Ricci curvature and $W_1$-exponential convergence of Markov processes on graphs

Liming Wu; Lingyan Cheng; Ruinan Li

arxiv: 1907.11036 · v1 · pith:DZ77FY5Enew · submitted 2019-07-25 · 🧮 math.PR

Ricci curvature and W₁-exponential convergence of Markov processes on graphs

Lingyan Cheng , Ruinan Li , Liming Wu This is my paper

Pith reviewed 2026-05-24 16:14 UTC · model grok-4.3

classification 🧮 math.PR

keywords Ricci curvatureMarkov processesgraphsWasserstein distanceexponential convergencecoupling generatorGlauber dynamicsdeath-birth process

0 comments

The pith

The lower bound on Ollivier Ricci curvature for continuous-time Markov jump processes on graphs is equivalent to the existence of an optimal coupling generator.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper shows that a lower bound on Ricci curvature, measured in the Wasserstein-1 sense of Ollivier, for a continuous-time jumping Markov process on a graph is exactly equivalent to the existence of an optimal coupling generator, and supplies an explicit construction of that generator. The work extends earlier discrete-time results to the continuous-time setting. Explicit exponential convergence rates to equilibrium follow from a comparison with a death-birth process on the natural numbers after a suitable change of metric. A counterpart of the Zhong-Yang estimate is proved when curvature is nonnegative, and the Lyapunov-function approach yields quantitative rates when curvature is bounded below by a negative constant. The results are applied to Glauber dynamics under dynamical versions of the Dobrushin uniqueness and analyticity conditions.

Core claim

The Ricci curvature lower bound in Ollivier's Wasserstein metric sense of a continuous time jumping Markov process on a graph can be characterized by some optimal coupling generator and we provide the construction of this latter. Previous results of Ollivier for discrete time Markov chains are generalized to the continuous time case. A comparison technique with some death-birth process on the natural numbers yields explicit exponential convergence rates after modifying the metric. A counterpart of Zhong-Yang's estimate holds when the Ricci curvature with respect to the graph metric is nonnegative. The Lyapunov function method works with explicit quantitative estimates once the Ricci curvatue

What carries the argument

The optimal coupling generator: a Markov generator on the product space whose marginals recover the original processes and whose action realizes the infimum that defines the Wasserstein-1 distance between measures.

If this is right

Explicit exponential convergence rates to equilibrium are obtained by comparing the process to a death-birth chain on the naturals after a metric change.
A Zhong-Yang-type diameter bound on the spectral gap holds whenever the curvature lower bound is nonnegative.
Quantitative exponential convergence follows from a Lyapunov-function argument once curvature is bounded below by any negative constant.
The same curvature bound controls mixing for Glauber dynamics under dynamical Dobrushin uniqueness or analyticity conditions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The explicit construction of the coupling generator may turn curvature verification into a finite linear-programming problem on small graphs.
The metric-modification step suggests that curvature bounds can be improved by rescaling edge lengths without changing the underlying process.
The continuous-time formulation opens a direct comparison route to diffusion processes on manifolds where Ricci curvature already controls Wasserstein convergence.

Load-bearing premise

The object is a continuous-time jumping Markov process on a finite or locally finite graph whose transitions are measured with the graph distance and whose Wasserstein-1 distance uses that same metric.

What would settle it

Exhibit a concrete finite graph, a continuous-time jump process on it, and a pair of measures whose Wasserstein-1 distance cannot be recovered by any coupling whose generator satisfies the curvature inequality implied by the Ollivier definition.

read the original abstract

In this paper, we show that the Ricci curvature lower bound in Ollivier's Wasserstein metric sense of a continuous time jumping Markov process on a graph can be characterized by some optimal coupling generator and provide the construction of this latter. Some previous results of Ollivier for discrete time Markov chains are generalized to the actual continuous time case. We propose a comparison technique with some death-birth process on $\mathbb N$ to obtain some explicit exponential convergence rate, by modifying the metric. A counterpart of Zhong-Yang's estimate is established in the case where the Ricci curvature with repsect to the graph metric is nonnegative. Moreover we show that the Lyapunov function method for the exponential convergence works with some explicit quantitative estimates, once if the Ricci curvature is bounded from below by a negative constant. Finally we present applications to Glauder dynamics under some dynamical versions of the Dobrushin uniqueness condition or of the Dobrushin-Shlosman analyticity condition.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper cleanly generalizes Ollivier's discrete Ricci curvature to continuous time on graphs with an explicit coupling generator and birth-death comparison for rates.

read the letter

Your colleague should know that this manuscript provides a continuous-time analogue of Ollivier's results on Ricci curvature bounds for Markov chains, using a constructed optimal coupling generator, and derives explicit W1 exponential convergence via comparison to a death-birth process on the naturals. The new elements are the generator construction for the continuous case and the modified metric for the comparison, which allows explicit rates. They also establish a counterpart to the Zhong-Yang estimate for nonnegative curvature and show how Lyapunov methods give quantitative bounds when curvature is negative. The applications section to Glauber dynamics under Dobrushin conditions ties it to statistical physics. The paper relies on standard properties of Wasserstein distance and generators on locally finite graphs. The stress-test confirms internal consistency without hidden assumptions like reversibility. The central claims appear to hold up based on the described steps. Soft spots are minor: the optimality of the coupling needs careful verification in the proofs, and the modification of the metric might require checking how it affects the curvature bound in practice. Nothing looks like a deal-breaker. This is for specialists in probability on graphs and discrete curvature. It is a useful technical extension that builds on prior work without reinventing the wheel. A serious referee should see it to validate the constructions and estimates. I recommend sending it for peer review.

Referee Report

2 major / 3 minor

Summary. The manuscript claims that the lower bound on Ollivier-Ricci curvature (in the W1 sense) for a continuous-time jumping Markov process on a (finite or locally finite) graph admits an explicit characterization via an optimal coupling generator, for which a construction is supplied. This generalizes Ollivier's earlier discrete-time results. A comparison argument with a death-birth process on N is used (after modifying the metric) to derive explicit exponential convergence rates in W1. A Zhong-Yang-type estimate is proved under nonnegative curvature with respect to the graph metric, and quantitative Lyapunov-function bounds are obtained when the curvature is bounded below by a negative constant. Applications to Glauber dynamics under dynamical Dobrushin uniqueness or Dobrushin-Shlosman conditions are given.

Significance. If the central claims hold, the work supplies a concrete bridge from the discrete-time Ollivier framework to continuous-time generators on graphs, together with explicit comparison and Lyapunov tools that yield quantitative W1-exponential convergence rates. The absence of extra assumptions (reversibility, bounded degree, jump-rate regularity) beyond the stated setup strengthens the applicability to sampling and statistical-mechanics models.

major comments (2)

[Section containing the generator construction (immediately following the statement of the main curvature theorem)] The construction of the optimal coupling generator (the central object used to characterize the curvature lower bound) must be shown to attain the infimum appearing in the definition of the curvature via the evolution of W1; without an explicit verification that the generator produces the claimed contraction rate, the generalization from the discrete-time case remains formal.
[Section on the comparison technique with death-birth processes] The comparison argument with the death-birth process on N (used to obtain the explicit rate after metric modification) relies on a specific domination of the jump rates; the paper should state the precise inequality between the original generator and the birth-death rates that justifies the comparison (e.g., the inequality used in the proof of the rate).

minor comments (3)

[Abstract] Abstract: 'repsect' should read 'respect'.
[Abstract] Abstract: 'Glauder dynamics' should read 'Glauber dynamics'.
[Section on metric modification] Notation for the modified metric used in the comparison argument should be introduced once and used consistently; currently the same symbol appears to denote both the original graph distance and the modified distance in different paragraphs.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading, positive evaluation, and constructive suggestions. We address the two major comments below and will incorporate the requested clarifications into the revised manuscript.

read point-by-point responses

Referee: The construction of the optimal coupling generator (the central object used to characterize the curvature lower bound) must be shown to attain the infimum appearing in the definition of the curvature via the evolution of W1; without an explicit verification that the generator produces the claimed contraction rate, the generalization from the discrete-time case remains formal.

Authors: We agree that an explicit verification is required for rigor. The manuscript already supplies the construction immediately after the main curvature theorem. In the revision we will add a short lemma that directly computes the evolution of W1 under this generator and confirms that it attains the infimum in the curvature definition, thereby establishing the contraction rate and completing the passage from the discrete-time setting. revision: yes
Referee: The comparison argument with the death-birth process on N (used to obtain the explicit rate after metric modification) relies on a specific domination of the jump rates; the paper should state the precise inequality between the original generator and the birth-death rates that justifies the comparison (e.g., the inequality used in the proof of the rate).

Authors: We accept the suggestion. The comparison relies on a domination of jump rates after the metric change, but the precise inequality is only implicit in the current proof. We will insert an explicit statement of this inequality (relating the original generator to the birth-death rates on N) at the beginning of the comparison section. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The derivation begins from the standard definition of Ollivier-Ricci curvature via W1 distance on the graph metric for a continuous-time jumping Markov process, then explicitly constructs the optimal coupling generator as a new object extending Ollivier's discrete-time framework. Subsequent comparison arguments with death-birth processes on N, Zhong-Yang estimates under nonnegative curvature, and Lyapunov bounds under negative curvature all proceed from this generator construction using standard semigroup and Wasserstein theory on locally finite graphs; none of these steps reduce by definition or self-citation to the target curvature bound itself. The cited prior work is external (Ollivier) and the constructions are presented as independent rather than tautological rewritings.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The work rests on standard properties of Wasserstein distance, optimal couplings, and generator theory for continuous-time Markov processes; no new free parameters or invented entities are introduced in the abstract.

axioms (2)

standard math Existence and properties of optimal couplings for Markov jump processes with respect to the graph distance
Invoked when characterizing the curvature bound via the optimal coupling generator (abstract).
domain assumption Comparison principle between the original process and a one-dimensional death-birth process after metric modification
Used to obtain explicit exponential rates (abstract).

pith-pipeline@v0.9.0 · 5693 in / 1321 out tokens · 23392 ms · 2026-05-24T16:14:25.785556+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

36 extracted references · 36 canonical work pages · 1 internal anchor

[1]

and Guillin, A

Bakry, D., Barthe, F., Cattiaux, P. and Guillin, A. (2008). A simple p roof of the Poincar´ e inequality for a large class of probability measures including the log-co ncave case. Electron. Commun. Probab. 13 60–66

work page 2008
[2]

Bakry, Z.M

D. Bakry, Z.M. Qian. (2000). Some new results on eigenvectors v ia dimension, diameter and Ricci curvature. Adv. Math. 155 98–153

work page 2000
[3]

Barlow, T

M. Barlow, T. Coulhon and A. Grigor’yan. (2001). Manifolds and gr aphs with slow heat kernel decay. Inventions Mathematicae 144 609–649

work page 2001
[4]

Caputo, P

P. Caputo, P. Dai Pra and G. Posta. (2009). Convex entropy d ecay via the Bochner-Bakry- Emery approach. Ann. Inst. H. Poincar´ e45 734–753

work page 2009
[5]

G. Y. Chen and Y. C. Sheu. (2003). On the log-Sobolev constant for the simple random walk on the n-cycle: the even cases. J. Funct. Anal . 202 473–485

work page 2003
[6]

M.F. Chen. (1992). From Markov chains to non-equilibrium partcile systems. World Scientiﬁc, Singapore-New Jersey

work page 1992
[7]

M. F. Chen. (2005). Eigenvalues, inequalities and ergodic theory . Springer

work page 2005
[8]

Chen, F.Y

M.F. Chen, F.Y. Wang. (1994). Application of coupling method to th e ﬁrst eigenvalue on manifold. Sci. China Ser. A (English edition) 37 1-14

work page 1994
[9]

Chen and F.Y

M.F. Chen and F.Y. Wang. (1997). General formula for lower boun d of the ﬁrst eigenvalue on Riemannian manifold. Science in China, Series A 40 384–394

work page 1997
[10]

Fan R. K. Chung. (1997). Spectral graph theory. CBMS Regional Conference Series in Math- ematics, Number 92, Washington, American Mathematical Society, Providence, RI

work page 1997
[11]

Dai Pra, A.M

P. Dai Pra, A.M. Paganoni and G. Posta. (2002). Entropy ineq ualities for unbounded spin systems. Ann. of Probab. 30 1959–1976

work page 2002
[12]

Diaconis and D

P. Diaconis and D. Stroock. (1991). Geometric bounds for eige nvalues of Markov chains. Ann. App. Probab. 1 36–62

work page 1991
[13]

R. L. Dobrushin. (1968). The description of a random ﬁeld by me ans of conditional proba- bilities and condition of its regularity. Theory Probab. Appl. 13 197–224

work page 1968
[14]

R. L. Dobrushin and S. Shlosman. (1985). Completely analytical Gibbs ﬁelds. Statistical physics and dynamical systems 371-403

work page 1985
[15]

A. Eberle. (2016). Reﬂection couplings and contraction rates for diﬀusions. Probab. Theory Relat. Fields. 164 851–886

work page 2016
[16]

Guillin, C

A. Guillin, C. L´ eonard, L. M. Wu and N. Yao. (2009). Transportation-information inequalities for Markov processes. Probab. Theory Relat. Fields 144 669–695

work page 2009
[17]

Transportation-information inequalities for Markov processes (II) : relations with other functional inequalities

A. Guillin, C. L´ eonard, F.Y. Wang and L. M. Wu. Transportation- information inequal- ities for Markov processes (II): relations with other functional in equalities. Available on arXiv:0902.2101

work page internal anchor Pith review Pith/arXiv arXiv
[18]

Guillin, A

A. Guillin, A. Joulin, C. L´ eonard and L. M. Wu, Transportation-in formation inequalities for Markov processes (III). Processes with jumps. Preprint 2008

work page 2008
[19]

Hairer and J

M. Hairer and J. C. Mattingly. (2011). Yet another look at Harr is’ ergodic theorem for Markov chains. Seminar on Stochastic Analysis, Random Fields and Applicat ions VI 109–117

work page 2011
[20]

Joulin and Y

A. Joulin and Y. Ollivier. (2010). Curvature, concentration and error estimates for Markov chain Monte Carlo. Ann. Probab. 38 2418–2442

work page 2010
[21]

G. F. Lawler and A. D. Sokal. (1988). Bounds on the L2 spectrum for Markov chains and Markov processes: a generalization of Cheeger’s inequalities. Trans. Amer. Math. Soc. 309 557–580

work page 1988
[22]

Lin and S.T

Y. Lin and S.T. Yau. (2010). Ricci curvature and eigenvalue est imate on locally ﬁnite graphs. Math. Res. Lett. 17 343–356

work page 2010
[23]

Y. Lin, L. Y. Lu, S-T Yau. (2011). Ricci curvature of graphs. Tohoku Math. J. (2) 63 605–627

work page 2011
[24]

Liu and Y

W. Liu and Y. T. Ma. (2009). Spectral gap and convex concent ration inequalities for birth- death processes. Ann.Inst. H. Poincar´ e Probab. Statist.45 58–69

work page 2009
[25]

Liu, Y.T

W. Liu, Y.T. Ma and L.M. Wu. (2016). Spectral gap, isoperimetry and concentration on trees. Sciences in China, Mathematics 59 539–556. RICCI CUR V ATURE AND W1-EXPONENTIAL CONVERGENCE 51

work page 2016
[26]

Luo and J

D. Luo and J. Wang. (2016). Exponential convergence in Wass erstein distance for diﬀusion processes without uniform dissipativity. Math. Nachr. 289 1909–1926

work page 2016
[27]

Y.T. Ma, R. Wang and L.M. Wu. (2016). Log-Sobolev, isoperimetr y and transport inequali- ties on graphs. Acta. Math. Sinica 32 1221–1236

work page 2016
[28]

Lectures on Probability Theory and Statistics

Martinelli, F., Lectures on Glauber dynamics for discrete spin models. Lectures on Probability Theory and Statistics. Springer Berlin Heidelberg 1999

work page 1999
[29]

Meyn and R.L

S.P. Meyn and R.L. Tweedie. (1993). Markov chains and stochas tic stability. Communica- tions and Control Engineering Series. Springer-Verlag

work page 1993
[30]

Ollivier

Y. Ollivier. (2009). Ricci curvature of Markov chains on metric s paces. J. Funct. Anal. 256 810–864

work page 2009
[31]

von Renesse and K.T

M.K. von Renesse and K.T. Sturm. (2010). Transport inequalitie s, gradient estimates, en- tropy, and Ricci curvature. Communications on Pure and Applied Mathematics 58 923–940

work page 2010
[32]

Saloﬀ-Coste

L. Saloﬀ-Coste. (1997). Lectures on ﬁnite Markov chains. Lecture Notes in Mathematics 1665, 301–413

work page 1997
[33]

Sammer, P

M. Sammer, P. Tetali. (2009). Concentration on the discrete t orus using transportation. Combin. Probab. Comput. 18 835–860

work page 2009
[34]

L. Wu. (2004). Essential spectral radius for Markov semigro ups (I): discrete time case. Probab. Th. Rel. Fields 128 255–321

work page 2004
[35]

L. Wu. (2006). Poincar´ e and transportation inequalities for G ibbs measures under the Do- brushin uniqueness condition. Ann. Probab. 34 1960–1989

work page 2006
[36]

J. Q. Zhong and H. C. Yang. (1984). On the estimate of the ﬁrs t eigenvalue of a compact Riemannian manifold. Sci. Sinica Ser. A 27 1265–1273. Lingyan Cheng. Center for Applied Mathematics, Tianjin Uni versity, Tianjin 300072, PR China. E-mail address : chengly@amss.ac.cn Ruinan Li. School of Statistics and Information, Shanghai U niversity of Inter- nati...

work page 1984

[1] [1]

and Guillin, A

Bakry, D., Barthe, F., Cattiaux, P. and Guillin, A. (2008). A simple p roof of the Poincar´ e inequality for a large class of probability measures including the log-co ncave case. Electron. Commun. Probab. 13 60–66

work page 2008

[2] [2]

Bakry, Z.M

D. Bakry, Z.M. Qian. (2000). Some new results on eigenvectors v ia dimension, diameter and Ricci curvature. Adv. Math. 155 98–153

work page 2000

[3] [3]

Barlow, T

M. Barlow, T. Coulhon and A. Grigor’yan. (2001). Manifolds and gr aphs with slow heat kernel decay. Inventions Mathematicae 144 609–649

work page 2001

[4] [4]

Caputo, P

P. Caputo, P. Dai Pra and G. Posta. (2009). Convex entropy d ecay via the Bochner-Bakry- Emery approach. Ann. Inst. H. Poincar´ e45 734–753

work page 2009

[5] [5]

G. Y. Chen and Y. C. Sheu. (2003). On the log-Sobolev constant for the simple random walk on the n-cycle: the even cases. J. Funct. Anal . 202 473–485

work page 2003

[6] [6]

M.F. Chen. (1992). From Markov chains to non-equilibrium partcile systems. World Scientiﬁc, Singapore-New Jersey

work page 1992

[7] [7]

M. F. Chen. (2005). Eigenvalues, inequalities and ergodic theory . Springer

work page 2005

[8] [8]

Chen, F.Y

M.F. Chen, F.Y. Wang. (1994). Application of coupling method to th e ﬁrst eigenvalue on manifold. Sci. China Ser. A (English edition) 37 1-14

work page 1994

[9] [9]

Chen and F.Y

M.F. Chen and F.Y. Wang. (1997). General formula for lower boun d of the ﬁrst eigenvalue on Riemannian manifold. Science in China, Series A 40 384–394

work page 1997

[10] [10]

Fan R. K. Chung. (1997). Spectral graph theory. CBMS Regional Conference Series in Math- ematics, Number 92, Washington, American Mathematical Society, Providence, RI

work page 1997

[11] [11]

Dai Pra, A.M

P. Dai Pra, A.M. Paganoni and G. Posta. (2002). Entropy ineq ualities for unbounded spin systems. Ann. of Probab. 30 1959–1976

work page 2002

[12] [12]

Diaconis and D

P. Diaconis and D. Stroock. (1991). Geometric bounds for eige nvalues of Markov chains. Ann. App. Probab. 1 36–62

work page 1991

[13] [13]

R. L. Dobrushin. (1968). The description of a random ﬁeld by me ans of conditional proba- bilities and condition of its regularity. Theory Probab. Appl. 13 197–224

work page 1968

[14] [14]

R. L. Dobrushin and S. Shlosman. (1985). Completely analytical Gibbs ﬁelds. Statistical physics and dynamical systems 371-403

work page 1985

[15] [15]

A. Eberle. (2016). Reﬂection couplings and contraction rates for diﬀusions. Probab. Theory Relat. Fields. 164 851–886

work page 2016

[16] [16]

Guillin, C

A. Guillin, C. L´ eonard, L. M. Wu and N. Yao. (2009). Transportation-information inequalities for Markov processes. Probab. Theory Relat. Fields 144 669–695

work page 2009

[17] [17]

Transportation-information inequalities for Markov processes (II) : relations with other functional inequalities

A. Guillin, C. L´ eonard, F.Y. Wang and L. M. Wu. Transportation- information inequal- ities for Markov processes (II): relations with other functional in equalities. Available on arXiv:0902.2101

work page internal anchor Pith review Pith/arXiv arXiv

[18] [18]

Guillin, A

A. Guillin, A. Joulin, C. L´ eonard and L. M. Wu, Transportation-in formation inequalities for Markov processes (III). Processes with jumps. Preprint 2008

work page 2008

[19] [19]

Hairer and J

M. Hairer and J. C. Mattingly. (2011). Yet another look at Harr is’ ergodic theorem for Markov chains. Seminar on Stochastic Analysis, Random Fields and Applicat ions VI 109–117

work page 2011

[20] [20]

Joulin and Y

A. Joulin and Y. Ollivier. (2010). Curvature, concentration and error estimates for Markov chain Monte Carlo. Ann. Probab. 38 2418–2442

work page 2010

[21] [21]

G. F. Lawler and A. D. Sokal. (1988). Bounds on the L2 spectrum for Markov chains and Markov processes: a generalization of Cheeger’s inequalities. Trans. Amer. Math. Soc. 309 557–580

work page 1988

[22] [22]

Lin and S.T

Y. Lin and S.T. Yau. (2010). Ricci curvature and eigenvalue est imate on locally ﬁnite graphs. Math. Res. Lett. 17 343–356

work page 2010

[23] [23]

Y. Lin, L. Y. Lu, S-T Yau. (2011). Ricci curvature of graphs. Tohoku Math. J. (2) 63 605–627

work page 2011

[24] [24]

Liu and Y

W. Liu and Y. T. Ma. (2009). Spectral gap and convex concent ration inequalities for birth- death processes. Ann.Inst. H. Poincar´ e Probab. Statist.45 58–69

work page 2009

[25] [25]

Liu, Y.T

W. Liu, Y.T. Ma and L.M. Wu. (2016). Spectral gap, isoperimetry and concentration on trees. Sciences in China, Mathematics 59 539–556. RICCI CUR V ATURE AND W1-EXPONENTIAL CONVERGENCE 51

work page 2016

[26] [26]

Luo and J

D. Luo and J. Wang. (2016). Exponential convergence in Wass erstein distance for diﬀusion processes without uniform dissipativity. Math. Nachr. 289 1909–1926

work page 2016

[27] [27]

Y.T. Ma, R. Wang and L.M. Wu. (2016). Log-Sobolev, isoperimetr y and transport inequali- ties on graphs. Acta. Math. Sinica 32 1221–1236

work page 2016

[28] [28]

Lectures on Probability Theory and Statistics

Martinelli, F., Lectures on Glauber dynamics for discrete spin models. Lectures on Probability Theory and Statistics. Springer Berlin Heidelberg 1999

work page 1999

[29] [29]

Meyn and R.L

S.P. Meyn and R.L. Tweedie. (1993). Markov chains and stochas tic stability. Communica- tions and Control Engineering Series. Springer-Verlag

work page 1993

[30] [30]

Ollivier

Y. Ollivier. (2009). Ricci curvature of Markov chains on metric s paces. J. Funct. Anal. 256 810–864

work page 2009

[31] [31]

von Renesse and K.T

M.K. von Renesse and K.T. Sturm. (2010). Transport inequalitie s, gradient estimates, en- tropy, and Ricci curvature. Communications on Pure and Applied Mathematics 58 923–940

work page 2010

[32] [32]

Saloﬀ-Coste

L. Saloﬀ-Coste. (1997). Lectures on ﬁnite Markov chains. Lecture Notes in Mathematics 1665, 301–413

work page 1997

[33] [33]

Sammer, P

M. Sammer, P. Tetali. (2009). Concentration on the discrete t orus using transportation. Combin. Probab. Comput. 18 835–860

work page 2009

[34] [34]

L. Wu. (2004). Essential spectral radius for Markov semigro ups (I): discrete time case. Probab. Th. Rel. Fields 128 255–321

work page 2004

[35] [35]

L. Wu. (2006). Poincar´ e and transportation inequalities for G ibbs measures under the Do- brushin uniqueness condition. Ann. Probab. 34 1960–1989

work page 2006

[36] [36]

J. Q. Zhong and H. C. Yang. (1984). On the estimate of the ﬁrs t eigenvalue of a compact Riemannian manifold. Sci. Sinica Ser. A 27 1265–1273. Lingyan Cheng. Center for Applied Mathematics, Tianjin Uni versity, Tianjin 300072, PR China. E-mail address : chengly@amss.ac.cn Ruinan Li. School of Statistics and Information, Shanghai U niversity of Inter- nati...

work page 1984