Particle filters or Sequential Monte Carlo (SMC) methods are a set of Monte Carlo algorithms used to solve filtering problems arising in signal processing and Bayesian statistical inference. The filtering problem consists of estimating the internal states in dynamical systems when partial observations are made, and random perturbations are present in the sensors as well as in the dynamical system. The objective is to compute the posterior distributions of the states of some Markov process, given some noisy and partial observations. The term "particle filters" was first coined in 1996 by Del Moral[1] in reference to mean field interacting particle methods used in fluid mechanics since the beginning of the 1960s. The terminology "sequential Monte Carlo" was proposed by Liu and Chen in 1998[2].

Particle filtering uses a set of particles (also called samples) to represent the posterior distribution of some stochastic process given noisy and/or partial observations. The state-space model can be nonlinear and the initial state and noise distributions can take any form required. Particle filter techniques provide a well-established methodology[1][3][4] for generating samples from the required distribution without requiring assumptions about the state-space model or the state distributions. However, these methods do not perform well when applied to very high-dimensional systems.

Particle filters implement the prediction-updating updates in an approximate manner. The samples from the distribution are represented by a set of particles; each particle has a likelihood weight assigned to it that represents the probability of that particle being sampled from the probability density function. Weight disparity leading to weight collapse is a common issue encountered in these filtering algorithms; however it can be mitigated by including a resampling step before the weights become too uneven. Several adaptive resampling criteria can be used, including the variance of the weights and the relative entropy with respect to the uniform distribution.[5] In the resampling step, the particles with negligible weights are replaced by new particles in the proximity of the particles with higher weights.

From the statistical and probabilistic point of view, particle filters can be interpreted as mean field particle interpretations of Feynman-Kac probability measures.[6][7][8][9][10] These particle integration techniques were developed in molecular chemistry and computational physics by Theodore E. Harris and Herman Kahn in 1951, Marshall N. Rosenbluth and Arianna W. Rosenbluth in 1955[11] and more recently by Jack H. Hetherington in 1984.[12] In computational physics, these Feynman-Kac type path particle integration methods are also used in Quantum Monte Carlo, and more specifically Diffusion Monte Carlo methods.[13][14][15] Feynman-Kac interacting particle methods are also strongly related to mutation-selection genetic algorithms currently used in evolutionary computing to solve complex optimization problems.

The particle filter methodology is used to solve Hidden Markov Model (HMM) and nonlinear filtering problems. With the notable exception of linear-Gaussian signal-observation models (Kalman filter) or wider classes of models (Benes filter[16]) Mireille Chaleyat-Maurel and Dominique Michel proved in 1984 that the sequence of posterior distributions of the random states of the signal given the observations (a.k.a. optimal filter) have no finitely recursive recursion.[17] Various numerical methods based on fixed grid approximations, Markov Chain Monte Carlo techniques (MCMC), conventional linearization, extended Kalman filters, or determining the best linear system (in expect cost-error sense) have never really coped with large scale systems, unstable processes or when the nonlinearities are not sufficiently smooth.

Particle filters and Feynman-Kac particle methodologies find application in signal and image processing, Bayesian inference, machine learning, risk analysis and rare event sampling, engineering and robotics, artificial intelligence, bioinformatics,[18] phylogenetics, computational science, Economics and mathematical finance, molecular chemistry, computational physics, pharmacokinetic and other fields.

HistoryEdit

Heuristic like algorithmsEdit

From the statistical and probabilistic viewpoint, particle filters belong to the class of branching/genetic type algorithms, and mean field type interacting particle methodologies. The interpretation of these particle methods depends on the scientific discipline. In Evolutionary Computing, mean field genetic type particle methodologies are often used as a heuristic and natural search algorithms (a.k.a. Metaheuristic). In computational physics and molecular chemistry they are used to solve Feynman-Kac path integration problems, or they compute Boltzmann-Gibbs measures, top eigenvalues and ground states of Schrödinger operators. In Biology and Genetics they also represent the evolution of a population of individuals or genes in some environment.

The origins of mean field type evolutionary computational techniques can be traced to 1950 and 1954 with the seminal work of Alan Turing on genetic type mutation-selection learning machines[19] and the articles by Nils Aall Barricelli at the Institute for Advanced Study in Princeton, New Jersey.[20][21] The first trace of particle filters in statistical methodology dates back to the mid-1950s; the 'Poor Man's Monte Carlo',[22] that was proposed by Hammersley et al., in 1954, contained hints of the genetic type particle filtering methods used today. In 1963, Nils Aall Barricelli simulated a genetic type algorithm to mimic the ability of individuals to play a simple game.[23] In evolutionary computing literature, genetic type mutation-selection algorithms became popular through the seminal work of John Holland in the early 1970s, and particularly his book[24] published in 1975.

In Biology and Genetics, the Australian geneticist Alex Fraser also published in 1957 a series of papers on the genetic type simulation of artificial selection of organisms.[25] The computer simulation of evolution by biologists became more common in the early 1960s, and the methods were described in books by Fraser and Burnell (1970)[26] and Crosby (1973).[27] Fraser's simulations included all of the essential elements of modern mutation-selection genetic particle algorithms.

From the mathematical viewpoint, the conditional distribution of the random states of a signal given some partial and noisy observations is described by a Feynman-Kac probability on the random trajectories of the signal weighted by a sequence of likelihood potential functions.[6][7] Quantum Monte Carlo, and more specifically Diffusion Monte Carlo methods can also be interpreted as a mean field genetic type particle approximation of Feynman-Kac path integrals.[6][7][8][12][13][28][29] The origins of Quantum Monte Carlo methods are often attributed to Enrico Fermi and Robert Richtmyer who developed in 1948 a mean field particle interpretation of neutron-chain reactions,[30] but the first heuristic-like and genetic type particle algorithm (a.k.a. Resampled or Reconfiguration Monte Carlo methods) for estimating ground state energies of quantum systems (in reduced matrix models) is due to Jack H. Hetherington in 1984.[12] We also quote an earlier seminal works of Theodore E. Harris and Herman Kahn in particle physics, published in 1951, using mean field but heuristic-like genetic methods for estimating particle transmission energies.[31] In molecular chemistry, the use of genetic heuristic-like particle methodologies (a.k.a. pruning and enrichment strategies) can be traced back to 1955 with the seminal work of Marshall. N. Rosenbluth and Arianna. W. Rosenbluth.[11]

The use of genetic particle algorithms in advanced signal processing and Bayesian inference is more recent. In January 1993, Genshiro Kitagawa developed a "Monte Carlo filter"[32], a slightly modified version of this article appearing in 1996[33]. In April 1993, Gordon et al., published in their seminal work[34] an application of genetic type algorithm in Bayesian statistical inference. The authors named their algorithm 'the bootstrap filter', and demonstrated that compared to other filtering methods, their bootstrap algorithm does not require any assumption about that state-space or the noise of the system. Independently, the ones by Pierre Del Moral[1] and Himilcon Carvalho, Pierre Del Moral, André Monin and Gérard Salut[35] on particle filters published in the mid-1990s. Particle filters were also developed in signal processing in the early 1989-1992 by P. Del Moral, J.C. Noyer, G. Rigal, and G. Salut in the LAAS-CNRS in a series of restricted and classified research reports with STCAN (Service Technique des Constructions et Armes Navales), the IT company DIGILOG, and the LAAS-CNRS (the Laboratory for Analysis and Architecture of Systems) on RADAR/SONAR and GPS signal processing problems.[36][37][38][39][40][41]

Mathematical foundationsEdit

From 1950 to 1996, all the publications on particle filters, genetic algorithms, including the pruning and resample Monte Carlo methods introduced in computational physics and molecular chemistry, present natural and heuristic-like algorithms applied to different situations without a single proof of their consistency, nor a discussion on the bias of the estimates and on genealogical and ancestral tree based algorithms.

The mathematical foundations and the first rigorous analysis of these particle algorithms are due to Pierre Del Moral[1][3] in 1996. The article[1] also contains a proof of the unbiased properties of a particle approximations of likelihood functions and unnormalized conditional probability measures. The unbiased particle estimator of the likelihood functions presented in this article is used today in Bayesian statistical inference.

Branching type particle methodologies with varying population sizes were also developed toward the end of the 1990s by Dan Crisan, Jessica Gaines and Terry Lyons,[42][43][44] and by Dan Crisan, Pierre Del Moral and Terry Lyons.[45] Further developments in this field were developed in 2000 by P. Del Moral, A. Guionnet and L. Miclo.[7][46][47] The first central limit theorems are due to Pierre Del Moral and Alice Guionnet[48] in 1999 and Pierre Del Moral and Laurent Miclo[7] in 2000. The first uniform convergence results with respect to the time parameter for particle filters were developed in the end of the 1990s by Pierre Del Moral and Alice Guionnet.[46][47] The first rigorous analysis of genealogical tree based particle filter smoothers is due to P. Del Moral and L. Miclo in 2001[49]

The theory on Feynman-Kac particle methodologies and related particle filters algorithms has been developed in 2000 and 2004 in the books.[7][4] These abstract probabilistic models encapsulate genetic type algorithms, particle and bootstrap filters, interacting Kalman filters (a.k.a. Rao–Blackwellized particle filter[50]), importance sampling and resampling style particle filter techniques, including genealogical tree based and particle backward methodologies for solving filtering and smoothing problems. Other classes of particle filtering methodologies includes genealogical tree based models,[9][4][51] backward Markov particle models,[9][52] adaptive mean field particle models,[5] island type particle models,[53][54] and particle Markov chain Monte Carlo methodologies.[55][56]

The filtering problemEdit

ObjectiveEdit

The objective of a particle filter is to estimate the posterior density of the state variables given the observation variables. The particle filter is designed for a hidden Markov Model, where the system consists of hidden and observable variables. The observable variables (observation process) are related to the hidden variables (state-process) by some functional form that is known. Similarly the dynamical system describing the evolution of the state variables is also known probabilistically.

A generic particle filter estimates the posterior distribution of the hidden states using the observation measurement process. Consider a state-space shown in the diagram below.

 

The filtering problem is to estimate sequentially the values of the hidden states  , given the values of the observation process   at any time step k.

All Bayesian estimates of   follow from the posterior density p(xk | y0,y1,…,yk). The particle filter methodology provides an approximation of these conditional probabilities using the empirical measure associated with a genetic type particle algorithm. In contrast, the MCMC or importance sampling approach would model the full posterior p(x0,x1,…,xk | y0,y1,…,yk).

The Signal-Observation ModelEdit

Particle methods often assume   and the observations   can be modeled in this form:

  •   is a Markov process on   (for some  ) that evolves according to the transition probability density  . This model is also often written in a synthetic way as
     
with an initial probability density  .
  • The observations   take values in some state space on   (for some  ) and are conditionally independent provided that   are known. In other words, each   only depends on  . In addition, we assume conditional distribution for   given   are absolutely continuous, and in a synthetic way we have
     

An example of system with these properties is:

 
 

where both   and   are mutually independent sequences with known probability density functions and g and h are known functions. These two equations can be viewed as state space equations and look similar to the state space equations for the Kalman filter. If the functions g and h in the above example are linear, and if both   and   are Gaussian, the Kalman filter finds the exact Bayesian filtering distribution. If not, Kalman filter based methods are a first-order approximation (EKF) or a second-order approximation (UKF in general, but if probability distribution is Gaussian a third-order approximation is possible).

The assumption that the initial distribution and the transitions of the Markov chain are absolutely continuous with respect to the Lebesgue measure can be relaxed. To design a particle filter we simply need to assume that we can sample the transitions   of the Markov chain   and to compute the likelihood function   (see for instance the genetic selection mutation description of the particle filter given below). The absolutely continuous assumption on the Markov transitions of   are only used to derive in an informal (and rather abusive) way different formulae between posterior distributions using the Bayes' rule for conditional densities.

Approximate Bayesian Computation modelsEdit

In some important problems, the conditional distribution of the observations given the random states of the signal may fail to have a density or may be impossible or too complex to compute.[18] In this situation, we need to resort to an additional level of approximation. One strategy is to replace the signal   by the Markov chain   and to introduce a virtual observation of the form

 

for some sequence of independent sequences with known probability density functions. The central idea is to observe that

 

The particle filter associated with the Markov process   given the partial observations   is defined in terms of particles evolving in   with a likelihood function given with some obvious abusive notation by  . These probabilistic techniques are closely related to Approximate Bayesian Computation (ABC). In the context of particle filters, these ABC particle filtering techniques were introduced in 1998 by P. Del Moral, J. Jacod and P. Protter.[57] They were further developed by P. Del Moral, A. Doucet and A. Jasra.[58][59]

The nonlinear filtering equationEdit

Bayes' rule for conditional probability gives:

 

where

 

Particle filters are also an approximation, but with enough particles they can be much more accurate.[1][3][4][46][47] The nonlinear filtering equation is given by the recursion

 

 

 

 

 

(Eq. 1)

with the convention   for k = 0. The nonlinear filtering problem consists in computing these conditional distributions sequentially.

Feynman-Kac formulationEdit

We fix a time horizon n and a sequence of observations  , and for each k = 0, ..., n we set:

 

In this notation, for any bounded function F on the set of trajectories of   from the origin k = 0 up to time k = n, we have the Feynman-Kac formula

 

These Feynman-Kac path integration models arise in a variety of scientific disciplines, including in computational physics, biology, information theory and computer sciences.[7][9][4] Their interpretations depend on the application domain. For instance, if we choose the indicator function   of some subset of the state space, they represent the conditional distribution of a Markov chain given it stays in a given tube; that is, we have:

 

and

 

as soon as the normalizing constant is strictly positive.

Particle filtersEdit

A Genetic type particle algorithmEdit

Initially we start with N independent random variables   with common probability density  . The genetic algorithm selection-mutation transitions

 

mimic/approximate the updating-prediction transitions of the optimal filter evolution (Eq. 1):

  • During the selection-updating transition we sample N (conditionally) independent random variables   with common (conditional) distribution
 
  • During the mutation-prediction transition, from each selected particle   we sample independently a transition
 

In the above displayed formulae   stands for the likelihood function   evaluated at  , and   stands for the conditional density   evaluated at  .

At each time k, we have the particle approximations

 

and

 

A detailed proof of these convergence results can be found in,[1][3] see also the more recent developments provided in the books.[9][4] In Genetic algorithms and Evolutionary computing community, the mutation-selection Markov chain described above is often called the genetic algorithm with proportional selection. Several branching variants, including with random population sizes have also been proposed in the articles.[4][42][45]

Monte Carlo principlesEdit

Particle methods, like all sampling-based approaches (e.g., MCMC), generate a set of samples that approximate the filtering density

 

For example, we may have N samples from the approximate posterior distribution of  , where the samples are labeled with superscripts as

 

Then, expectations with respect to the filtering distribution are approximated by

 

 

 

 

 

(Eq. 2)

with

 

where   stands for the Dirac measure at a given state a. The function f, in the usual way for Monte Carlo, can give all the moments etc. of the distribution up to some approximation error. When the approximation equation (Eq. 2) is satisfied for any bounded function f we write

 

Particle filters can be interpreted as a genetic type particle algorithm evolving with mutation and selection transitions. We can keep track of the ancestral lines

 

of the particles  . The random states  , with the lower indices l=0,...,k, stands for the ancestor of the individual   at level l=0,...,k. In this situation, we have the approximation formula

 

 

 

 

 

(Eq. 3)

with the empirical measure

 

Here F stands for any founded function on the path space of the signal. In a more synthetic form (Eq. 3) is equivalent to

 

Particle filters can be interpreted in many different ways. From the probabilistic point of view they coincide with a mean field particle interpretation of the nonlinear filtering equation. The updating-prediction transitions of the optimal filter evolution can also be interpreted as the classical genetic type selection-mutation transitions of individuals. The sequential importance resampling technique provides another interpretation of the filtering transitions coupling importance sampling with the bootstrap resampling step. Last, but not least, particle filters can be seen as an acceptance-rejection methodology equipped with a recycling mechanism.[9][4]

Mean field particle simulationEdit

The general probabilistic principleEdit

The nonlinear filtering evolution can be interpreted as a dynamical system in the set of probability measures of the following form   where   stands for some mapping from the set of probability distribution into itself. For instance, the evolution of the one-step optimal predictor  

satisfies a nonlinear evolution starting with the probability distribution  . One of the simplest ways to approximate these probability measures is to start with N independent random variables   with common probability distribution   . Suppose we have defined a sequence of N random variables   such that

 

At the next step we sample N (conditionally) independent random variables   with common law .

 

A particle interpretation of the filtering equationEdit

We illustrate this mean field particle principle in the context of the evolution of the one step optimal predictors

 

 

 

 

 

(Eq. 4)

For k = 0 we use the convention  .

By the law of large numbers, we have

 

in the sense that

 

for any bounded function  . We further assume that we have constructed a sequence of particles   at some rank k such that

 

in the sense that for any bounded function   we have

 

In this situation, replacing   by the empirical measure   in the evolution equation of the one-step optimal filter stated in (Eq. 4) we find that

 

Notice that the right hand side in the above formula is a weighted probability mixture

 

where   stands for the density   evaluated at  , and   stands for the density   evaluated at   for  

Then, we sample N independent random variable   with common probability density   so that

 

Iterating this procedure, we design a Markov chain such that

 

Notice that the optimal filter is approximated at each time step k using the Bayes' formulae

 

The terminology "mean field approximation" comes from the fact that we replace at each time step the probability measure   by the empirical approximation  . The mean field particle approximation of the filtering problem is far from being unique. Several strategies are developed in the books.[9][4]

Some convergence resultsEdit

The analysis of the convergence of particle filters was started in 1996[1][3] and in 2000 in the book[7] and the series of articles.[45][46][47][48][49][60][61] More recent developments can be found in the books,[9][4] When the filtering equation is stable (in the sense that it corrects any erroneous initial condition), the bias and the variance of the particle particle estimates

 

are controlled by the non asymptotic uniform estimates

 
 

for any function f bounded by 1, and for some finite constants   In addition, for any  :

 

for some finite constants   related to the asymptotic bias and variance of the particle estimate, and some finite constant c. The same results are satisfied if we replace the one step optimal predictor by the optimal filter approximation.

Genealogical trees and Unbiasedness propertiesEdit

Genealogical tree based particle smoothingEdit

Tracing back in time the ancestral lines

 

of the individuals   and   at every time step k, we also have the particle approximations

 

These empirical approximations are equivalent to the particle integral approximations

 

for any bounded function F on the random trajectories of the signal. As shown in[51] the evolution of the genealogical tree coincides with a mean field particle interpretation of the evolution equations associated with the posterior densities of the signal trajectories. For more details on these path space models, we refer to the books.[9][4]

Unbiased particle estimates of likelihood functionsEdit

We use the product formula

 

with

 

and the conventions   and   for k = 0. Replacing   by the empirical approximation

 

in the above displayed formula, we design the following unbiased particle approximation of the likelihood function

 

with

 

where   stands for the density   evaluated at  . The design of this particle estimate and the unbiasedness property has been proved in 1996 in the article.[1] Refined variance estimates can be found in[4] and.[9]

Backward particle smoothersEdit

Using Bayes' rule, we have the formula

 

Notice that

 

This implies that

 

Replacing the one-step optimal predictors   by the particle empirical measures

 

we find that

 

We conclude that

 

with the backward particle approximation

 

The probability measure

 

is the probability of the random paths of a Markov chain  running backward in time from time k=n to time k=0, and evolving at each time step k in the state space associated with the population of particles  

  • Initially (at time k=n) the chain   chooses randomly a state with the distribution
 
  • From time k to the time (k-1), the chain starting at some state   for some   at time k moves at time (k-1) to a random state   chosen with the discrete weighted probability
 

In the above displayed formula,   stands for the conditional distribution   evaluated at  . In the same vein,   and   stand for the conditional densities   and   evaluated at   and   These models allows to reduce integration with respect to the densities   in terms of matrix operations with respect to the Markov transitions of the chain described above.[52] For instance, for any function   we have the particle estimates

 

where

 

This also shows that if

 

then

 

Some convergence resultsEdit

We shall assume that filtering equation is stable, in the sense that it corrects any erroneous initial condition.

In this situation, the particle approximations of the likelihood functions are unbiased and the relative variance is controlled by

 

for some finite constant c. In addition, for any  :

 

for some finite constants   related to the asymptotic bias and variance of the particle estimate, and for some finite constant c.

The bias and the variance of the particle particle estimates based on the ancestral lines of the genealogical trees

 

are controlled by the non asymptotic uniform estimates

 

for any function F bounded by 1, and for some finite constants   In addition, for any  :

 

for some finite constants   related to the asymptotic bias and variance of the particle estimate, and for some finite constant c. The same type of bias and variance estimates hold for the backward particle smoothers. For additive functionals of the form

 

with

 

with functions   bounded by 1, we have

 

and

 

for some finite constants   More refined estimates including exponentially small probability of errors are developed in.[9]

Sequential Importance Resampling (SIR)Edit

Monte Carlo filter and bootstrap filterEdit

Sequential importance Resampling (SIR), Monte Carlo filtering (Kitagawa 1993[32]) and the bootstrap filtering algorithm (Gordon et al. 1993[34]), is also a very commonly used filtering algorithm, which approximates the filtering probability density   by a weighted set of N samples

 

The importance weights   are approximations to the relative posterior probabilities (or densities) of the samples such that

 

Sequential importance sampling (SIS) is a sequential (i.e., recursive) version of importance sampling. As in importance sampling, the expectation of a function f can be approximated as a weighted average

 

For a finite set of samples, the algorithm performance is dependent on the choice of the proposal distribution

 .

The "optimal" proposal distribution is given as the target distribution

 

This particular choice of proposal transition has been proposed by P. Del Moral in 1996 and 1998.[3] When it is difficult to sample transitions according to the distribution   one natural strategy is to use the following particle approximation

 

with the empirical approximation

 

associated with N (or any other large number of samples) independent random samples  with the conditional distribution of the random state   given  . The consistency of the resulting particle filter of this approximation and other extensions are developed in.[3] In the above display   stands for the Dirac measure at a given state a.

However, the transition prior probability distribution is often used as importance function, since it is easier to draw particles (or samples) and perform subsequent importance weight calculations:

 

Sequential Importance Resampling (SIR) filters with transition prior probability distribution as importance function are commonly known as bootstrap filter and condensation algorithm.

Resampling is used to avoid the problem of degeneracy of the algorithm, that is, avoiding the situation that all but one of the importance weights are close to zero. The performance of the algorithm can be also affected by proper choice of resampling method. The stratified sampling proposed by Kitagawa (1993[32]) is optimal in terms of variance.

A single step of sequential importance resampling is as follows:

1) For   draw samples from the proposal distribution
 
2) For   update the importance weights up to a normalizing constant:
 
Note that when we use the transition prior probability distribution as the importance function,
 
this simplifies to the following :
 
3) For   compute the normalized importance weights:
 
4) Compute an estimate of the effective number of particles as
 
This criterion reflects the variance of the weights, other criteria can be found in the article,[5] including their rigorous analysis and central limit theorems.
5) If the effective number of particles is less than a given threshold  , then perform resampling:
a) Draw N particles from the current particle set with probabilities proportional to their weights. Replace the current particle set with this new one.
b) For   set  

The term Sampling Importance Resampling is also sometimes used when referring to SIR filters.

Sequential importance sampling (SIS)Edit

  • Is the same as sequential importance resampling, but without the resampling stage.

"direct version" algorithmEdit

The "direct version" algorithm[citation needed] is rather simple (compared to other particle filtering algorithms) and it uses composition and rejection. To generate a single sample x at k from  :

1) Set n=0 (This will count the number of particles generated so far)
2) Uniformly choose an index i from the range  
3) Generate a test   from the distribution   with  
4) Generate the probability of   using   from   where   is the measured value
5) Generate another uniform u from   where  
6) Compare u and  
6a) If u is larger then repeat from step 2
6b) If u is smaller then save   as   and increment n
7) If n == N then quit

The goal is to generate P "particles" at k using only the particles from  . This requires that a Markov equation can be written (and computed) to generate a   based only upon  . This algorithm uses composition of the P particles from   to generate a particle at k and repeats (steps 2–6) until P particles are generated at k.

This can be more easily visualized if x is viewed as a two-dimensional array. One dimension is k and the other dimensions is the particle number. For example,   would be the ith particle at   and can also be written   (as done above in the algorithm). Step 3 generates a potential   based on a randomly chosen particle ( ) at time   and rejects or accepts it in step 6. In other words, the   values are generated using the previously generated  .

Other particle filtersEdit

See alsoEdit

ReferencesEdit

  1. ^ a b c d e f g h i j Del Moral, Pierre (1996). "Non Linear Filtering: Interacting Particle Solution" (PDF). Markov Processes and Related Fields. 2 (4): 555–580.
  2. ^ Liu, Jun S.; Chen, Rong (1998-09-01). "Sequential Monte Carlo Methods for Dynamic Systems". Journal of the American Statistical Association. 93 (443): 1032–1044. doi:10.1080/01621459.1998.10473765. ISSN 0162-1459.
  3. ^ a b c d e f g Del Moral, Pierre (1998). "Measure Valued Processes and Interacting Particle Systems. Application to Non Linear Filtering Problems". Annals of Applied Probability (Publications du Laboratoire de Statistique et Probabilités, 96-15 (1996) ed.). 8 (2): 438–495. doi:10.1214/aoap/1028903535.
  4. ^ a b c d e f g h i j k l m Del Moral, Pierre (2004). Feynman-Kac formulae. Genealogical and interacting particle approximations. https://www.springer.com/gp/book/9780387202686: Springer. Series: Probability and Applications. p. 556. ISBN 978-0-387-20268-6.CS1 maint: location (link)
  5. ^ a b c Del Moral, Pierre; Doucet, Arnaud; Jasra, Ajay (2012). "On Adaptive Resampling Procedures for Sequential Monte Carlo Methods" (PDF). Bernoulli. 18 (1): 252–278. doi:10.3150/10-bej335.
  6. ^ a b c Del Moral, Pierre (2004). Feynman-Kac formulae. Genealogical and interacting particle approximations. Probability and its Applications. Springer. p. 575. ISBN 9780387202686. Series: Probability and Applications
  7. ^ a b c d e f g h Del Moral, Pierre; Miclo, Laurent (2000). Branching and Interacting Particle Systems Approximations of Feynman-Kac Formulae with Applications to Non-Linear Filtering (PDF). Lecture Notes in Mathematics. 1729. pp. 1–145. doi:10.1007/bfb0103798. ISBN 978-3-540-67314-9.
  8. ^ a b Del Moral, Pierre; Miclo, Laurent (2000). "A Moran particle system approximation of Feynman-Kac formulae". Stochastic Processes and Their Applications. 86 (2): 193–216. doi:10.1016/S0304-4149(99)00094-0.
  9. ^ a b c d e f g h i j k l Del Moral, Pierre (2013). Mean field simulation for Monte Carlo integration. Chapman & Hall/CRC Press. p. 626. Monographs on Statistics & Applied Probability
  10. ^ Moral, Piere Del; Doucet, Arnaud (2014). "Particle methods: An introduction with applications". ESAIM: Proc. 44: 1–46. doi:10.1051/proc/201444001.
  11. ^ a b Rosenbluth, Marshall, N.; Rosenbluth, Arianna, W. (1955). "Monte-Carlo calculations of the average extension of macromolecular chains". J. Chem. Phys. 23 (2): 356–359. Bibcode:1955JChPh..23..356R. doi:10.1063/1.1741967.
  12. ^ a b c Hetherington, Jack, H. (1984). "Observations on the statistical iteration of matrices". Phys. Rev. A. 30 (2713): 2713–2719. Bibcode:1984PhRvA..30.2713H. doi:10.1103/PhysRevA.30.2713.
  13. ^ a b Del Moral, Pierre (2003). "Particle approximations of Lyapunov exponents connected to Schrödinger operators and Feynman-Kac semigroups". ESAIM Probability & Statistics. 7: 171–208. doi:10.1051/ps:2003001.
  14. ^ Assaraf, Roland; Caffarel, Michel; Khelif, Anatole (2000). "Diffusion Monte Carlo Methods with a fixed number of walkers" (PDF). Phys. Rev. E. 61 (4): 4566–4575. Bibcode:2000PhRvE..61.4566A. doi:10.1103/physreve.61.4566. PMID 11088257. Archived from the original (PDF) on 2014-11-07.
  15. ^ Caffarel, Michel; Ceperley, David; Kalos, Malvin (1993). "Comment on Feynman-Kac Path-Integral Calculation of the Ground-State Energies of Atoms". Phys. Rev. Lett. 71 (13): 2159. Bibcode:1993PhRvL..71.2159C. doi:10.1103/physrevlett.71.2159. PMID 10054598.
  16. ^ Ocone, D. L. (January 1, 1999). "Asymptotic stability of beneš filters". Stochastic Analysis and Applications. 17 (6): 1053–1074. doi:10.1080/07362999908809648. ISSN 0736-2994.
  17. ^ Maurel, Mireille Chaleyat; Michel, Dominique (January 1, 1984). "Des resultats de non existence de filtre de dimension finie". Stochastics. 13 (1–2): 83–102. doi:10.1080/17442508408833312. ISSN 0090-9491.
  18. ^ a b Hajiramezanali, E. & Imani, M. & Braga-Neto, U. & Qian, X & Dougherty, E. R., "Scalable Optimal Bayesian Classification of Single-Cell Trajectories under Regulatory Model Uncertainty", Proceedings of the 2018 ACM International Conference on Bioinformatics, Computational Biology, and Health Informatics, Washington, DC, USA. https://arxiv.org/pdf/1902.03188.pdf
  19. ^ Turing, Alan M. (October 1950). "Computing machinery and intelligence". Mind. LIX (238): 433–460. doi:10.1093/mind/LIX.236.433.
  20. ^ Barricelli, Nils Aall (1954). "Esempi numerici di processi di evoluzione". Methodos: 45–68.
  21. ^ Barricelli, Nils Aall (1957). "Symbiogenetic evolution processes realized by artificial methods". Methodos: 143–182.
  22. ^ Hammersley, J. M.; Morton, K. W. (1954). "Poor Man's Monte Carlo". Journal of the Royal Statistical Society. Series B (Methodological). 16 (1): 23–38. doi:10.1111/j.2517-6161.1954.tb00145.x. JSTOR 2984008.
  23. ^ Barricelli, Nils Aall (1963). "Numerical testing of evolution theories. Part II. Preliminary tests of performance, symbiogenesis and terrestrial life". Acta Biotheoretica. 16 (3–4): 99–126. doi:10.1007/BF01556602.
  24. ^ "Adaptation in Natural and Artificial Systems | The MIT Press". mitpress.mit.edu. Retrieved 2015-06-06.
  25. ^ Fraser, Alex (1957). "Simulation of genetic systems by automatic digital computers. I. Introduction". Aust. J. Biol. Sci. 10 (4): 484–491. doi:10.1071/BI9570484.
  26. ^ Fraser, Alex; Burnell, Donald (1970). Computer Models in Genetics. New York: McGraw-Hill. ISBN 978-0-07-021904-5.
  27. ^ Crosby, Jack L. (1973). Computer Simulation in Genetics. London: John Wiley & Sons. ISBN 978-0-471-18880-3.
  28. ^ Assaraf, Roland; Caffarel, Michel; Khelif, Anatole (2000). "Diffusion Monte Carlo Methods with a fixed number of walkers" (PDF). Phys. Rev. E. 61 (4): 4566–4575. Bibcode:2000PhRvE..61.4566A. doi:10.1103/physreve.61.4566. PMID 11088257. Archived from the original (PDF) on 2014-11-07.
  29. ^ Caffarel, Michel; Ceperley, David; Kalos, Malvin (1993). "Comment on Feynman-Kac Path-Integral Calculation of the Ground-State Energies of Atoms". Phys. Rev. Lett. 71 (13): 2159. Bibcode:1993PhRvL..71.2159C. doi:10.1103/physrevlett.71.2159. PMID 10054598.
  30. ^ Fermi, Enrique; Richtmyer, Robert, D. (1948). "Note on census-taking in Monte Carlo calculations" (PDF). LAM. 805 (A). Declassified report Los Alamos Archive
  31. ^ Herman, Kahn; Harris, Theodore, E. (1951). "Estimation of particle transmission by random sampling" (PDF). Natl. Bur. Stand. Appl. Math. Ser. 12: 27–30.
  32. ^ a b c Kitagawa, G. (January 1993). "A Monte Carlo Filtering and Smoothing Method for Non-Gaussian Nonlinear State Space Models" (PDF). Proceedings of the 2nd U.S.-Japan Joint Seminar on Statistical Time Series Analysis: 110–131.
  33. ^ Kitagawa, G. (1996). "Monte carlo filter and smoother for non-Gaussian nonlinear state space models". Journal of Computational and Graphical Statistics. 5 (1): 1–25. doi:10.2307/1390750. JSTOR 1390750.
  34. ^ a b Gordon, N.J.; Salmond, D.J.; Smith, A.F.M. (April 1993). "Novel approach to nonlinear/non-Gaussian Bayesian state estimation". IEE Proceedings F - Radar and Signal Processing. 140 (2): 107–113. doi:10.1049/ip-f-2.1993.0015. ISSN 0956-375X.
  35. ^ Carvalho, Himilcon; Del Moral, Pierre; Monin, André; Salut, Gérard (July 1997). "Optimal Non-linear Filtering in GPS/INS Integration" (PDF). IEEE Transactions on Aerospace and Electronic Systems. 33 (3): 835. Bibcode:1997ITAES..33..835C. doi:10.1109/7.599254.
  36. ^ P. Del Moral, G. Rigal, and G. Salut. Estimation and nonlinear optimal control : An unified framework for particle solutions
    LAAS-CNRS, Toulouse, Research Report no. 91137, DRET-DIGILOG- LAAS/CNRS contract, April (1991).
  37. ^ P. Del Moral, G. Rigal, and G. Salut. Nonlinear and non Gaussian particle filters applied to inertial platform repositioning.
    LAAS-CNRS, Toulouse, Research Report no. 92207, STCAN/DIGILOG-LAAS/CNRS Convention STCAN no. A.91.77.013, (94p.) September (1991).
  38. ^ P. Del Moral, G. Rigal, and G. Salut. Estimation and nonlinear optimal control : Particle resolution in filtering and estimation. Experimental results.
    Convention DRET no. 89.34.553.00.470.75.01, Research report no.2 (54p.), January (1992).
  39. ^ P. Del Moral, G. Rigal, and G. Salut. Estimation and nonlinear optimal control : Particle resolution in filtering and estimation. Theoretical results
    Convention DRET no. 89.34.553.00.470.75.01, Research report no.3 (123p.), October (1992).
  40. ^ P. Del Moral, J.-Ch. Noyer, G. Rigal, and G. Salut. Particle filters in radar signal processing : detection, estimation and air targets recognition.
    LAAS-CNRS, Toulouse, Research report no. 92495, December (1992).
  41. ^ P. Del Moral, G. Rigal, and G. Salut. Estimation and nonlinear optimal control : Particle resolution in filtering and estimation.
    Studies on: Filtering, optimal control, and maximum likelihood estimation. Convention DRET no. 89.34.553.00.470.75.01. Research report no.4 (210p.), January (1993).
  42. ^ a b Crisan, Dan; Gaines, Jessica; Lyons, Terry (1998). "Convergence of a branching particle method to the solution of the Zakai". SIAM Journal on Applied Mathematics. 58 (5): 1568–1590. doi:10.1137/s0036139996307371.
  43. ^ Crisan, Dan; Lyons, Terry (1997). "Nonlinear filtering and measure-valued processes". Probability Theory and Related Fields. 109 (2): 217–244. doi:10.1007/s004400050131.
  44. ^ Crisan, Dan; Lyons, Terry (1999). "A particle approximation of the solution of the Kushner–Stratonovitch equation". Probability Theory and Related Fields. 115 (4): 549–578. doi:10.1007/s004400050249.
  45. ^ a b c Crisan, Dan; Del Moral, Pierre; Lyons, Terry (1999). "Discrete filtering using branching and interacting particle systems" (PDF). Markov Processes and Related Fields. 5 (3): 293–318.
  46. ^ a b c d Del Moral, Pierre; Guionnet, Alice (1999). "On the stability of Measure Valued Processes with Applications to filtering". C. R. Acad. Sci. Paris. 39 (1): 429–434.
  47. ^ a b c d Del Moral, Pierre; Guionnet, Alice (2001). "On the stability of interacting processes with applications to filtering and genetic algorithms". Annales de l'Institut Henri Poincaré. 37 (2): 155–194. Bibcode:2001AnIHP..37..155D. doi:10.1016/s0246-0203(00)01064-5.
  48. ^ a b Del Moral, P.; Guionnet, A. (1999). "Central limit theorem for nonlinear filtering and interacting particle systems". The Annals of Applied Probability. 9 (2): 275–297. doi:10.1214/aoap/1029962742. ISSN 1050-5164.
  49. ^ a b Del Moral, Pierre; Miclo, Laurent (2001). "Genealogies and Increasing Propagation of Chaos For Feynman-Kac and Genetic Models". The Annals of Applied Probability. 11 (4): 1166–1198. doi:10.1214/aoap/1015345399. ISSN 1050-5164.
  50. ^ a b Doucet, A.; De Freitas, N.; Murphy, K.; Russell, S. (2000). Rao–Blackwellised particle filtering for dynamic Bayesian networks. Proceedings of the Sixteenth conference on Uncertainty in artificial intelligence. pp. 176–183. CiteSeerX 10.1.1.137.5199.
  51. ^ a b Del Moral, Pierre; Miclo, Laurent (2001). "Genealogies and Increasing Propagations of Chaos for Feynman-Kac and Genetic Models". Annals of Applied Probability. 11 (4): 1166–1198.
  52. ^ a b Del Moral, Pierre; Doucet, Arnaud; Singh, Sumeetpal, S. (2010). "A Backward Particle Interpretation of Feynman-Kac Formulae" (PDF). M2AN. 44 (5): 947–976. doi:10.1051/m2an/2010048.
  53. ^ Vergé, Christelle; Dubarry, Cyrille; Del Moral, Pierre; Moulines, Eric (2013). "On parallel implementation of Sequential Monte Carlo methods: the island particle model". Statistics and Computing. 25 (2): 243–260. arXiv:1306.3911. Bibcode:2013arXiv1306.3911V. doi:10.1007/s11222-013-9429-x.
  54. ^ Chopin, Nicolas; Jacob, Pierre, E.; Papaspiliopoulos, Omiros (2011). "SMC^2: an efficient algorithm for sequential analysis of state-space models". arXiv:1101.1528v3 [stat.CO].
  55. ^ Andrieu, Christophe; Doucet, Arnaud; Holenstein, Roman (2010). "Particle Markov chain Monte Carlo methods". Journal of the Royal Statistical Society, Series B. 72 (3): 269–342. doi:10.1111/j.1467-9868.2009.00736.x.
  56. ^ Del Moral, Pierre; Patras, Frédéric; Kohn, Robert (2014). "On Feynman-Kac and particle Markov chain Monte Carlo models". arXiv:1404.5733 [math.PR].
  57. ^ Del Moral, Pierre; Jacod, Jean; Protter, Philip (2001-07-01). "The Monte-Carlo method for filtering with discrete-time observations". Probability Theory and Related Fields. 120 (3): 346–368. doi:10.1007/PL00008786. hdl:1813/9179. ISSN 0178-8051.
  58. ^ Del Moral, Pierre; Doucet, Arnaud; Jasra, Ajay (2011). "An adaptive sequential Monte Carlo method for approximate Bayesian computation". Statistics and Computing. 22 (5): 1009–1020. CiteSeerX 10.1.1.218.9800. doi:10.1007/s11222-011-9271-y. ISSN 0960-3174.
  59. ^ Martin, James S.; Jasra, Ajay; Singh, Sumeetpal S.; Whiteley, Nick; Del Moral, Pierre; McCoy, Emma (May 4, 2014). "Approximate Bayesian Computation for Smoothing". Stochastic Analysis and Applications. 32 (3): 397–420. arXiv:1206.5208. doi:10.1080/07362994.2013.879262. ISSN 0736-2994.
  60. ^ Del Moral, Pierre; Rio, Emmanuel (2011). "Concentration inequalities for mean field particle models". The Annals of Applied Probability. 21 (3): 1017–1052. arXiv:1211.1837. doi:10.1214/10-AAP716. ISSN 1050-5164.
  61. ^ Del Moral, Pierre; Hu, Peng; Wu, Liming (2012). On the Concentration Properties of Interacting Particle Processes. Hanover, MA, USA: Now Publishers Inc. ISBN 978-1601985125.
  62. ^ Zand, G.; Taherkhani, M.; Safabakhsh, R. (2015). "Exponential Natural Particle Filter". arXiv:1511.06603 [cs.LG].
  63. ^ Pitt, M.K.; Shephard, N. (1999). "Filtering Via Simulation: Auxiliary Particle Filters". Journal of the American Statistical Association. 94 (446): 590–591. doi:10.2307/2670179. JSTOR 2670179. Retrieved 2008-05-06.
  64. ^ Liu, J.; Wang, W.; Ma, F. (2011). "A Regularized Auxiliary Particle Filtering Approach for System State Estimation and Battery Life Prediction". Smart Materials and Structures. 20 (7): 1–9. Bibcode:2011SMaS...20g5021L. doi:10.1088/0964-1726/20/7/075021.
  65. ^ Canton-Ferrer, C.; Casas, J.R.; Pardàs, M. (2011). "Human Motion Capture Using Scalable Body Models". Computer Vision and Image Understanding. 115 (10): 1363–1374. doi:10.1016/j.cviu.2011.06.001. hdl:2117/13393.
  66. ^ Blanco, J.L.; Gonzalez, J.; Fernandez-Madrigal, J.A. (2008). An Optimal Filtering Algorithm for Non-Parametric Observation Models in Robot Localization. IEEE International Conference on Robotics and Automation (ICRA'08). pp. 461–466. CiteSeerX 10.1.1.190.7092.
  67. ^ Blanco, J.L.; Gonzalez, J.; Fernandez-Madrigal, J.A. (2010). "Optimal Filtering for Non-Parametric Observation Models: Applications to Localization and SLAM". The International Journal of Robotics Research (IJRR). 29 (14): 1726–1742. CiteSeerX 10.1.1.1031.4931. doi:10.1177/0278364910364165.

BibliographyEdit

External linksEdit