Simulation of rare events by Adaptive Multilevel Splitting algorithms Charles-Edouard Bréhier CNRS & Institut Camille Jordan, Université Lyon 1 Joint works with M. Gazeau (Borealis AI, Toronto) , L. Goudenège (CNRS, Centrale Paris) , T. Lelièvre (Ecole des Ponts, Paris) , M. Rousset (Inria, Rennes) F. Bouchet, C. Herbert, T. Lestang (ENS Lyon, Physics Dept) , F. Ragone (Department of Earth and Environmental Sciences, Milano)
Motivations: rare event in metastable dynamics Goal: simulation of metastable stochastic processes = small noise perturbations of dynamics having multiple stationary points.
Motivations: rare event in metastable dynamics Goal: simulation of metastable stochastic processes = small noise perturbations of dynamics having multiple stationary points. Examples: √ ◮ SDE: dX t = −∇ V ( X t ) dt + 2 ǫ dW ( t ) ◮ Double-well potential: V ( x ) = x 4 4 − x 2 2 .
Motivations: rare event in metastable dynamics Goal: simulation of metastable stochastic processes = small noise perturbations of dynamics having multiple stationary points. Exemples: ◮ ◮ SPDE: Allen-Cahn equation √ ∂ u ( t , x ) = γ ∆ u ( t , x ) − ∇ V ( u ( t , x )) + 2 ǫξ ( t , x ) ∂ t
Transitions between metastable states One goal is to estimate p = P x 0 ( τ B < τ A ) ⊥ΧΘΕ∴ A B where A and B are metastable states. ◮ Direct numerical simulations : prohibitive, need N ∼ Cp − 1 independent realizations. ◮ Analytic approach : Large Deviations, Potential Theory . In the limit ǫ → 0, ǫ → 0 ce − C /ǫ . ǫ log ( p ( ǫ )) → ǫ → 0 − C , p ( ǫ ) ∼ Transition times of the order e C /ǫ .
Splitting algorithms Introduced in the 1950s: H. Kahn and T. E. Harris. Estimation of particle transmission by random sampling. National Bureau of Standards , 12:27–30, 1951.
Splitting algorithms Introduced in the 1950s: H. Kahn and T. E. Harris. Estimation of particle transmission by random sampling. National Bureau of Standards , 12:27–30, 1951. Revival in the 1990s and 2000s: P. Glasserman, P. Heidelberger, P. Shahabuddin, and T. Zajic. Multilevel splitting for estimating rare event probabilities. Oper. Res. , 47(4):585–600, 1999. with many variants: M. Villén-Altamirano and J. Villén-Altamirano. RESTART: A method for accelerating rare events simulations. In Proceeding of the thirteenth International Teletraffic Congress , volume Copenhagen, Denmark, June 19-26 of Queueing, performance and control in ATM: ITC-13 workshops , pages 71–76. North-Holland, Amsterdam-New York, 1991. S. K. Au and J. L. Beck. Estimation of small failure probabilities in high dimensions by subset simulation. Journal of Probabilistic Engineering Mechanics , 16:263–277, 2001. J. Skilling. Nested sampling for general Bayesian computation. Bayesian Anal. , 1(4):833–859 (electronic), 2006. P. Del Moral, A. Doucet, and A. Jasra. Sequential Monte Carlo samplers. J. R. Stat. Soc. Ser. B Stat. Methodol. , 68(3):411–436, 2006.
Adaptive Multilevel Splitting algorithms Introduced in F. Cérou and A. Guyader. Adaptive multilevel splitting for rare event analysis. Stoch. Anal. Appl. , 25(2):417–443, 2007. Analysis of the method: series of recent works (non exhaustive) E. Simonnet. Combinatorial analysis of the adaptive last particle method. Statistics and Computing , pages 1–20, 2014. C.-E. Bréhier, T. Lelièvre, and M. Rousset. Analysis of adaptive multilevel splitting algorithms in an idealized setting. ESAIM Probability and Statistics , 19:361–394, 2015. F. Cérou and A. Guyader. Fluctuation analysis of adaptive multilevel splitting. Ann. Appl. Probab. , 26(6):3319–3380, 2016. C.-E. Bréhier, M. Gazeau, L. Goudenège, T. Lelièvre, and M. Rousset. Unbiasedness of some generalized adaptive multilevel splitting algorithms. Ann. Appl. Probab. , 26(6):3559–3601, 2016.
Applications of AMS algorithms J. Rolland and E. Simonnet. Statistical behaviour of adaptive multilevel splitting algorithms in simple models. Journal of Computational Physics , 283:541 – 558, 2015. D. Aristoff, T. Lelièvre, C. G. Mayne, and I. Teo. Adaptive multilevel splitting in molecular dynamics simulations. In CEMRACS 2013—modelling and simulation of complex systems: stochastic and deterministic approaches , volume 48 of ESAIM Proc. Surveys , pages 215–225. EDP Sci., Les Ulis, 2015. Several recently defended and ongoing PhD Thesis: R. Poncet (Polytechnique 10-2017), H. Louvin (CEA 10-2017), L. J. Lopes (Ponts-Sanofi), T. Lestang (ENS Lyon, Physics). H. Louvin, E. Dumonteil, T. Lelièvre, M. Rousset, and C. M. Diop. Adaptive multilevel splitting for monte carlo particle transport. In EPJ Web of Conferences , volume 153, page 06006. EDP Sciences, 2017. L. J. Lopes, C. Chipot, and T. Lelièvre. Adaptive multilevel splitting method: Isomerization of the alanine dipeptide. 2017. Collaboration with F. Bouchet, C. Herbert, T. Lestang, F. Ragone.
Principle of splitting algorithms Principles: iterative algorithm ◮ System of interacting replicas: copies of the Markov trajectories are simulated (in parallel). ◮ Adaptive Splitting: selection of replicas using a score function. ◮ Partial resampling: new copies are simulated, by branching replicas (mutation). ◮ Weighting: remain consistent when removing/creating replicas. An important property: the (Markov) model is a black box. Simpler setting: estimation of p = P ( Y > a ) . with ◮ Y positive real-valued random variable (with continuous cdf) ◮ a threshold.
General notions: estimators, consistency, efficiency An estimator of θ ∈ R is a random variable ˆ θ .
General notions: estimators, consistency, efficiency An estimator of θ ∈ R is a random variable ˆ θ . About consistency: ◮ The bias of the estimator is defined as b (ˆ θ ) = E [ˆ θ ] − θ . ◮ We say that ˆ θ is an unbiased estimator of θ if b (ˆ θ ) = 0. � ˆ ◮ Let � θ N N ∈ N be a sequence of estimators of θ . It is called consistent if ˆ N →∞ θ in probability. It is called asymptotically unbiased if → θ N b (ˆ θ N ) N →∞ 0. →
General notions: estimators, consistency, efficiency An estimator of θ ∈ R is a random variable ˆ θ . About consistency: ◮ The bias of the estimator is defined as b (ˆ θ ) = E [ˆ θ ] − θ . ◮ We say that ˆ θ is an unbiased estimator of θ if b (ˆ θ ) = 0. � ˆ ◮ Let � θ N N ∈ N be a sequence of estimators of θ . It is called consistent if ˆ N →∞ θ in probability. It is called asymptotically unbiased if → θ N b (ˆ θ N ) N →∞ 0. → About efficiency: mean-square error � 2 + E | ˆ θ − θ | 2 = θ | 2 = b (ˆ θ ) 2 + Var (ˆ E | ˆ E ˆ θ − E ˆ � θ − θ θ ) sum of bias squared (consistency error) and variance (statistical error). Example: N m N = 1 � ˆ Z n N n = 1 � � with i.i.d. Z n n ∈ N is a consistent unbiased estimator of m = E [ Z 1 ] . m N ) = Var ( Z 1 ) Statistical error: Var ( ˆ . More precisely: Central Limit Theorem. N
The rare event case Estimation of p = P ( X ∈ A ) = E [ 1 X ∈A ] ? Direct approach: crude Monte-Carlo simulation Take Z = 1 X ∈A , and i.i.d. realizations Z 1 , . . . , Z n = 1 X n ∈A , . . . , and compute the empirical average N N p N = 1 Z n = 1 � � ˆ 1 X n ∈A . N N n = 1 n = 1 Then ˆ p N is a consistent unbiased estimator of p = E [ Z ] . Relative error: � � E | ˆ p N − p | 2 p ( 1 − p ) 1 = √ ∼ √ pN . p p N p → 0
The rare event case Estimation of p = P ( X ∈ A ) = E [ 1 X ∈A ] ? Direct approach: crude Monte-Carlo simulation Take Z = 1 X ∈A , and i.i.d. realizations Z 1 , . . . , Z n = 1 X n ∈A , . . . , and compute the empirical average N N p N = 1 Z n = 1 � � ˆ 1 X n ∈A . N N n = 1 n = 1 Then ˆ p N is a consistent unbiased estimator of p = E [ Z ] . Relative error: � � E | ˆ p N − p | 2 p ( 1 − p ) 1 = √ ∼ √ pN . p p N p → 0 Variance reduction strategy: find a family of simulatable estimators ˆ θ to reduce the cost. Requirements (to also remain unbiased) Var (ˆ E [ˆ θ ) ≪ Var (ˆ p N ) θ ] = p , where simulations of ˆ θ and ˆ p N have the same average cost .
Splitting algorithm Case study: p = P ( Y > a ) , with Y > 0. Splitting: N N � � � � � p = Y > a j � Y > a j − 1 = p j P j = 1 j = 1 with a non-decreasing sequence of levels 0 = a 0 < a 1 < . . . < a N − 1 < a N = a . p = � N Estimator: ˆ j = 1 p j , n rep , ( N independent estimators, with n rep independent replicas per level) is unbiased. Optimal choice of levels a j : such that p 1 = . . . = p N = p 1 / N . √ | log ( p ) | Relative error, N → + ∞ : ∼ . √ N
Splitting algorithm Case study: p = P ( Y > a ) , with Y > 0. Splitting: N N � � � � � p = Y > a j � Y > a j − 1 = p j P j = 1 j = 1 with a non-decreasing sequence of levels 0 = a 0 < a 1 < . . . < a N − 1 < a N = a . p = � N Estimator: ˆ j = 1 p j , n rep , ( N independent estimators, with n rep independent replicas per level) is unbiased. Optimal choice of levels a j : such that p 1 = . . . = p N = p 1 / N . √ | log ( p ) | Relative error, N → + ∞ : ∼ . √ N Adaptive version: computes a random sequence Z 0 , . . . of levels with k p j = 1 − n rep .
Recommend
More recommend