Propagators and Green’s Functions • Diffusion equation (B 175) • Fick’s law combined with continuity equation j(r, t ) D (r, t ) (r, t ) .j(r, t ) (r, t ) t (r, t ) D. (r, t ) (r, t ) t L L D 2 t j D Fick’s Law Continuity Equation flux of solute, heat, etc. solute, heat, etc concentration solute, heat, etc source density diffusion constant Propagators and Green’s Functions • • • • • Propagator Ko(x,t,x’,t’) for linear pde in 1-D Evolves solution forward in time from t’ to t (x, t) dx' K o (x, t, x' , t' )a(x' ) Governs how any initial conditions (IC) will evolve Solutions to homogeneous problem (r, t ) 0 for particular IC, a(x) Subject to specific boundary conditions (BC) L 0 ( x, t 0) a( x ) L D 2 t • • • • Ko satisfies LKo(x,t,x’,t’) = 0 Ko(x,t,x’,t’) = δ(x-x’) Ko(x,t,x’,t’) →0 as |x| → ∞ t > t’ t = t’ (equal times) open BC Propagators and Green’s Functions • If propagator satisfies defining relations, solution is generated using (x, t) dx' K o (x, t, x' , t' )a(x' ) Check L 0 and (x, t -) a(x) L dx' LK o (x, t, x' , t' )a(x' ) 0 t t' (x, t t o ) dx' K o (x, t o , x' , t o )a(x' ) dx' (x - x' )a(x' ) a(x) t t' Propagators and Green’s Functions • Propagator for 1-D diffusion equation with open BC 1/ 2 1 K o ( , ) 4 D x x ' t - t' 2 e 4 D • Representations of Dirac delta function in 1-D Gaussian (x) lim 0 1 e x / / Lorentzian ( x ) 0 x2 2 lim 2 2 Fourier c 1 (x) 2 a f(x) (x b)dx f(b) ikx dk e - iff a b c Propagators and Green’s Functions • Check that propagator satisfies the defining relation 1/ 2 1 K o ( , ) 4D x x ' t - t' 2 e 4 D LK o ( , ) D 2 K o ( , ) t 2 4 D e 1 1 3 / 2 K o ( , ) 4 D 5 / 2 4D 2 2 D 2 4 D e 1 1 K ( , ) o 3/ 2 2 4 D 5 / 2 4D 2 2 2 D 2 K o ( , ) LK o ( , ) 0 2 Propagators and Green’s Functions Ko(x,>0.00001) • Consider limit of Ko as tends to zero 1/ 2 1 K o ( , ) 4 D x x ' t - t' 2 e 4 D Ko(x,>0.001) 2 D 2 4D K o ( , 0) lim 2 1 0 2 e K o ( , ) x x ' at equal times D0 Ko(x,>0.1) 1; Plot3D Sqrt 1 PlotRange 4 Pi D0 t 0, 1 Exp xx 4 D0 t , x, 1, 1 , t, 0.1, 1 , Propagators and Green’s Functions • Solution of diffusion equation by separation of variables 2 • Expansion of propagator in eigenfunctions of Sin(kx)e-k2Dt k=10 2 2 D 2 ψ(x, t) D 2 X(x)T(t) 0 x x t t X' ' (x) k 2 X T' (t) k 2 DT X(x) e ikx T(t) e ψ(x, t) X(x)T(t) e k 2Dt 2 ikx k Dt e K o (x, t; x' , t' ) φ p (x)φ (x' )e * p Sin(kx)e-k2Dt k=15 k 2p D t -t' p φ p (x) is an eigenfunct ion of 2 k p is wavenumbe r allowed by the BC Propagators and Green’s Functions For open BC there is no restrictio n on k p K o (x, t; x' , t' ) dk e ik(x -x' ) k 2p D t -t' e For equal times t t' K o (x, t; x' , t' t) dk eik(x -x' ) (x x' ) Satisfacti on of defining relation 2 ik(x -x' ) k 2p D t -t' e LK o (x, t; x' , t' ) 2 dk e t x k 2 D D(ik) 2 K o (x, t; x' , t' ) 0 Propagators and Green’s Functions • • • • • Green’s function Go(r,t,r’,t’) for linear pde in 3-D (B 188) Evolves solution forward in time from t’ to t in presence of sources Solutions to inhomogeneous problem for particular IC a(r) Subject to specific boundary conditions (BC) Heat is added or removed after initial time ( ≠ 0) L (r, t 0) a(r ) L D 2 t • Go satisfies LG o 2 G o (r, t, r ' , t' ) δ(r r ' ) δ(t t' ) t • Go(x,t,x’,t’) = 0 t < t’ • Go(x,t,x’,t’) = δ(x-x’) t = t’ (equal times) • Go(x,t,x’,t’) →0 as |x| → ∞ open BC Propagators and Green’s Functions • Translational invariance of space and time G o G o R, R r - r' t - t' • Defining relation 2 D R G o R, R t G o R, 0 0 causality G o 0 R • Solution in terms of propagator (x) 1 G o R, K o R, 1 0 Heaviside step function 0 0 d d 0 (-x) 1 0 Propagators and Green’s Functions • Check that defining relation is satisfied 2 K o R, D 2R K o R, D R K o R, t t K o R, LK o R, K o R, 0 0 R • Exercise: Show that the solution at time t is t ψ(r, t) dt' dr 'G o r, t, r ' , t' ρr ' , t' dr 'G o r, t, r ' , t o ψr ' , t o to Green’s Function for Schrödinger Equation • Time-dependent single-particle Schrödinger Equation i Ĥ ( r ) Ψ(r, t) 0 t 2 Ĥ(r ) V̂(r ) 2 • Solution by separation of variables Ψ(r, t) c n n (r )e n Ĥ(r ) n (r ) n n (r ) n n 1 - i n t Green’s Function for Schrödinger Equation • Defining relation for Green’s function i Ĥ(r ) G o (r, r ' , t, t' ) r - r ' t - t' t • Eigenfunction expansion of Go iG o r, r' , t t' n (r) n* (r' )e - i n t - t' t - t' n • Exercise: Verify that Go satisfies the defining relation LGo= Green’s Function for Schrödinger Equation • Single-particle Green’s function G o r, r' , t t' θt t' K o r, r' , t t' Add particle Remove particle t’ Remove particle t > t’ time Add particle t t’ > t time Green’s Function for Schrödinger Equation • Eigenfunction expansion of Go for an added particle (M 40) G o r, r ' , t t' i * ( r ) n n (r' )e - i n t - t' t - t' nunocc G o r, r ' , d t - t' G o r, r ' , t t' e i t - t' i n (r ) n* (r ' ) d e nunocc - i n - -1 - i n * i n (r ) n (r ' ) e i n nunocc e- i n 1 * n (r ) n (r ' ) n nunocc n 0 Green’s Function for Schrödinger Equation • Eigenfunction expansion of Go for an added particle G o r, r ' , i n (r ) n* (r ' ) d e nunocc - i n i - i.e. n n i positive infinitesi mal t - t' 0 for t - t' finite t - t' for t - t' i n (r ) n* (r ' ) nunocc -1 - i n i e i n i n (r ) n* (r ' ) G r, r ' , Retarded Green' s Function nunocc n i o poles at n i i.e. below real axis in complex plane 0 Green’s Function for Schrödinger Equation • Eigenfunction expansion of Go for an added hole - i t - t' G - r, r ' , t t' i (r ) * (r ' )e n t'-t o nocc n n G -o r, r ' , i n (r ) n* (r ' ) d e nocc - i n i - 0 i n (r ) n* (r ' ) d e nocc - i n i - -1 i n (r ) n* (r ' ) 0 nocc i n i * ( r ) n (r ' ) G -o r, r ' , n Advanced Green' s Function nocc n i Poles at n i i.e. above real axis in complex plane Green’s Function for Schrödinger Equation • Poles of Go in the complex energy plane Im() F Advanced (holes) x x xx xxx x x xxx xxx xx xx xxx x x xxx Retarded (particles) Re() Contour Integrals in the Complex Plane • Exercise: Fourier back-transform the retarded Green’s function d i G (r, r ' , ) e G o (r, r ' , ) 2 o d e i G (r, r ' , ) ψ n (r ) ψ(r ' ) 2 - n i nunocc o * n Fill in the missing steps to obtain i n i G o (r, r ' , ) i ψ n (r )ψ*n (r ' )e ( ) nunocc Green’s Function for Schrödinger Equation • Spatial Fourier transform of Go for translationally invariant system 1 ik.r ψ k (r ) e ' box normalisat ion' on V V 2 L ik .( r -r ') 1 e (k - k F ) G o (r, r ' , ) V k k i D k d Dk L3 V 1 d 3k V k 2 3 ik .( r -r ') 3 d k e (k - k F ) -ik '.(r-r ') G o (k ' , ) d( r r ' ) e 3 k i - 2 d(r r ' ) i (k-k ').( r-r ') (k - k F ) e 3 k i 2 d 3k - (k - k F ) d 3k (k k ' ) k i - (k k F ) (k F - k ) G o (k , ) G o (k , ) k i k i d(r r ' ) i (k-k ').( r -r ') (k k ' ) 2 3 e Functions of a Complex Variable • Cauchy-Riemann Conditions for differentiability (A 399) z x iy f(z) u(x, y) iv(x, y) u iv u iv u iv lim f(z δz) f(z) f ' (z) δz0 δz δx iy δx iy iy δx limit must be independen t of direction in which δz δx iy approaches zero u v v u and x y x y y x Complex plane f(z) = u(x,y)+iv(x,y) z = x + iy = reif are the Cauchy - Riemann conditions If f is differenti able in a small region about z o , f(z) is analytic at zo If f is differenti able in the whole complex plane, f(z) is an entire function Functions of a Complex Variable • Non-analytic behaviour 20 10 I1 dx f(x) x - x o 1.0 0.5 0.5 1.0 1.5 2.0 I2 dx f(x) x - x o i 10 20 Plot Sin 4 x x 1 , x, 1, 2 , PlotRange A pole in a function renders the function non-analytic at that point 20, 20 Functions of a Complex Variable • Cauchy Integral Theorem (A 404) f(z) analytic (and single - valued) in some simply connected region, R For every closed path C in R f(z)dz 0 y C C x Proved using Stokes' theorem Functions of a Complex Variable • Cauchy Integral Formula (A 411) C is a closed contour in the complex plane f(z) analytic on the contour f(z)dz C z - zo is well defined since z is on C and zo is not f(z) is not analytic at z z o unless f(z o ) 0 z - zo y 1 f(z)dz f(z o ) 2i C z - z o Direction of integratio n is anti - clockwise Changing direction of integratio n changes sign of integral zo C x Functions of a Complex Variable • Cauchy Integral Formula (A 411) From Cauchy integral theorem f(z)dz f(z)dz C z - z o C z - z o 0 2 y zo C2 z - z o r ei dz ir ei d lim f(z r ei )ir ei d f(z)dz r0 r0 o 2if(z o ) i z - zo re C2 C2 lim f(z)dz C z - z o 2if(z o ) 1 f(z)dz f(z o ) 2i C z - z o If z o lies outside C then f(z) f(z)dz g(z) is analytic and 0 z - zo z - zo C C x Functions of a Complex Variable • Taylor Series (A 416) • When a function is analytic on and within C containing a point zo it may be expanded about zo in a Taylor series of the form f(z) a n (z - z o ) n 0 n f n (z o ) 1 f(z' )dz' an n! 2i C (z'-zo ) n 1 y zo C x f(z) a o a1 (z - z o ) a 2 (z - z o ) 2 ... • Expansion applies for |z-zo| < |z-z1| where z1 is nearest non-analytic point • See exercises for proof of expansion coefficients Functions of a Complex Variable • Laurent Series (A 416) • When a function is analytic in an annular region about a point zo it may be expanded in a Laurent series of the form y zo f(z) a n (z - z o ) n n an 1 f(z' )dz' 2i C (z'-zo ) n 1 f(z) a o a1 (z - z o ) a 2 (z - z o ) 2 ... C x a -1 a -2 ... 2 z - z o z - z o • If an = 0 for n < -m < 0 and a-m = 0, f(z) has a pole of order m at zo • If m = 1 then it is a simple pole • Analytic functions whose only singularities are separate poles are termed meromorphic functions Contour Integrals in the Complex Plane • Cauchy Residue Theorem (A 444) f(z)dz a (z - z n C n m th order pole n o ) dz C n 1 z1 (z - z o ) a n (z - z o ) dz a n n 1 C n 0 n 1 z1 Let z - z o r ei 2 i ir e d 2i a -1 n 1 i re 0 a -1 (z - z o ) -1 dz a -1 C a -1 is the residue of f(z) at z o f(z)dz 2i residues C 1 d m-1 m a 1 z z f(z) o m -1 (m 1)! dz simple pole (m 1) a 1 z z o f(z) z z o See tutorial sheet for proof z zo Contour Integrals in the Complex Plane • Cauchy Residue Theorem (A 444) Example : f(z) has two simple poles f(z) 1 z a z b A B z a z b by partial fractions lim z a z a A z a B z a f(z) lim z a z a z b 1 z b 1 A a 1 a b lim z a Contour Integrals in the Complex Plane • Integration along real axis in complex plane f(z)dz lim R R f(x)dx lim -R R i i f(Re ) i Re d y 0 2i residues in UHP (upper half plane) enclosed pole -R • Provided: • f(z) is analytic in the UHP • f(z) vanishes faster than 1/z f(x)dx 2i residues in UHP (upper half plane) - • Can use LHP (lower half plane) if f(z) vanishes faster than 1/z and f(z) is analytic there • Usually can do one or the other, same result if possible either way x +R Contour Integrals in the Complex Plane • Integration along real axis in complex plane • Theta function (M40) d e i ( ) 1 ( 0) 2 i i y 0 ( 0) i i e e pole at i residue 2i 2i close contour in lower 1/2 plane for 0 t >0 x i C d e i d e i d e i C 2i i 2i i LHP 2i i d e i e 0 2i e ( 0) 2i i 2i close contour in upper 1/2 plane for 0 0 ( 0) Contour Integrals in the Complex Plane • Integration along real axis in complex plane • Principal value integrals – first order pole on real axis • What if the pole lies on the integration contour? x o δ f(z)dz f(x)dx f(z)dz - C1 x o δ y C2 C1 -R f(x)dx f(z)dz 2i enclosed residues C2 Let z - x o re i dz irei d dz ire i d C z - x o rei i clockwise, similarly i anticlockw ise 1 0 • If small semi-circle C1 in/excludes pole contribution appears twice/once lim 0 x o δ - x o δ - f(x)dx f(x)dx P f(x)dx x +R Contour Integrals in the Complex Plane y • Kramers-Kronig Relations (A 469) 1 f(z)dz f(z o ) 2i C z - z o 1 2i 1 2i zo Cauchy integral formula -R→- f(x)dx - x - z o f(z o ) - f(x)dx _ 0 pole at z o x o iy o in side contour _ pole at z o x o iy o outside contour x - zo 1 1 1 f(z o ) f(x)dx _ 2i - x - zo x - zo 1 f(z o ) i 1 x - zo 1 _ x - zo 2x - x o x - x o 2 y o2 f(x) x - x o dx 1 f(x)dx (y 0) P (y o 0) - x - x o 2 yo2 o i - x - x o žo x +R →+ Contour Integrals in the Complex Plane • Kramers-Kronig Relations 1 x-x o f(x)dx f(z o ) f(x, y) u(x, y) iv(x, y) 2 2 i - x-x o y o 1 f(z o ) u(x o , y o ) iv(x o , y o ) i x-x o v(x,0)dx - x-x o 2 y o2 1 x-x o u(x,0)dx v(x o , y o ) - x-x o 2 y o2 u(x o , y o ) 1 1 v(x o ,0) Equate real parts (y o 0) Equate imaginary parts (y o 0) v(x,0)dx u(x o ,0) P - x-x o 1 x-x o (u(x,0) iv(x,0))dx - x-x o 2 y o2 Equate real parts (y o 0) u(x,0)dx x-x o - P Equate imaginary parts (y o 0)