2021

Schr{\"o}dinger-F{\"o}llmer Sampler: Sampling without Ergodicity

Huang, Jian, Jiao, Yuling, Kang, Lican et al.

Understand

Sampling from probability distributions is an important problem in statistics and machine learning, specially in Bayesian inference when integration with respect to posterior distribution is intractable and sampling from the posterior is the only viable option for inference.

In this paper, we propose Schr\"{o}dinger-F\"{o}llmer sampler (SFS), a novel approach for sampling from possibly unnormalized distributions.
The proposed SFS is based on the Schr\"{o}dinger-F\"{o}llmer diffusion process on the unit interval with a time dependent drift term, which transports the degenerate distribution at time zero to the target distribution at time one.
Comparing with the existing Markov chain Monte Carlo samplers that require ergodicity, no such requirement is needed for SFS.

Built on

On stochastic gradient Langevin dynamics with dependent data streams: the fully non-convex case
Original
Chau, N. H., Moulines, É., Rásonyi, M., Sabanis, S., and Zhang, Y. (2019) · 1905
Earlier work this paper cites.
Improved bounds for discretization of Langevin diffusions: Near-optimal rates without convexity
Original
Mou, W., Flammarion, N., Wainwright, M. J., and Bartlett, P. L. (2019) · 1907
Earlier work this paper cites.
Nonasymptotic estimates for stochastic gradient Langevin dynamics under local conditions in nonconvex optimization
Original
Zhang, Y., Akyildiz, Ö. D., Damoulas, T., and Sabanis, S. (2019b) · 1910
Earlier work this paper cites.
Sur la théorie relativiste de l’électron et l’interprétation de la mécanique quantique
Schrödinger, E. (1932) · 1932
Earlier work this paper cites.
Equation of state calculations by fast computing machines
Metropolis, N., Rosenbluth, A. W., Rosenbluth, M. N., Teller, A. H., and Teller, E. (1953) · 1953
Earlier work this paper cites.
A relationship between arbitrary positive matrices and doubly stochastic matrices
Sinkhorn, R. (1964) · 1964
Earlier work this paper cites.
Cluster analysis of multivariate data: efficiency versus interpretability of classifications
Forgy, E. W. (1965) · 1965
Earlier work this paper cites.
Some methods for classification and analysis of multivariate observations
MacQueen, J. et al. (1967) · 1967
Earlier work this paper cites.
Monte Carlo sampling methods using Markov chains and their applications
Hastings, W. K. (1970) · 1970
Earlier work this paper cites.
A bound for the error in the normal approximation to the distribution of a sum of dependent random variables
Stein, C. (1972) · 1972
Earlier work this paper cites.
The Markov processes of Schrödinger
Jamison, B. (1975) · 1975
Earlier work this paper cites.
Generalized residuals for log-linear models
Haberman, S. J. (1976) · 1976
Earlier work this paper cites.
Least squares quantization in PCM
Lloyd, S. (1982) · 1982
Earlier work this paper cites.
Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images
Geman, S. and Geman, D. (1984) · 1984
Earlier work this paper cites.
An entropy approach to the time reversal of diffusion processes
Föllmer, H. (1985) · 1985
Earlier work this paper cites.
Random fields and diffusion processes
Föllmer, H. (1988) · 1985
Earlier work this paper cites.
Time reversal on Wiener space
Föllmer, H. (1986) · 1986
Earlier work this paper cites.
Approximate Computations of Expectations
Stein, C. (1986) · 1986
Earlier work this paper cites.
Hybrid Monte Carlo
Duane, S., Kennedy, A. D., Pendleton, B. J., and Roweth, D. (1987) · 1987
Earlier work this paper cites.
Using the adap learning algorithm to forecast the onset of diabetes mellitus
Smith, J. W., Everhart, J. E., Dickson, W., Knowler, W. C., and Johannes, R. S. (1988) · 1988
Earlier work this paper cites.
Sampling-based approaches to calculating marginal densities
Gelfand, A. E. and Smith, A. F. M. (1990) · 1990
Earlier work this paper cites.
A stochastic control approach to reciprocal diffusion processes
Dai Pra, P. (1991) · 1991
Earlier work this paper cites.
Stochastic differential equations
Kloeden, P. E. and Platen, E. (1992) · 1992
Earlier work this paper cites.
Markov chains for exploring posterior distributions
Tierney, L. (1994) · 1994
Earlier work this paper cites.
Understanding the Metropolis-Hastings algorithm
Chib, S. and Greenberg, E. (1995) · 1995
Earlier work this paper cites.
Exponential convergence of Langevin distributions and their discrete approximations
Roberts, G. O., Tweedie, R. L., et al. (1996) · 1996
Earlier work this paper cites.

Similar

The Metropolis-Hastings algorithm
Robert, C. P. and Casella, G. (1999) · 1999
Cited alongside, same era.
An adaptive Metropolis algorithm
Haario, H., Saksman, E., Tamminen, J., et al. (2001) · 2001
Cited alongside, same era.
Computing bayes: Bayesian computation from 1763 to the 21st century
Original
Martin, G. M., Frazier, D. T., and Robert, C. P. (2020) · 2004
Cited alongside, same era.
Analysis of the sagittal balance of the spine and pelvis using shape and orientation parameters
Berthonnaud, E., Dimnet, J., Roussouly, P., and Labelle, H. (2005) · 2005
Cited alongside, same era.
Stochastic control liasons: Richard Sinkhorn meets Gaspard Monge on a Schrödinger bridge
Original
Chen, Y., Georgiou, T. T., and Pavon, M. (2020) · 2005
Cited alongside, same era.
Rate of convergence for ergodic continuous Markov processes: Lyapunov versus Poincaré
Bakry, D., Cattiaux, P., and Guillin, A. (2008) · 2008
Cited alongside, same era.
Stein’s lemma for elliptical random vectors
Landsman, Z. and Nešlehová, J. (2008) · 2008
Cited alongside, same era.
Trends to equilibrium in total variation distance
Cattiaux, P. and Guillin, A. (2009) · 2009
Cited alongside, same era.
Simulation and inference for stochastic differential equations: with R examples
Iacus, S. M. (2009) · 2009
Cited alongside, same era.
Log-Sobolev inequalities: different roles of Ric and Hess
Wang, F.-Y. et al. (2009) · 2009
Cited alongside, same era.
Asymptotic behavior of dissipative systems
Hale, J. K. (2010) · 2010
Cited alongside, same era.
Handbook of markov chain monte carlo
Brooks, S., Gelman, A., Jones, G., and Meng, X.-L. (2011) · 2011
Cited alongside, same era.

Then

A conceptual introduction to Hamiltonian Monte Carlo
Original
Betancourt, M. (2017) · 2017
Later among the works it cites.
UCI machine learning repository
Dua, D. and Graff, C. (2017) · 2017
Later among the works it cites.
Nonasymptotic convergence analysis for the unadjusted Langevin algorithm
Durmus, A., Moulines, E., et al. (2017) · 2017
Later among the works it cites.
Non-convex learning via stochastic gradient Langevin dynamics: a nonasymptotic analysis
Raginsky, M., Rakhlin, A., and Telgarsky, M. (2017) · 2017
Later among the works it cites.
The bouncy particle sampler: A nonreversible rejection-free markov chain monte carlo method
Bouchard-Côté, A., Vollmer, S. J., and Doucet, A. (2018) · 2018
Later among the works it cites.
Convergence of Langevin MCMC in KL-divergence
Cheng, X. and Bartlett, P. (2018) · 2018
Later among the works it cites.
Sharp convergence rates for Langevin dynamics in the nonconvex setting
Original
Cheng, X., Chatterji, N. S., Abbasi-Yadkori, Y., Bartlett, P. L., and Jordan, M. I. (2018) · 2018
Later among the works it cites.
Simulation and Inference for Stochastic Processes with YUIMA
Iacus, S. M. and Yoshida, N. (2018) · 2018
Later among the works it cites.
Does Hamiltonian monte carlo mix faster than a random walk on multimodal densities?
Original
Mangoubi, O., Pillai, N. S., and Smith, A. (2018) · 2018
Later among the works it cites.
Schrödinger bridge samplers
Bernton, E., Heng, J., Doucet, A., and Jacob, P. E. (2019) · 2019
Later among the works it cites.
The zig-zag process and super-efficient sampling for Bayesian analysis of big data
Bierkens, J., Fearnhead, P., Roberts, G., et al. (2019) · 2019
Later among the works it cites.
Probabilistic inference on noisy time series (pints)
Clerx, M., Robinson, M., Lambert, B., Lei, C. L., Ghosh, S., Mirams, G. R., and Gavaghan, D. J. (2019) · 2019
Later among the works it cites.
User-friendly guarantees for the Langevin Monte Carlo with inaccurate gradient
Dalalyan, A. S. and Karagulyan, A. G. (2019) · 2019
Later among the works it cites.
High-dimensional Bayesian inference via the unadjusted Langevin algorithm
Durmus, A., Moulines, E., et al. (2019) · 2019
Later among the works it cites.
Applied stochastic analysis
E, W., Li, T., and Vanden-Eijnden, E. (2019) · 2019
Later among the works it cites.
Sampling can be faster than optimization
Ma, Y.-A., Chen, Y., Jin, C., Flammarion, N., and Jordan, M. I. (2019) · 2019
Later among the works it cites.
Theoretical guarantees for sampling and inference in generative models with latent diffusions
Tzen, B. and Raginsky, M. (2019) · 2019
Later among the works it cites.
Coupling and convergence for Hamiltonian Monte Carlo
Bou-Rabee, N., Eberle, A., Zimmer, R., et al. (2020) · 2020
Later among the works it cites.
Markov chain monte carlo algorithms for bayesian computation, a survey and some generalisation
Changye, W. and Robert, C. P. (2020) · 2020
Later among the works it cites.
The Hastings algorithm at fifty
Dunson, D. B. and Johndrow, J. (2020) · 2020
Later among the works it cites.
Stability of the logarithmic Sobolev inequality via the Föllmer process
Eldan, R., Lehec, J., Shenfeld, Y., et al. (2020) · 2020
Later among the works it cites.
Computational optimal transport
Peyre, G. and Cuturi, M. (2020) · 2020
Later among the works it cites.
On stochastic gradient Langevin dynamics with dependent data streams in the logconcave case
Barkhagen, M., Chau, N. H., Moulines, É., Rásonyi, M., Sabanis, S., and Zhang, Y. (2021) · 2021
Closest in time.
Diffusion schrodinger bridge with applications to score-based generative modeling
De Bortoli, V., Thornton, J., Heng, J., and Doucet, A. (2021) · 2021
Closest in time.
The data-driven schroedinger bridge
Pavon, M., Tabak, E. G., and Trigila, G. (2021) · 2021
Closest in time.
Deep generative learning via schrodinger bridge
Wang, G., Jiao, Y., Xu, Q., Wang, Y., and Yang, C. (2021) · 2021
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…

Reading the bibliography…

Schr{\&#34;o}dinger-F{\&#34;o}llmer Sampler: Sampling without Ergodicity

Understand

Built on

Similar

Then

Beyond the bibliography

Built on

Similar

Then

Beyond the bibliography

Schr{\"o}dinger-F{\"o}llmer Sampler: Sampling without Ergodicity