by a non-zero vector in Ext2(O Z,O Z) and thefibreΠS over S has tangent space given by z∈Z λz ,whereλz generates the tangent space to S at z.Butλz∪λz=0in Ext2(O Z,O Z)for each z and soΩI(t1,t2)=0as required.In the case of a product abelian surface we have natural hyperk¨a hler actions of the torus and its dual via translations and twisting byflat line bundles.These act naturally on thefibration structure and the resulting quotient is also a Lagrangianfibration.We can say a little more in the particular case of the product of two elliptic curves.For T=V×W,observe that M T(r,k)has two suchfibration。

相关文档
最新文档