Guiding Monte Carlo Simulations with Machine Learning
|
|
- William Harris
- 5 years ago
- Views:
Transcription
1 Guiding Monte Carlo Simulations with Machine Learning Yang Qi Department of Physics, Massachusetts Institute of Technology Joining Fudan University in 2017 KITS, June / 33
2 References Works led by Liang Fu, Zi-Yang Meng, and Lei Wang. Jun-Wei Liu, Xiao-Yan Xu, Hui-Tao Shen, Yuri Nagai, Li Huang, Yi-Feng Yang etc. References: 1. J Liu, YQ, Z-Y Meng and L Fu, PRB 95, (R) (2017). 2. J Liu, H Shen, YQ, Z-Y Meng and L Fu, PRB 95, (R) (2017). 3. X-Y Xu, YQ, J Liu, L Fu and Z-Y Meng, arxiv: [cond-mat.str-el]. 4. Y Nagai, H Shen, YQ and L Fu, arxiv: [cond-mat.str-el]. 5. Li Huang, Lei Wang, PRB 95, (R) (2017). 6. Li Huang, Yi-Feng Yang, Lei Wang, PRE 95, (R)(2017). 7. Lei Wang, arxiv: [cond-mat.str-el]. 2/ 33
3 Outline 1 Introduction to Markov Chain Monte Carlo 2 The Importance of Update 3 Example: Ising Model and Cluster Update 4 Example: Bosonic Model and Cumulative Update 3/ 33
4 Outline 1 Introduction to Markov Chain Monte Carlo 2 The Importance of Update 3 Example: Ising Model and Cluster Update 4 Example: Bosonic Model and Cumulative Update 4/ 33
5 Monte Carlo simulation: an unbiased method A widely used numerical method in statistical physics and quantum many-body physics. Unbiased: reliable statistical error bar. Fast: error decreases as 1/ N. Universal: applies to any model without the sign program. 5/ 33
6 Introduction to MCMC Consider a statistical mechanics model: Z = C e βh[c] = C W (C). The Markov-chain Monte Carlo (MCMC) is a way to do importance sampling. A Markov chain is constructed, C i 1 C i C i+1 Markov chain: p(c i C j ) only depends on C i (no memory). Goal: distribution of C converges to the Boltzmann distribution W (C). Any observable can be measured from a Markov chain, O(C)W (C) O = 1 O(C i ). W (C) N i 6/ 33
7 Detailed balance Detailed balance: p(c D) p(d C) = W (D) W (C). Detailed balance (and ergodicity) guarantees that IF the MC converges, it converges to the desired distribution W (C). Metropolis-Hastings algorithm: propose accept/reject. p(c D) = q(c D)α(C D). { α(c D) = min 1, W (D) } q(d C). W (C) q(c D) 7/ 33
8 Autocorrelation time Autocorrelation time measures the efficiency of the update algorithm. time sequence: O(t 1) O(t) O(t + 1), O(t) = O[C(t)]. Autocorrelation function A O ( t) = O(t)O(t + t) O(t) 2 e t/τ. Complexity τ O = 1 O(C i ). N Statistical error δo 1 N i only if O(i) are M(t)M(s + t) M Local Self-learning Naive-Wolff independent t 8/ 33
9 Outline 1 Introduction to Markov Chain Monte Carlo 2 The Importance of Update 3 Example: Ising Model and Cluster Update 4 Example: Bosonic Model and Cumulative Update 9/ 33
10 Metropolis algorithm Local update: randomly select a site and flip the spin. q(c D) = q(d C) = 1 N { }. α(c D) = min. 1, W (D) W (C) N trials are counted as one MC step. Very general: applies to any model. N Metropolis, A W Rosenbluth, M N Rosenbluth, A H Teller, and E Teller, J Chem Phys 21, 1087 (1953). 10/ 33
11 Metropolis algorithm Local update: randomly select a site and flip the spin. q(c D) = q(d C) = 1 N { }. α(c D) = min. 1, W (D) W (C) N trials are counted as one MC step. Very general: applies to any model. N Metropolis, A W Rosenbluth, M N Rosenbluth, A H Teller, and E Teller, J Chem Phys 21, 1087 (1953). 10/ 33
12 Critical slowing down The real dynamical relaxation time diverges at the critical point: a critical system is slow to equilibrate. The local update mimics the real relaxation process: also exhibits the critical slowing down phenomena. τ L z, z = for the 2D Ising model τ L z There s a way around this: MCMC simulation does not have to mimic the real dynamics... 11/ 33
13 Wolff algorithm: a cluster update A cluster is built one bond at a time. Probability of activating a bond is cleverly designed, such that q(c D) q(d C) = W (D) W (C). Thus we have the ideal acceptance ratio α = 1. Very efficient update: τ L Only works with two-body interactions: H = ij J ijs i s j. U Wolff, PRL. 62, 361 (1989). 12/ 33
14 Wolff algorithm: a cluster update A cluster is built one bond at a time. Probability of activating a bond is cleverly designed, such that q(c D) q(d C) = W (D) W (C). Thus we have the ideal acceptance ratio α = 1. Very efficient update: τ L Only works with two-body interactions: H = ij J ijs i s j. U Wolff, PRL. 62, 361 (1989). 12/ 33
15 Wolff algorithm: a cluster update A cluster is built one bond at a time. Probability of activating a bond is cleverly designed, such that q(c D) q(d C) = W (D) W (C). Thus we have the ideal acceptance ratio α = 1. Very efficient update: τ L Only works with two-body interactions: H = ij J ijs i s j. U Wolff, PRL. 62, 361 (1989). 12/ 33
16 Wolff algorithm: a cluster update A cluster is built one bond at a time. Probability of activating a bond is cleverly designed, such that q(c D) q(d C) = W (D) W (C). Thus we have the ideal acceptance ratio α = 1. Very efficient update: τ L Only works with two-body interactions: H = ij J ijs i s j. U Wolff, PRL. 62, 361 (1989). 12/ 33
17 Wolff algorithm: a cluster update A cluster is built one bond at a time. Probability of activating a bond is cleverly designed, such that q(c D) q(d C) = W (D) W (C). Thus we have the ideal acceptance ratio α = 1. Very efficient update: τ L Only works with two-body interactions: H = ij J ijs i s j. U Wolff, PRL. 62, 361 (1989). 12/ 33
18 Wolff algorithm: a cluster update A cluster is built one bond at a time. Probability of activating a bond is cleverly designed, such that q(c D) q(d C) = W (D) W (C). Thus we have the ideal acceptance ratio α = 1. Very efficient update: τ L Only works with two-body interactions: H = ij J ijs i s j. U Wolff, PRL. 62, 361 (1989). 12/ 33
19 Wolff algorithm: a cluster update A cluster is built one bond at a time. Probability of activating a bond is cleverly designed, such that q(c D) q(d C) = W (D) W (C). Thus we have the ideal acceptance ratio α = 1. Very efficient update: τ L Only works with two-body interactions: H = ij J ijs i s j. U Wolff, PRL. 62, 361 (1989). 12/ 33
20 Wolff algorithm: a cluster update A cluster is built one bond at a time. Probability of activating a bond is cleverly designed, such that q(c D) q(d C) = W (D) W (C). Thus we have the ideal acceptance ratio α = 1. Very efficient update: τ L Only works with two-body interactions: H = ij J ijs i s j. U Wolff, PRL. 62, 361 (1989). 12/ 33
21 Challenge Local update is too slow for many models: critical slowing down, glassy behavior, separation of energy scales, etc. Global update is only available for certain models. Like Wolff algorithm for two-body interactions. A good update algorithm for generic models? 13/ 33
22 A good update samples low-energy states A dilemma of local updates: Step is too small: high acceptance, small difference. Step is too big: low acceptance, big difference. Global updates: explore the low-energy configurations. 14/ 33
23 A good update samples low-energy states A dilemma of local updates: Step is too small: high acceptance, small difference. Step is too big: low acceptance, big difference. Global updates: explore the low-energy configurations. 14/ 33
24 A good update samples low-energy states A dilemma of local updates: Step is too small: high acceptance, small difference. Step is too big: low acceptance, big difference. Global updates: explore the low-energy configurations. 14/ 33
25 Modeling the update process A stochastic function f : C D. Parameterized: f = f (p 1, p 2,..., p n ). Analytically known q(c f D). Optimize p i, such that Choose the acceptance ratio α(c D) = min The approximation q(c D) f W (D) q(d C) f W (C) long as we know q(c f D) accurately. q(c f D) q(d f C) W (D) W (C). { } 1, W (D) q(d f C) 1. W (C) q(c f D) does not introduce error in the MC simulation, as 15/ 33
26 Design (i) Trial simulation by local update (ii) Learning Machine Learning Ηeff (iii) (iv) Simulating Ηeff Η Propose trial Conf. Detailed balance 16/ 33
27 Outline 1 Introduction to Markov Chain Monte Carlo 2 The Importance of Update 3 Example: Ising Model and Cluster Update 4 Example: Bosonic Model and Cumulative Update 17/ 33
28 Example: The model: The effective model: H = J 1 s i s j J 2 ij H eff = J eff 1 ijkl s i s j. ij s i s j s k s l. We consider J 1 = 1, J 2 = 0.2. An Ising transition at T c = Ising model without J 2 has T c = / 33
29 The self-learning update The update: cluster is constructed using the effective model. q(c D) q(d C) = W eff(d) W eff (C). The acceptance ratio: { α(c D) = min 1, W (D) W (C) W eff (C) W eff (D) }. The algorithm still satisfies the detailed balance exactly, although the effective model is approximate. 19/ 33
30 Learning the effective model 1. Generate a sample using the local update. 2. Perform a linear regression, H eff = J eff 1 S i S j + E 0. ij 20/ 33
31 Unsupervised machine learning The linear regression we just did can be viewed as an unsupervised machine learning. Unsupervised ML: learning the underlying distribution from a sample of configurations. We pretend that we didn t know H, and learn a simpler model H eff. The learned model describes the low-energy effective theory, where the high-energy fluctuations are treated as noise in the sample. Can incorporate more advanced ML models and algorithms. 21/ 33
32 Learning the effective model 1. Generate a sample using the local update, at T = 5 > T c. 2. Perform a linear regression, H eff = J eff 1 S i S j + E J eff 1 = Generate another sample using the self-learning update, at T = T c. 5. J eff 1 = ij 22/ 33
33 Autocorrelation time M(t)M(s + t) M Local Self-learning Naive-Wolff t System size / 33
34 SLMC works well at moderate sizes 2000 Local Self-Learned τ L The results look good at moderate sizes L = 20 80, but the scaling behavior is exponential... Eventually, the self-learning update becomes slower than the local update. 24/ 33
35 Limiting the cluster size H = H H eff scales with the perimeter of the cluster L. τ e L/L0 because α e β H e L/L0. A simple effective model is not accurate: need more accurate model, with a hierarchy of energy scales. Quick fix: limit the cluster size during cluster growing. Still satisfies the detailed balance, if α is corrected correspondingly. A block update: same scaling as the local update, but with a smaller prefactor. 25/ 33
36 Restricted-cluster-size update τ Local Self-Learned Restricted τ L L τ R L L We restrict the cluster size to r = / 33
37 Summary τ Local Self-learning Restricted τ L L τ R L L 27/ 33
38 Outline 1 Introduction to Markov Chain Monte Carlo 2 The Importance of Update 3 Example: Ising Model and Cluster Update 4 Example: Bosonic Model and Cumulative Update 28/ 33
39 Determinant Monte Carlo: computational cost Fermion coupling to (auxiliary) bosonic field. H = ij t ij c i c j + i σ i c i c i + H[σ]. Simulation: use bosonic field as configurations. C = {σ i (t)}, i = 1,... N; 0 < t < β. Weight of each configuration W (C): integrate out the fermions. Computational cost: βn 3. Fast update: βn 3 for βn steps of local updates. Generating two independent configurations: βn 3 τ L. 29/ 33
40 Cumulative update Idea: replace W (C) by W (C), which is much faster to evaluate. Local updates guided by W (C): C D : C C 1 C i 1 C i C i+1 C M D. Detailed balance: p(c i C i+1 ) p(c i+1 C i ) = W (C i+1) W (C i ) For each path: p(c C 1 ) p(c i C i+1 )p(c M D) p(c 1 C) p(c i+1 C i )p(d C M ) = W (C 1 ) W (C 2 ) W (C) W (C 1 ) W (C M ) W (D) W (C M 1 ) W (C M ) = W (D) W (C). Summing over all paths: q(c D) p(c D) = q(d C) p(d C) = C i p(c C 1 ) p(c i C i+1 )p(c M D) C i p(c 1 C) p(c i+1 C i )p(d C M ) = W (D) W (C). 30/ 33
41 Cumulative update Approximated weight: Update scheme: Acceptance ratio: Choose good W and M τ L : τ CU 1. Computational cost: βn 3 τ βn 3. W (C) W (C). q(c D) q(d C) = W (D) W (C) W (D) W (C). { α(c D) = min 1, W (D) W } (C) W (C) W 1. (D) Again, W (C) W (C) does not introduce approximation. 31/ 33
42 Constructing effective weights 1. Physical intuition: bosonic effective model of W eff [σ] + Linear Regression. Works best for gapped fermions. H = ij t ij c iα c jα + i σ i c iα σz αβc iβ + H[σ] W eff [σ] = ij J ij σ i σ j. 2. Machine Learning models: Restricted Boltzmann Machine, Neural Networks, etc. 32/ 33
43 Self-Learning MC in action Zi-Hong Liu, Xiao-Yan Xu, Yang Qi, Kai Sun and Zi-Yang Meng, to appear. Fermion + boson model, new critical universality. Simulated at the critical point. SLMC: L = superposition of spin up and spin down Ising site Ising spin 3 b coupling L=24, β=8 Fermi pockets Hot spots fermion 2 c a λ=2 k=q ky -t BKT phase Bare Fermi surface ln(χ -1 (q,0)- χ -1 (0,0)) fermion site L=30, β=8 a ln( q )+ln(c ) q q λ=1 SDW Induced shallow band Clock phase kx PM phase a q =1.997 ± ln(c )= x10 q ln( q ) ± / 33
Self-learning Monte Carlo Method
Self-learning Monte Carlo Method Zi Yang Meng ( 孟子杨 ) http://ziyangmeng.iphy.ac.cn/ Know thyself "Know thyself" (Greek: γνῶθι σεαυτόν, gnothi seauton) one of the Delphic maxims and was inscribed in the
More informationRevealing fermionic quantum criticality from new Monte Carlo techniques. Zi Yang Meng ( 孟子杨 )
Revealing fermionic quantum criticality from new Monte Carlo techniques Zi Yang Meng ( 孟子杨 ) http://ziyangmeng.iphy.ac.cn Collaborators and References Xiao Yan Xu Zi Hong Liu Chuang Chen Gao Pei Pan Yang
More informationQuantum Monte Carlo investigations of correlated electron systems, present and future. Zi Yang Meng ( 孟子杨 )
Quantum Monte Carlo investigations of correlated electron systems, present and future Zi Yang Meng ( 孟子杨 ) http://ziyangmeng.iphy.ac.cn Collaborators Xiao Yan Xu Yoni Schattner Zi Hong Liu Erez Berg Chuang
More informationMonte Caro simulations
Monte Caro simulations Monte Carlo methods - based on random numbers Stanislav Ulam s terminology - his uncle frequented the Casino in Monte Carlo Random (pseudo random) number generator on the computer
More informationA Video from Google DeepMind.
A Video from Google DeepMind http://www.nature.com/nature/journal/v518/n7540/fig_tab/nature14236_sv2.html Can machine learning teach us cluster updates? Lei Wang Institute of Physics, CAS https://wangleiphy.github.io
More informationClassical Monte Carlo Simulations
Classical Monte Carlo Simulations Hyejin Ju April 17, 2012 1 Introduction Why do we need numerics? One of the main goals of condensed matter is to compute expectation values O = 1 Z Tr{O e βĥ} (1) and
More informationAccelerated Monte Carlo simulations with restricted Boltzmann machines
Accelerated simulations with restricted Boltzmann machines Li Huang 1 Lei Wang 2 1 China Academy of Engineering Physics 2 Institute of Physics, Chinese Academy of Sciences July 6, 2017 Machine Learning
More information4. Cluster update algorithms
4. Cluster update algorithms Cluster update algorithms are the most succesful global update methods in use. These methods update the variables globally, in one step, whereas the standard local methods
More informationMarkov Processes. Stochastic process. Markov process
Markov Processes Stochastic process movement through a series of well-defined states in a way that involves some element of randomness for our purposes, states are microstates in the governing ensemble
More informationA Monte Carlo Implementation of the Ising Model in Python
A Monte Carlo Implementation of the Ising Model in Python Alexey Khorev alexey.s.khorev@gmail.com 2017.08.29 Contents 1 Theory 1 1.1 Introduction...................................... 1 1.2 Model.........................................
More informationCritical Dynamics of Two-Replica Cluster Algorithms
University of Massachusetts Amherst From the SelectedWorks of Jonathan Machta 2001 Critical Dynamics of Two-Replica Cluster Algorithms X. N. Li Jonathan Machta, University of Massachusetts Amherst Available
More informationRandom Walks A&T and F&S 3.1.2
Random Walks A&T 110-123 and F&S 3.1.2 As we explained last time, it is very difficult to sample directly a general probability distribution. - If we sample from another distribution, the overlap will
More informationMonte Carlo Study of Planar Rotator Model with Weak Dzyaloshinsky Moriya Interaction
Commun. Theor. Phys. (Beijing, China) 46 (2006) pp. 663 667 c International Academic Publishers Vol. 46, No. 4, October 15, 2006 Monte Carlo Study of Planar Rotator Model with Weak Dzyaloshinsky Moriya
More informationMachine Learning with Quantum-Inspired Tensor Networks
Machine Learning with Quantum-Inspired Tensor Networks E.M. Stoudenmire and David J. Schwab Advances in Neural Information Processing 29 arxiv:1605.05775 RIKEN AICS - Mar 2017 Collaboration with David
More informationICCP Project 2 - Advanced Monte Carlo Methods Choose one of the three options below
ICCP Project 2 - Advanced Monte Carlo Methods Choose one of the three options below Introduction In statistical physics Monte Carlo methods are considered to have started in the Manhattan project (1940
More informationQuantum and classical annealing in spin glasses and quantum computing. Anders W Sandvik, Boston University
NATIONAL TAIWAN UNIVERSITY, COLLOQUIUM, MARCH 10, 2015 Quantum and classical annealing in spin glasses and quantum computing Anders W Sandvik, Boston University Cheng-Wei Liu (BU) Anatoli Polkovnikov (BU)
More informationNumerical methods for lattice field theory
Numerical methods for lattice field theory Mike Peardon Trinity College Dublin August 9, 2007 Mike Peardon (Trinity College Dublin) Numerical methods for lattice field theory August 9, 2007 1 / 24 Numerical
More informationMonte Carlo and cold gases. Lode Pollet.
Monte Carlo and cold gases Lode Pollet lpollet@physics.harvard.edu 1 Outline Classical Monte Carlo The Monte Carlo trick Markov chains Metropolis algorithm Ising model critical slowing down Quantum Monte
More informationMarkov Chain Monte Carlo The Metropolis-Hastings Algorithm
Markov Chain Monte Carlo The Metropolis-Hastings Algorithm Anthony Trubiano April 11th, 2018 1 Introduction Markov Chain Monte Carlo (MCMC) methods are a class of algorithms for sampling from a probability
More informationOptimized statistical ensembles for slowly equilibrating classical and quantum systems
Optimized statistical ensembles for slowly equilibrating classical and quantum systems IPAM, January 2009 Simon Trebst Microsoft Station Q University of California, Santa Barbara Collaborators: David Huse,
More informationQuantum Annealing in spin glasses and quantum computing Anders W Sandvik, Boston University
PY502, Computational Physics, December 12, 2017 Quantum Annealing in spin glasses and quantum computing Anders W Sandvik, Boston University Advancing Research in Basic Science and Mathematics Example:
More informationMachine learning quantum phases of matter
Machine learning quantum phases of matter Summer School on Emergent Phenomena in Quantum Materials Cornell, May 2017 Simon Trebst University of Cologne Machine learning quantum phases of matter Summer
More informationQuantum simulation with string-bond states: Joining PEPS and Monte Carlo
Quantum simulation with string-bond states: Joining PEPS and Monte Carlo N. Schuch 1, A. Sfondrini 1,2, F. Mezzacapo 1, J. Cerrillo 1,3, M. Wolf 1,4, F. Verstraete 5, I. Cirac 1 1 Max-Planck-Institute
More informationComputer Vision Group Prof. Daniel Cremers. 11. Sampling Methods: Markov Chain Monte Carlo
Group Prof. Daniel Cremers 11. Sampling Methods: Markov Chain Monte Carlo Markov Chain Monte Carlo In high-dimensional spaces, rejection sampling and importance sampling are very inefficient An alternative
More informationParallel Tempering Algorithm in Monte Carlo Simulation
Parallel Tempering Algorithm in Monte Carlo Simulation Tony Cheung (CUHK) Kevin Zhao (CUHK) Mentors: Ying Wai Li (ORNL) Markus Eisenbach (ORNL) Kwai Wong (UTK/ORNL) Metropolis Algorithm on Ising Model
More informationThe underdoped cuprates as fractionalized Fermi liquids (FL*)
The underdoped cuprates as fractionalized Fermi liquids (FL*) R. K. Kaul, A. Kolezhuk, M. Levin, S. Sachdev, and T. Senthil, Physical Review B 75, 235122 (2007) R. K. Kaul, Y. B. Kim, S. Sachdev, and T.
More informationNematic and Magnetic orders in Fe-based Superconductors
Nematic and Magnetic orders in Fe-based Superconductors Cenke Xu Harvard University Collaborators: Markus Mueller, Yang Qi Subir Sachdev, Jiangping Hu Collaborators: Subir Sachdev Markus Mueller Yang Qi
More informationWang-Landau sampling for Quantum Monte Carlo. Stefan Wessel Institut für Theoretische Physik III Universität Stuttgart
Wang-Landau sampling for Quantum Monte Carlo Stefan Wessel Institut für Theoretische Physik III Universität Stuttgart Overview Classical Monte Carlo First order phase transitions Classical Wang-Landau
More informationPattern Recognition and Machine Learning. Bishop Chapter 11: Sampling Methods
Pattern Recognition and Machine Learning Chapter 11: Sampling Methods Elise Arnaud Jakob Verbeek May 22, 2008 Outline of the chapter 11.1 Basic Sampling Algorithms 11.2 Markov Chain Monte Carlo 11.3 Gibbs
More informationA Review of Pseudo-Marginal Markov Chain Monte Carlo
A Review of Pseudo-Marginal Markov Chain Monte Carlo Discussed by: Yizhe Zhang October 21, 2016 Outline 1 Overview 2 Paper review 3 experiment 4 conclusion Motivation & overview Notation: θ denotes the
More informationBrief introduction to Markov Chain Monte Carlo
Brief introduction to Department of Probability and Mathematical Statistics seminar Stochastic modeling in economics and finance November 7, 2011 Brief introduction to Content 1 and motivation Classical
More information3.320 Lecture 18 (4/12/05)
3.320 Lecture 18 (4/12/05) Monte Carlo Simulation II and free energies Figure by MIT OCW. General Statistical Mechanics References D. Chandler, Introduction to Modern Statistical Mechanics D.A. McQuarrie,
More informationGeneral Construction of Irreversible Kernel in Markov Chain Monte Carlo
General Construction of Irreversible Kernel in Markov Chain Monte Carlo Metropolis heat bath Suwa Todo Department of Applied Physics, The University of Tokyo Department of Physics, Boston University (from
More informationQuantum Monte Carlo Simulations in the Valence Bond Basis. Anders Sandvik, Boston University
Quantum Monte Carlo Simulations in the Valence Bond Basis Anders Sandvik, Boston University Outline The valence bond basis for S=1/2 spins Projector QMC in the valence bond basis Heisenberg model with
More informationGraphical Representations and Cluster Algorithms
Graphical Representations and Cluster Algorithms Jon Machta University of Massachusetts Amherst Newton Institute, March 27, 2008 Outline Introduction to graphical representations and cluster algorithms
More informationTheir Statistical Analvsis. With Web-Based Fortran Code. Berg
Markov Chain Monter rlo Simulations and Their Statistical Analvsis With Web-Based Fortran Code Bernd A. Berg Florida State Univeisitfi USA World Scientific NEW JERSEY + LONDON " SINGAPORE " BEIJING " SHANGHAI
More informationGlobal phase diagrams of two-dimensional quantum antiferromagnets. Subir Sachdev Harvard University
Global phase diagrams of two-dimensional quantum antiferromagnets Cenke Xu Yang Qi Subir Sachdev Harvard University Outline 1. Review of experiments Phases of the S=1/2 antiferromagnet on the anisotropic
More informationMarkov chain Monte Carlo Lecture 9
Markov chain Monte Carlo Lecture 9 David Sontag New York University Slides adapted from Eric Xing and Qirong Ho (CMU) Limitations of Monte Carlo Direct (unconditional) sampling Hard to get rare events
More informationNumerical Analysis of 2-D Ising Model. Ishita Agarwal Masters in Physics (University of Bonn) 17 th March 2011
Numerical Analysis of 2-D Ising Model By Ishita Agarwal Masters in Physics (University of Bonn) 17 th March 2011 Contents Abstract Acknowledgment Introduction Computational techniques Numerical Analysis
More informationStatistics and Quantum Computing
Statistics and Quantum Computing Yazhen Wang Department of Statistics University of Wisconsin-Madison http://www.stat.wisc.edu/ yzwang Workshop on Quantum Computing and Its Application George Washington
More informationIntroduction to Machine Learning
Introduction to Machine Learning Brown University CSCI 1950-F, Spring 2012 Prof. Erik Sudderth Lecture 25: Markov Chain Monte Carlo (MCMC) Course Review and Advanced Topics Many figures courtesy Kevin
More informationComputer Vision Group Prof. Daniel Cremers. 10a. Markov Chain Monte Carlo
Group Prof. Daniel Cremers 10a. Markov Chain Monte Carlo Markov Chain Monte Carlo In high-dimensional spaces, rejection sampling and importance sampling are very inefficient An alternative is Markov Chain
More informationSampling Methods (11/30/04)
CS281A/Stat241A: Statistical Learning Theory Sampling Methods (11/30/04) Lecturer: Michael I. Jordan Scribe: Jaspal S. Sandhu 1 Gibbs Sampling Figure 1: Undirected and directed graphs, respectively, with
More informationBayesian Methods for Machine Learning
Bayesian Methods for Machine Learning CS 584: Big Data Analytics Material adapted from Radford Neal s tutorial (http://ftp.cs.utoronto.ca/pub/radford/bayes-tut.pdf), Zoubin Ghahramni (http://hunch.net/~coms-4771/zoubin_ghahramani_bayesian_learning.pdf),
More informationCopyright 2001 University of Cambridge. Not to be quoted or copied without permission.
Course MP3 Lecture 4 13/11/2006 Monte Carlo method I An introduction to the use of the Monte Carlo method in materials modelling Dr James Elliott 4.1 Why Monte Carlo? The name derives from the association
More informationSession 3A: Markov chain Monte Carlo (MCMC)
Session 3A: Markov chain Monte Carlo (MCMC) John Geweke Bayesian Econometrics and its Applications August 15, 2012 ohn Geweke Bayesian Econometrics and its Session Applications 3A: Markov () chain Monte
More informationDiagrammatic Monte Carlo
Sign Problems and Complex Actions, ECT*, Trento, March 2-6, 2009 Diagrammatic Monte Carlo Boris Svistunov University of Massachusetts, Amherst Nikolay Prokof ev Kris Van Houcke (Umass/Ghent) Evgeny Kozik
More informationCluster Algorithms to Reduce Critical Slowing Down
Cluster Algorithms to Reduce Critical Slowing Down Monte Carlo simulations close to a phase transition are affected by critical slowing down. In the 2-D Ising system, the correlation length ξ becomes very
More informationMonte Carlo Methods in High Energy Physics I
Helmholtz International Workshop -- CALC 2009, July 10--20, Dubna Monte Carlo Methods in High Energy Physics CALC2009 - July 20 10, Dubna 2 Contents 3 Introduction Simple definition: A Monte Carlo technique
More informationMarkov chain Monte Carlo methods for visual tracking
Markov chain Monte Carlo methods for visual tracking Ray Luo rluo@cory.eecs.berkeley.edu Department of Electrical Engineering and Computer Sciences University of California, Berkeley Berkeley, CA 94720
More informationRenormalization Group: non perturbative aspects and applications in statistical and solid state physics.
Renormalization Group: non perturbative aspects and applications in statistical and solid state physics. Bertrand Delamotte Saclay, march 3, 2009 Introduction Field theory: - infinitely many degrees of
More informationMachine Learning with Tensor Networks
Machine Learning with Tensor Networks E.M. Stoudenmire and David J. Schwab Advances in Neural Information Processing 29 arxiv:1605.05775 Beijing Jun 2017 Machine learning has physics in its DNA # " # #
More informationMetropolis/Variational Monte Carlo. Microscopic Theories of Nuclear Structure, Dynamics and Electroweak Currents June 12-30, 2017, ECT*, Trento, Italy
Metropolis/Variational Monte Carlo Stefano Gandolfi Los Alamos National Laboratory (LANL) Microscopic Theories of Nuclear Structure, Dynamics and Electroweak Currents June 12-30, 2017, ECT*, Trento, Italy
More informationA Bayesian Approach to Phylogenetics
A Bayesian Approach to Phylogenetics Niklas Wahlberg Based largely on slides by Paul Lewis (www.eeb.uconn.edu) An Introduction to Bayesian Phylogenetics Bayesian inference in general Markov chain Monte
More informationJ ij S i S j B i S i (1)
LECTURE 18 The Ising Model (References: Kerson Huang, Statistical Mechanics, Wiley and Sons (1963) and Colin Thompson, Mathematical Statistical Mechanics, Princeton Univ. Press (1972)). One of the simplest
More informationLecture 7 and 8: Markov Chain Monte Carlo
Lecture 7 and 8: Markov Chain Monte Carlo 4F13: Machine Learning Zoubin Ghahramani and Carl Edward Rasmussen Department of Engineering University of Cambridge http://mlg.eng.cam.ac.uk/teaching/4f13/ Ghahramani
More informationQuantum Monte Carlo Simulations in the Valence Bond Basis
NUMERICAL APPROACHES TO QUANTUM MANY-BODY SYSTEMS, IPAM, January 29, 2009 Quantum Monte Carlo Simulations in the Valence Bond Basis Anders W. Sandvik, Boston University Collaborators Kevin Beach (U. of
More informationQuantum disordering magnetic order in insulators, metals, and superconductors
Quantum disordering magnetic order in insulators, metals, and superconductors Perimeter Institute, Waterloo, May 29, 2010 Talk online: sachdev.physics.harvard.edu HARVARD Cenke Xu, Harvard arxiv:1004.5431
More informationDirectional Ordering in the Classical Compass Model in Two and Three Dimensions
Institut für Theoretische Physik Fakultät für Physik und Geowissenschaften Universität Leipzig Diplomarbeit Directional Ordering in the Classical Compass Model in Two and Three Dimensions vorgelegt von
More informationUnitary Fermi Gas: Quarky Methods
Unitary Fermi Gas: Quarky Methods Matthew Wingate DAMTP, U. of Cambridge Outline Fermion Lagrangian Monte Carlo calculation of Tc Superfluid EFT Random matrix theory Fermion L Dilute Fermi gas, 2 spins
More informationGround State Projector QMC in the valence-bond basis
Quantum Monte Carlo Methods at Work for Novel Phases of Matter Trieste, Italy, Jan 23 - Feb 3, 2012 Ground State Projector QMC in the valence-bond basis Anders. Sandvik, Boston University Outline: The
More informationCS 781 Lecture 9 March 10, 2011 Topics: Local Search and Optimization Metropolis Algorithm Greedy Optimization Hopfield Networks Max Cut Problem Nash
CS 781 Lecture 9 March 10, 2011 Topics: Local Search and Optimization Metropolis Algorithm Greedy Optimization Hopfield Networks Max Cut Problem Nash Equilibrium Price of Stability Coping With NP-Hardness
More informationPY 502, Computational Physics, Fall 2017 Monte Carlo simulations in classical statistical physics
PY 502, Computational Physics, Fall 2017 Monte Carlo simulations in classical statistical physics Anders W. Sandvik, Department of Physics, Boston University 1 Introduction Monte Carlo simulation is a
More informationMetropolis Monte Carlo simulation of the Ising Model
Metropolis Monte Carlo simulation of the Ising Model Krishna Shrinivas (CH10B026) Swaroop Ramaswamy (CH10B068) May 10, 2013 Modelling and Simulation of Particulate Processes (CH5012) Introduction The Ising
More informationImprovement of Monte Carlo estimates with covariance-optimized finite-size scaling at fixed phenomenological coupling
Improvement of Monte Carlo estimates with covariance-optimized finite-size scaling at fixed phenomenological coupling Francesco Parisen Toldin Max Planck Institute for Physics of Complex Systems Dresden
More informationFig. 1 Cluster flip: before. The region inside the dotted line is flipped in one Wolff move. Let this configuration be A.
Physics 6562: Statistical Mechanics http://www.physics.cornell.edu/sethna/teaching/562/ In Class Exercises Last correction at March 25, 2017, 1:37 pm c 2017, James Sethna, all rights reserved 1. Detailed
More informationQuantum phase transitions of insulators, superconductors and metals in two dimensions
Quantum phase transitions of insulators, superconductors and metals in two dimensions Talk online: sachdev.physics.harvard.edu HARVARD Outline 1. Phenomenology of the cuprate superconductors (and other
More informationA = {(x, u) : 0 u f(x)},
Draw x uniformly from the region {x : f(x) u }. Markov Chain Monte Carlo Lecture 5 Slice sampler: Suppose that one is interested in sampling from a density f(x), x X. Recall that sampling x f(x) is equivalent
More informationQuantum Simulation Studies of Charge Patterns in Fermi-Bose Systems
Quantum Simulation Studies of Charge Patterns in Fermi-Bose Systems 1. Hubbard to Holstein 2. Peierls Picture of CDW Transition 3. Phonons with Dispersion 4. Holstein Model on Honeycomb Lattice 5. A New
More informationProbabilistic Machine Learning
Probabilistic Machine Learning Bayesian Nets, MCMC, and more Marek Petrik 4/18/2017 Based on: P. Murphy, K. (2012). Machine Learning: A Probabilistic Perspective. Chapter 10. Conditional Independence Independent
More informationCPSC 540: Machine Learning
CPSC 540: Machine Learning MCMC and Non-Parametric Bayes Mark Schmidt University of British Columbia Winter 2016 Admin I went through project proposals: Some of you got a message on Piazza. No news is
More informationQuantum Monte Carlo. Matthias Troyer, ETH Zürich
Quantum Monte Carlo Matthias Troyer, ETH Zürich 1 1. Monte Carlo Integration 2 Integrating a function Convert the integral to a discrete sum b! f (x)dx = b " a N a N ' i =1 # f a + i b " a % $ N & + O(1/N)
More informationMarkov Chain Monte Carlo Simulations and Their Statistical Analysis An Overview
Markov Chain Monte Carlo Simulations and Their Statistical Analysis An Overview Bernd Berg FSU, August 30, 2005 Content 1. Statistics as needed 2. Markov Chain Monte Carlo (MC) 3. Statistical Analysis
More informationA quick introduction to Markov chains and Markov chain Monte Carlo (revised version)
A quick introduction to Markov chains and Markov chain Monte Carlo (revised version) Rasmus Waagepetersen Institute of Mathematical Sciences Aalborg University 1 Introduction These notes are intended to
More informationKyle Reing University of Southern California April 18, 2018
Renormalization Group and Information Theory Kyle Reing University of Southern California April 18, 2018 Overview Renormalization Group Overview Information Theoretic Preliminaries Real Space Mutual Information
More informationPhysics 115/242 Monte Carlo simulations in Statistical Physics
Physics 115/242 Monte Carlo simulations in Statistical Physics Peter Young (Dated: May 12, 2007) For additional information on the statistical Physics part of this handout, the first two sections, I strongly
More informationan introduction to bayesian inference
with an application to network analysis http://jakehofman.com january 13, 2010 motivation would like models that: provide predictive and explanatory power are complex enough to describe observed phenomena
More informationClassical boson sampling algorithms and the outlook for experimental boson sampling
Classical boson sampling algorithms and the outlook for experimental boson sampling A. Neville, C. Sparrow, R. Clifford, E. Johnston, P. Birchall, A. Montanaro, A. Laing University of Bristol & P. Clifford
More informationGPU-based computation of the Monte Carlo simulation of classical spin systems
Perspectives of GPU Computing in Physics and Astrophysics, Sapienza University of Rome, Rome, Italy, September 15-17, 2014 GPU-based computation of the Monte Carlo simulation of classical spin systems
More informationProbabilistic Graphical Models
10-708 Probabilistic Graphical Models Homework 3 (v1.1.0) Due Apr 14, 7:00 PM Rules: 1. Homework is due on the due date at 7:00 PM. The homework should be submitted via Gradescope. Solution to each problem
More informationMonte Carlo Simulation of the Ising Model. Abstract
Monte Carlo Simulation of the Ising Model Saryu Jindal 1 1 Department of Chemical Engineering and Material Sciences, University of California, Davis, CA 95616 (Dated: June 9, 2007) Abstract This paper
More informationTemperature-dependence of magnetism of free Fe clusters
Temperature-dependence of magnetism of free Fe clusters O. Šipr 1, S. Bornemann 2, J. Minár 2, S. Polesya 2, H. Ebert 2 1 Institute of Physics, Academy of Sciences CR, Prague, Czech Republic 2 Universität
More informationOn Markov Chain Monte Carlo
MCMC 0 On Markov Chain Monte Carlo Yevgeniy Kovchegov Oregon State University MCMC 1 Metropolis-Hastings algorithm. Goal: simulating an Ω-valued random variable distributed according to a given probability
More informationOn the Optimal Scaling of the Modified Metropolis-Hastings algorithm
On the Optimal Scaling of the Modified Metropolis-Hastings algorithm K. M. Zuev & J. L. Beck Division of Engineering and Applied Science California Institute of Technology, MC 4-44, Pasadena, CA 925, USA
More information3.320: Lecture 19 (4/14/05) Free Energies and physical Coarse-graining. ,T) + < σ > dµ
3.320: Lecture 19 (4/14/05) F(µ,T) = F(µ ref,t) + < σ > dµ µ µ ref Free Energies and physical Coarse-graining T S(T) = S(T ref ) + T T ref C V T dt Non-Boltzmann sampling and Umbrella sampling Simple
More informationMonte Carlo Simulation of the 2D Ising model
Monte Carlo Simulation of the 2D Ising model Emanuel Schmidt, F44 April 6, 2 Introduction Monte Carlo methods are a powerful tool to solve problems numerically which are dicult to be handled analytically.
More informationSpin liquids on the triangular lattice
Spin liquids on the triangular lattice ICFCM, Sendai, Japan, Jan 11-14, 2011 Talk online: sachdev.physics.harvard.edu HARVARD Outline 1. Classification of spin liquids Quantum-disordering magnetic order
More informationarxiv: v2 [cond-mat.stat-mech] 13 Oct 2010
Markov Chain Monte Carlo Method without Detailed Balance Hidemaro Suwa 1 and Synge Todo 1,2 1 Department of Applied Physics, University of Tokyo, Tokyo 113-8656, Japan 2 CREST, Japan Science and Technology
More informationQuantum Metropolis Sampling. K. Temme, K. Vollbrecht, T. Osborne, D. Poulin, F. Verstraete arxiv:
Quantum Metropolis Sampling K. Temme, K. Vollbrecht, T. Osborne, D. Poulin, F. Verstraete arxiv: 09110365 Quantum Simulation Principal use of quantum computers: quantum simulation Quantum Chemistry Determination
More informationMonte Carlo. Lecture 15 4/9/18. Harvard SEAS AP 275 Atomistic Modeling of Materials Boris Kozinsky
Monte Carlo Lecture 15 4/9/18 1 Sampling with dynamics In Molecular Dynamics we simulate evolution of a system over time according to Newton s equations, conserving energy Averages (thermodynamic properties)
More informationQuantum phase transitions of insulators, superconductors and metals in two dimensions
Quantum phase transitions of insulators, superconductors and metals in two dimensions Talk online: sachdev.physics.harvard.edu HARVARD Outline 1. Phenomenology of the cuprate superconductors (and other
More informationComparison of Modern Stochastic Optimization Algorithms
Comparison of Modern Stochastic Optimization Algorithms George Papamakarios December 214 Abstract Gradient-based optimization methods are popular in machine learning applications. In large-scale problems,
More informationL échantillonnage parfait: un nouveau paradigme des calculs de Monte Carlo
L échantillonnage parfait: un nouveau paradigme des calculs de Monte Carlo Séminaire général Département de Physique Ecole normale supérieure Werner Krauth Laboratoire de physique statistique Ecole normale
More informationQuantum criticality in the cuprate superconductors. Talk online: sachdev.physics.harvard.edu
Quantum criticality in the cuprate superconductors Talk online: sachdev.physics.harvard.edu The cuprate superconductors Destruction of Neel order in the cuprates by electron doping, R. K. Kaul, M. Metlitksi,
More informationState-Space Methods for Inferring Spike Trains from Calcium Imaging
State-Space Methods for Inferring Spike Trains from Calcium Imaging Joshua Vogelstein Johns Hopkins April 23, 2009 Joshua Vogelstein (Johns Hopkins) State-Space Calcium Imaging April 23, 2009 1 / 78 Outline
More informationSign-problem-free Quantum Monte Carlo of the onset of antiferromagnetism in metals
Sign-problem-free Quantum Monte Carlo of the onset of antiferromagnetism in metals Subir Sachdev sachdev.physics.harvard.edu HARVARD Max Metlitski Erez Berg HARVARD Max Metlitski Erez Berg Sean Hartnoll
More informationSTA 4273H: Statistical Machine Learning
STA 4273H: Statistical Machine Learning Russ Salakhutdinov Department of Computer Science! Department of Statistical Sciences! rsalakhu@cs.toronto.edu! h0p://www.cs.utoronto.ca/~rsalakhu/ Lecture 7 Approximate
More informationMarkov Chain Monte Carlo (MCMC)
Markov Chain Monte Carlo (MCMC Dependent Sampling Suppose we wish to sample from a density π, and we can evaluate π as a function but have no means to directly generate a sample. Rejection sampling can
More informationBalancing and Control of a Freely-Swinging Pendulum Using a Model-Free Reinforcement Learning Algorithm
Balancing and Control of a Freely-Swinging Pendulum Using a Model-Free Reinforcement Learning Algorithm Michail G. Lagoudakis Department of Computer Science Duke University Durham, NC 2778 mgl@cs.duke.edu
More informationIrreversible Markov-Chain Monte Carlo methods
Irreversible Markov-Chain Monte Carlo methods Koji HUKUSHIMA Department of Basic Science, University of Tokyo 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902 Abstract We review irreversible Markov chain Monte
More information