Classical and Quantum Monte Carlo Methods Or: Why we know as little as we do about interacting fermions Erez Berg Student/Postdoc Journal Club, Oct. 2007 Outline • Introduction to MC • Quantum and classical statistical mechanics • Classical Monte Carlo algorithm for the Ising model • Quantum Monte Carlo algorithm for the Hubbard model • “Sign problems” Introduction: Monte Carlo Monte Carlo, Monaco www.wikipedia.org Introduction: Monte Carlo Suppose we are given the problem of calculating S 10000000 i 1 ai And have nothing but a pen and paper. “Monte Carlo” solution: … And we may need to sum much fewer numbers. i= Statistical mechanics Z Tre Thermodynamic quantities F 1 ln Z F F CV T 2 2 , M ,... T H 2 Hˆ Correlation functions 1 ˆ Hˆ ˆ O Tr Oe Z Statistical mechanics Example: Classical Ising model Hˆ { 1 ,..., N } J i j ij Problem: calculate 1 k 1 Hˆ {1 ... L } 1 k e Z 1 ... L 2D lattice with 10x10 sites: Number of terms = 2100=1030 On a supercomputer that does 1015 summations/sec, this takes 107 years… Stochastic summation S f { } { } f { } f Trick: write S P{ } P { } P{ } P P{ } Is an arbitrary probability distribution Pick N configurations { }1...{ }N randomly with probability P{ } Calculate 1 S N i 1... N f { }i P{ }i Stochastic summation (cont.) Mean and standard deviation: (Central limit theorem) f { }i 1 f S N N i 1... N P{ }i P std{S } S S 2 P 1 f std N N P S for any choice of P. So… S N How to choose P? Importance Sampling f We should choose P such that std is minimized. P f For example, if P f , then std 0 ! P … This is a cheat, because to normalize P we need to sum over f. But it shows the correct trend: choose P which is large where f is large. Sampling Technique Back to the Ising model: A natural choice of P: 1 k 1 Hˆ {1 ... L } 1 k e Z 1 ... L P{ } e Hˆ { } Z How to choose random configurations with probability P{ }? Solution: Generate a Markov process that converges to P{ } The Metropolis Algorithm “Random walk” in configuration space: { }1 { }2 { }3 ... 1. Start from a random configuration { }1 2. Pick a spin j. Propose a new configuration { }' that differs by one spin flip 3. If P{ }' P{ }1 , accept the new configuration: { }2 : { }' 4. If P{ }' P{ }1, accept the new configuration with probability P{ }'/ P{ }1 5. And back to step 2… Outline • Introduction to MC • Quantum and classical statistical mechanics • Classical Monte Carlo algorithm for the Ising model • Quantum Monte Carlo algorithm for the Hubbard model • “Sign problems” Quantum statistical mechanics Z Tre Hˆ …But now, H is an operator. In general, we don’t even know how to calculate exp(-H). Example: Single particle Schrodinger equation Hˆ 2 2m 2R V R Quantum statistical mechanics Path integral formulation: Z Tre Hˆ Tre Hˆ e Hˆ ....e Hˆ 2 m ... DR exp d R V R 0 2 Discrete time version: 2 P / m( Ri 1 Ri ) Z dR1...dRP exp V ( R) 2 i 1 P ri ,1 2 1 P ri , 2 The Hubbard Model 1 1 † ˆ H t ci c j H .c. ni U ni , ni , 2 2 ij i i •“Prototype” model for correlated electrons •Relation to real materials: HTC, organic SC,… •No exact (or even approximate) solution for D>1 How to formulate QMC algorithm? Determinantal MC Hˆ Kˆ Vˆ Kˆ t ci† c j H .c. ni ij i 1 1 ˆ V U ni , ni , 2 2 i Z Tre Hˆ Tre Hˆ Hˆ e ....e Hˆ Trotter-Suzuki decomposition: e Hˆ e Kˆ Vˆ e O 2 Blankenbecler, Scalapino, Sugar (1981) Determinantal MC (2) Z Tr e Kˆ Vˆ e e Kˆ Vˆ e ....e Kˆ Vˆ e The K̂ term is quadratic, and can be handled exactly. What to do with the Vˆ term? Hubbard-Stratonovich transformation: e 1 1 U n n 2 2 1 e 2 1 U 4 e s n n s 1 cosh eU / 2 Note that this works only for U>0 Determinantal MC (3) Hubbard-Stratonovich transformation for any U: e 1 1 U n n 2 2 1 e 2 1 U 4 cosh e e 1 1 U n n 2 2 e s n n 1 U<0 s 1 U / 2 1 14U s n n e e 2 s 1 U>0 Determinantal MC (4) For the U>0 case, the partition function becomes: Z e 1 Ld U 4 Here Tr e sik ci† Kij c j e ij ci†Vij 1c j ci† Kij c j ci†Vij N c j ij e ij .... e ij † † † c K c t c c H . c . c i ij j i j i ci ij ij i † c i Vij k c j sik ni, ni, ij k+1 k k-1 i sik i i+1 Determinantal MC (5) Now, since the action is quadratic, the fermions can be traced out. Tr e ci† Aij c j det 1 e Aij ci† Aij k c j Aij k Tr e det 1 e k k Z e 1 Ld U 4 det M s det M s sik M , sik 1 e k ik K V , sik e ik Monte Carlo Evaluation Z e 1 Ld U 4 det M s det M s sik ik ik And, by a variation of Wick’s theorem, Oˆ O s det M s det M s det M s det M s Hˆ ˆ Tr Oe Tr e Hˆ ik ik ik sik sik ik ik How to calculate this sum? Monte Carlo: interpret det M det M Z as a probability P{s} Sign Problem Problem: det M det M is not necessarily positive. Z Solution: det M det M sgn det M det M Probability distribution: Oˆ P s det M det M Z O s sgn sgn P P And evaluate the numerator and denominator by MC! Sign Problem (2) But… At low temperatures and large U, the denominator sgn P becomes extremely small, causing large errors in Ô . 4x4 Hubbard model (Loh et al., 1990) Sign Problem (3) Note that for U<0, det M det M Therefore det M det M 0 And there is no sign problem! Summary • “Sign problem free” models can be considered as essentially solved! • In models with sign problems, in many cases, the low temperature physics is still unclear. • Unfortunately, many interesting models belong to the second type. No sign problem Sign problem Hubbard model (U<0) Hubbard model (U>0): generic filling Hubbard model (U>0): half filling Heisenberg model, triangular lattice “Bose-Hubbard” model (any U) Most “frustrated” spin models Heisenberg model, square lattice Summary Quoting M. Troyer: “If you want you can try your luck: the person who finds a general solution to the sign problem will surely get a Nobel prize!” References •M. Troyer, “Quantum and classical monte carlo algorithms, www.itp.phys.ethz.ch/staff/troyer/publications/troyerP27.pdf • N. Prokofiev, lecture notes on “Worm algorithms for classical and quantum statistical models”, Les Houches summer school on quantum magnetism (2006). •R. R. Dos Santos, Braz. J. Phys. 33, 36 (2003). •R. T. Scalettar, “How to write a determinant QMC code”, http://leopard.physics.ucdavis.edu/rts/p210/howto1.pdf •E. Dagotto, Rev. Mod. Phys. 66, 763 (1994). •J. W. Negele and H. Orland, “Quantum many particle systems”, Addison-Wesley (1988).