×

Weak semiconvexity estimates for Schrödinger potentials and logarithmic Sobolev inequality for Schrödinger bridges. (English) Zbl 07891478

Summary: We investigate the quadratic Schrödinger bridge problem, a.k.a. Entropic Optimal Transport problem, and obtain weak semiconvexity and semiconcavity bounds on Schrödinger potentials under mild assumptions on the marginals that are substantially weaker than log-concavity. We deduce from these estimates that Schrödinger bridges satisfy a logarithmic Sobolev inequality on the product space. Our proof strategy is based on a second order analysis of coupling by reflection on the characteristics of the Hamilton-Jacobi-Bellman equation that reveals the existence of new classes of invariant functions for the corresponding flow.

MSC:

49Q22 Optimal transportation
49J45 Methods involving semicontinuity and convergence; relaxation
35G50 Systems of nonlinear higher-order PDEs
60J60 Diffusion processes
39B62 Functional inequalities, including subadditivity, convexity, etc.

References:

[1] Aida, S.; Shigekawa, I., Logarithmic Sobolev inequalities and spectral gaps: perturbation theory, J. Funct. Anal., 126, 2, 448-475, 1994 · Zbl 0846.46019 · doi:10.1006/jfan.1994.1154
[2] Bakry, D.; Gentil, Ivan; Ledoux, M., Analysis and geometry of Markov diffusion operators, 2013, Springer Science & Business Media
[3] Bayraktar, E., Eckstein, S. and Zhang, X.: Stability and sample complexity of divergence regularized optimal transport. arXiv preprint arXiv:2212.00367, (2022)
[4] Benamou, J-D, Optimal transportation, modelling and numerical simulation, Acta Numer, 30, 249-325, 2021 · Zbl 1519.65013 · doi:10.1017/S0962492921000040
[5] Benamou, J-D; Carlier, G.; Cuturi, M.; Nenna, L.; Peyré, G., Iterative Bregman projections for regularized transportation problems, SIAM J. Sci. Comput., 37, 2, A1111-A1138, 2015 · Zbl 1319.49073 · doi:10.1137/141000439
[6] Bodineau, T.; Helffer, B., The log-sobolev inequality for unbounded spin systems, J. Funct. Anal., 166, 1, 168-178, 1999 · Zbl 0972.82035 · doi:10.1006/jfan.1999.3419
[7] Brascamp, HJ; Lieb, EH, On extensions of the Brunn-Minkowski and Prékopa-leindler theorems, including inequalities for log concave functions, and with an application to the diffusion equation, J. Funct. Anal., 22, 4, 366-389, 1976 · Zbl 0334.26009 · doi:10.1016/0022-1236(76)90004-5
[8] Caffarelli, LA, Monotonicity properties of optimal transportation and the FKG and related inequalities, Commun. Math. Phys., 214, 3, 547-563, 2000 · Zbl 0978.60107 · doi:10.1007/s002200000257
[9] Chen, Y., Georgiou, T.T. and Pavon, M.: On the relation between optimal transport and Schrödinger bridges: A stochastic control viewpoint. preprint arXiv:1412.4430, (2014)
[10] Chen, Y.; Georgiou, TT; Pavon, M., Stochastic control liaisons: richard Sinkhorn meets gaspard monge on a schrodinger bridge, SIAM Rev., 63, 2, 249-313, 2021 · Zbl 1465.49016 · doi:10.1137/20M1339982
[11] Chewi, S.; Pooladian, AA, An entropic generalization of caffarelli’s contraction theorem via covariance inequalities, Comp. Rendus. Math., 361, G9, 1471-1482, 2023 · Zbl 1534.49041 · doi:10.5802/crmath.486
[12] Chiarini, A.; Conforti, G.; Greco, G.; Tamanini, L., Gradient estimates for the Schrödinger potentials: convergence to the Brenier map and quantitative stability, Commun. Part. Diff. Equ., 48, 6, 895-943, 2023 · Zbl 1525.49033 · doi:10.1080/03605302.2023.2215527
[13] Conforti, G., A second order equation for Schrödinger bridges with applications to the hot gas experiment and entropic transportation cost, Probab. Theory Relat. Fields, 174, 1-2, 1-47, 2019 · Zbl 1481.60152 · doi:10.1007/s00440-018-0856-7
[14] Conforti, G., Coupling by reflection for controlled diffusion processes: turnpike property and large time behavior of Hamilton Jacobi Bellman equations, Ann. Appl. Probab., 33, 6, 4608-4644, 2022 · Zbl 1538.60095
[15] Cuturi., M.: Sinkhorn distances: Lightspeed computation of optimal transport. In Advances in Neural Information Processing Systems, pp. 2292-2300, (2013)
[16] Dai Pra, P.; Pavon, M., On the Markov processes of Schrödinger, the Feynman-Kac formula and stochastic control, Real. Modell. Syst. Theory, 3, 497-504, 1990 · Zbl 0726.60066
[17] Heng, J., De Bortoli, V., Doucet, A. and Thornton, J.: Simulating diffusion bridges with score matching. arXiv preprint arXiv:2111.07243, (2021)
[18] De Bortoli, V., Thornton, J., Heng, J. and Doucet, A.: Diffusion schrödinger bridge with applications to score-based generative modeling. Advances in Neural Information Processing Systems, 34, (2021)
[19] Deligiannidis, G., De Bortoli, V. and Doucet, A.: Quantitative uniform stability of the iterative proportional fitting procedure. arXiv preprint arXiv:2108.08129, (2021)
[20] Djellout, H.; Guillin, A.; Liming, W., Transportation cost-information inequalities and applications to random dynamical systems and diffusions, Ann. Probab., 32, 3, 2702-2732, 2004 · Zbl 1061.60011 · doi:10.1214/009117904000000531
[21] Doob, J., Conditional Brownian motion and the boundary limits of harmonic functions, Bull. Soc. Math. France, 85, 431-458, 1957 · Zbl 0097.34004 · doi:10.24033/bsmf.1494
[22] Eberle, A., Reflection couplings and contraction rates for diffusions, Probab. Theory Relat. Fields, 166, 3-4, 851-886, 2016 · Zbl 1367.60099 · doi:10.1007/s00440-015-0673-1
[23] Eckstein, S. and Nutz, M.: Quantitative stability of regularized optimal transport and convergence of Sinkhorn’s algorithm. arXiv preprint arXiv:2110.06798, (2021) · Zbl 1507.90127
[24] Fathi, M.; Gozlan, N.; Prod’homme, M., A proof of the Caffarelli contraction theorem via entropic regularization, Calculus Variat. Partial Differ. Equ., 59, 96, 1-18, 2020 · Zbl 1458.49038
[25] Gentil, I.; Léonard, C.; Ripani, L., Dynamical aspects of the generalized schrödinger problem via otto calculus-a heuristic point of view, Revista Matemática Iberoamericana, 36, 4, 1071-1112, 2020 · Zbl 1462.37009 · doi:10.4171/rmi/1159
[26] Ghosal, P. and Nutz, M.: On the convergence rate of sinkhorn’s algorithm. arXiv preprint arXiv:2212.06000, (2022)
[27] Grunewald, N., Otto, F., Villani, C. and Westdickenberg, M.G.: A two-scale approach to logarithmic sobolev inequalities and the hydrodynamic limit. In Annales de l’IHP Probabilités et statistiques, vol. 45, pp. 302-351, (2009) · Zbl 1179.60068
[28] Holley, R. and Stroock, D.W.: Logarithmic Sobolev inequalities and stochastic Ising models. (1986)
[29] Léonard, C.: Girsanov theory under a finite entropy condition. In C. Donati-Martin, A. Lejay, and A. Rouault, editors, Séminaire de Probabilités XLIV, volume 2046 of Lecture Notes in Mathematics, pp. 429-465. Springer, (2012) · Zbl 1253.60051
[30] Léonard, C.: Some properties of path measures. In Séminaire de Probabilités XLVI, pages 207-230. Springer, (2014) · Zbl 1390.60279
[31] Léonard, C., A survey of the Schrödinger problem and some of its connections with optimal transport, Discrete Contin. Dynam. Systems, 34, 4, 1533-1574, 2014 · Zbl 1277.49052 · doi:10.3934/dcds.2014.34.1533
[32] Lindvall, T.; Rogers, LCG, Coupling of multidimensional diffusions by reflection, Annal. Prob., 1, 860-872, 1986 · Zbl 0593.60076
[33] Mikulincer, D. and Shenfeld, Y.: The Brownian transport map. arXiv preprint arXiv:2111.11521, (2021)
[34] Mikulincer, D. and Shenfeld, Y.: On the Lipschitz properties of transportation along heat flows. arXiv preprint arXiv:2201.01382, 2022
[35] Nutz, M.; Wiesel, J., Entropic optimal transport: convergence of potentials, Probab. Theory Relat. Fields, 184, 1, 401-424, 2022 · Zbl 1504.90095 · doi:10.1007/s00440-021-01096-8
[36] Peng, S., A general stochastic maximum principle for optimal control problems, SIAM J. Control. Optim., 28, 4, 966-979, 1990 · Zbl 0712.93067 · doi:10.1137/0328054
[37] Peyré, G.; Cuturi, M., Computational optimal transport, Found. Trends Mach. Learn., 11, 5-6, 355-607, 2019 · Zbl 1475.68011 · doi:10.1561/2200000073
[38] Schrödinger, Erwin, Über die Umkehrung der Naturgesetze. Sitzungsberichte Preuss, Akad. Wiss. Berlin. Phys. Math., 144, 144-153, 1931 · JFM 57.1147.01
[39] Shi, Y., De Bortoli, V., Deligiannidis, G. and Doucet, A.: Conditional simulation using diffusion schrödinger bridges. arXiv preprint arXiv:2202.13460, (2022)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.