Next Article in Journal
Exploring Flatter Loss Landscape Surface via Sharpness-Aware Minimization with Linear Mode Connectivity
Previous Article in Journal
Prime Strictly Concentric Magic Squares of Odd Order
Previous Article in Special Issue
Revisiting the Group Classification of the General Nonlinear Heat Equation ut = (K(u)ux)x
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Harnack Inequality for Self-Repelling Diffusions Driven by Subordinated Brownian Motion

School of Mathematics and Statistics, Donghua University, 2999 North Renmin Rd., Shanghai 201620, China
*
Author to whom correspondence should be addressed.
These authors contributed equally to this work.
Mathematics 2025, 13(8), 1262; https://doi.org/10.3390/math13081262
Submission received: 18 February 2025 / Revised: 2 April 2025 / Accepted: 6 April 2025 / Published: 11 April 2025
(This article belongs to the Special Issue Mathematics and Applications)

Abstract

:
In this paper, we consider a self-repelling diffusion driven by the Lévy process. By using the coupling argument, we establish the corresponding Bismut formula and Harnack inequality.

1. Introduction

In 1992, Durrett and Rogers [1] introduced a model for the shape of a growing polymer (Brownian polymer):
X t = X 0 + B t + 0 t 0 s f ( X s X u ) d u d s ,
where B is a d-dimensional standard Brownian motion and f is Lipschitz continuous. X t corresponds to the location of the end of the polymer at time t. In general, the Equation (1) defines a self-interacting diffusion without any assumptions on f. We will call it self-repelling (resp. self-attracting) if, for all x R d , x · f ( x ) 0 (resp. 0 ), in other words, if it is more likely to stay away from (resp. come back to) the places it has already visited before. Numerous disciplines, including biophysics and financial mathematics, have made extensive use of the concept. For instance, the motion and interactions of colloid particles in a solution are similar to those of Brownian polymers in the field of colloid chemistry. Similar to the influence of the d-dimensional standard Brownian motion B t in the equation, colloid particles are susceptible to the Brownian motion of solvent molecules. The self-interaction term 0 t 0 s f ( X s X u ) d u d s can be compared to the attractive or repulsive forces that exist among the particles. Determining the stability, aggregation behavior, and phase separation of colloid systems, among other phenomena, requires an understanding of these interactions.
In 2002, Benaïm et al. [2] considered a self-interacting diffusion with dependence on the (convolved) empirical measure as follows:
d X t = 2 d B t 1 t 0 t W ( X t X s ) d s d t ,
where W is an interaction potential function. The fact that the drift term is divided by t distinguishes these diffusions from Brownian polymers. More works can be found in Benaïm et al. [3], He and Guo [4], Pan and Jiang [5], and the references therein.
In 2015, Benaïm et al. [3] considered the equation
X t = x + B t + 0 t b X s d s 0 t 0 s f X s X u d u d s ,
where x R , B = { B t , t 0 } is a Brownian motion on R 1 , and f is a 2 π -periodic function with sufficient regularity. For some suitable functions, they showed the existence and uniqueness of a solution with the Markov property and that it has the Feller property and an explicit form of an invariant probability. In this short note, we establish a Bismut-type formula and a Harnack inequality for self-interaction stochastic differential equations:
X t = x + W S t + 0 t b X s d s 0 t 0 s f X s X u d u d s + v t ,
where x , v R d , W = { W t , t 0 } is a Brownian motion on R d and S t is the subordinator induced by a Bernstein function B of independent W, and b and f are two measurable functions. In order to obtain our results, we need the following condition:
There exist positive constants K i , i = 1 , 2 , such that
b ( · ) K 1 , f ( · ) K 2 ,
where b ( x ) = ( x 1 b ( x ) , , x d b ( x ) ) , f ( x ) = ( x 1 f ( x ) , , x d f ( x ) ) and · denotes the uniform norm with respect to x.
It is important to note that the solution of (3) is not a Markov process in general because of the explicit dependence of its dynamics on the entire historical path of X, rather than solely on its current state. On the other hand, as is well known, the functional inequalities and derivative formulas for Markov-type stochastic differential equations have been widely studied (see, for instance, Forte [6], Vespri et al. [7], Negro [8]), while the research on functional inequalities and derivative formulas related to non-Markov-type stochastic differential equations is still very scarce. Thus, the Bismut-type formula and the Harnack inequality for such equations as (3) appear to be worth studying. For simplicity, we will use C to represent a positive constant that depends only on the subscripts, and its value may vary depending on how it is presented.
Define the class { P t , t 0 } of operators by
P t g ( x ) : = E g X t ( x ) , t 0 , g C b 1 ( R d ) ,
where X t ( x ) is the solution of (3) with the initial value x R d , and C b 1 ( R d ) is the space of all bounded functions on R d with one-order continuous derivatives. At the same time, we can define
h g ( x ) = lim ε 0 g ( x + ε h ) g ( x ) ε , x , h R d , g C b 1 ( R d ) .
The main purpose of this note is to expound and prove the following theorems.
Theorem 1. 
Under the condition ( C ) , for any function g C b 1 ( R d ) , T > 0 and x , h R d , we have
h P T g ( x ) = E g ( X T ) N T ,
where
N T = 0 T S T s S T h b ( X s ) + h S T 0 s s u S T h f ( X s X u ) d u d W S s .
Theorem 2. 
Under condition ( C ) , for any non-negative g C b 1 R d , p > 1 , T > 0 and x , h R d , the operator P T satisfies that
P T g ( x ) p P T g p ( h ) exp p p 1 C T , K 1 , K 2 h x 2 .

2. Preliminaries

Consider the equation
X t = x + W t + 0 t b ( X s ) d s 0 t 0 s f ( X s X u ) d u d s + v t , t 0 ,
where W = { W t , t 0 } is a standard Brownian motion on R d and is an absolutely continuous and strictly increasing function on [ 0 , + ) with 0 = 0 . Clearly, under the above assumption, this equation has a unique solution X x and the following Bismut formula holds. For more details, see Bao [9], Bismut [10], Fan [11], Guillin and Wang [12], and the references therein.
Lemma 1. 
Under the condition ( C ) , we have
h P T g ( x ) = E g ( X T ) L T
for any function g C b 1 ( R d ) , T > 0 and x , h R d , where
L T = 0 T T s T h b ( X s ) + h T 0 s s u T h f ( X s X u ) d u d W s .
Proof. 
For ε > 0 , we let Y solve the equation
d Y s = d W s + b ( X s ) d s ε h T d s 0 s f ( X s X u ) d u + v
with Y 0 = x + ε h , x , h R d . Then Y s = X s + ε ( T s ) T h , s [ 0 , T ] . Denote
u s = b ( X s ) b ( Y s ) ε h T + 0 s f ( Y s Y u ) f ( X s X u ) d u
with s [ 0 , T ] . It follows from the Girsanov theorem that W ˜ s = W t 0 t u s d s , s [ 0 , T ] is also a Brownian motion under the probability d Q : = R ε ( T ) d P , where
R ε ( T ) = exp 0 T u s , d W s 1 2 0 T | u s | 2 d s .
Thus, under d Q : = R ε d P , the distribution of Y y coincides with that of X x + ε h under P by reformulating the equation for Y s as follows:
d Y s = d W s + b ( X s ) d s ε h T d s 0 s f ( X s X u ) d u + v = d W ˜ s + b ( Y s ) d s 0 s f ( Y s Y u ) d u + v ,
which implies that E g ( Y t ) = E R ε g ( X t ) . Consider now the local martingale
M T = 0 T u s d W s = 0 T ( T s ) ε T h b ( X s ) + ε h T 0 s ( s u ) ε T h f ( X s X u ) d u d W s : = ε L T .
According to the condition ( C ) , we obtain that
M T = 0 T ( T s ) ε T h b ( X s ) + ε h T 0 s ( s u ) ε T h f ( X s X u ) d u 2 d s C K 1 , K 2 , T ε 2
and
h P T g ( x ) = lim ε 0 P T g ( x + ε h ) P T g ( x ) ε = lim ε 0 E g ( X T ) R ε 1 ε = lim ε 0 E g ( X T ) M T ε = E g ( X T ) L T ,
and the theorem follows. □
Lemma 2. 
Under the condition ( C ) , the operator P T satisfies that
P T g ( x ) p P T g p ( h ) exp p p 1 C T , K 1 , K 2 h x 2
for all p > 1 , T > 0 , x , h R d and all non-negative g C b 1 ( R d ) .
Proof. 
By Young’s inequality in Arnaudon et al. [13], we have, for all ρ > 0 ,
h P T g ( x ) ρ P T ( g log g ) ( x ) P T g ( x ) log P T g ( x ) + P T g ( x ) ρ log E e L T ρ .
Define β ( s ) = 1 + s ( p 1 ) , δ ( s ) = x + s ( h x ) , s [ 0 , 1 ] and ρ = p 1 β ( s ) . Then, we have
d d s log P T g β ( s ) p β ( s ) ( δ ( s ) ) p β ( s ) P T g β ( s ) ( δ ( s ) ) p 1 β ( s ) P T g β ( s ) log g β ( s ) ( δ ( s ) ) P T g β ( s ) ( δ ( s ) ) log P T g β ( s ) ( δ ( s ) ) h x P T g β ( s ) ( δ ( s ) ) p ( p 1 ) β 2 ( s ) log E e 2 L T ρ 2 1 2 p C K 1 , K 2 , T p 1 h x 2 .
It follows from (7) that
L T = 0 T 1 ε 2 b ( Y s ) b ( X s ) + ε h T 0 s f ( Y s Y u ) d u + 0 s f ( X s X u ) d u 2 d s 0 T 1 ε 2 b ε ( T s ) h T + ε h T + 0 s f ε ( T u ) h T ε ( T s ) h T d u 2 d s 0 T K 1 T s T + 1 T + 0 s K 2 s u T d u 2 h 2 d s C K 1 , K 2 , T h 2 .
This completes the proof. □
For the subordinator process S = { S t , t 0 } , by Wang [14] and Zhang [15], we define the its regularization as follows:
S t ε : = 1 ε t t + ε S s d s , t 0 , ε > 0 .
Then, S ε is strictly increasing, absolutely continuous and S ε S a.s. as ε 0 . For each ε > 0 , we consider the following approximation equation of (3):
X t ε = x + W S t ε + 0 t b X s ε d s 0 t 0 s f X s ε X u ε d u d s + v t , t 0 .
In order to prove Theorem 1, we need some necessary lemmas.
Lemma 3. 
For any p 1 , T 0 and x R d , we have
lim ε 0 sup 0 t T E X t ε ( x ) X t ( x ) p = 0 .
Proof. 
Using (10), we have
sup 0 t T E X t ε X t sup 0 t T E W S t ε W S t + b 0 t sup 0 t T E X s ε X s d s + f 0 t 0 s sup 0 t T E | ( X s ε X u ε ) ( X s X u ) | d u d s sup 0 t T E W S t ε W S t + K 1 0 t sup 0 t T E | X s ε X s | d s + K 2 0 t 0 s sup 0 t T E | X s ε X s | + sup 0 t T E | X u ε X u | d u d s sup 0 t T E W S t ε W S t + C K 1 , K 2 , T 0 t sup 0 t T E | X s ε X s | d s + C K 2 , T 0 t 0 s sup 0 t T E | X u ε X u | d u d s .
According to the Gronwall’s inequality (given in Dragomir [16]), the lemma follows since S t ε S t , a.s. as ε 0 . □
Lemma 4 
(Zhang [15]). Assume that ξ t is a bounded, continuous and F t W -adapted R d -valued process; we then have
lim ε 0 E 0 t ξ s d W S t ε 0 t ξ s d W S t p = 0
for each p , t > 0 , where S t ε is defined by (9).
Lemma 5. 
For t > 0 and h R d , we have
lim ε 0 E 0 t h b ( X s ε ) d W S s ε = E 0 t h b ( X s ) d W S s
and
lim ε 0 E 0 t 0 s h f ( X s ε X u ε ) d u d W S s ε = E 0 t 0 s h f ( X s X u ) d u d W S s .
Proof. 
The lemma follows from Lemmas 3 and 4. □

3. Proofs

In this section, we give proofs of Theorems 1 and 2. Consider the operators
P t ε f ( x ) = E f X t ε ( x ) , t 0 , f C b 1 ( R d ) ,
where X ε ( x ) is the solution of (10) with the initial value x R d . Then, Lemma 3 implies that
lim ε 0 P T ε g ( x ) = P T g ( x ) , g C b 1 ( R d ) .
According to Lemma 1, we have
h P T g ( x ) = E g ( X T ) L T
This holds true for both , where is an absolutely continuous and strictly increasing function on [ 0 , + ) with 0 = 0 , In particular, we take = S ε ,
h P T ε g ( x ) = E g ( X T S ε ) L T = E g ( X T ε ) L T = E g ( X T ε ) 0 T S T ε s S T ε h b ( X s ε ) + h S T ε 0 s s u S T ε h f ( X s ε X u ε ) d u d W S ε
for any function g C b 1 ( R d ) and h R d under the condition (C). Assuming ε 0 and using Lemmas 3–5, and (14), we obtain that
h P T g ( x ) = E g ( X T ) N T
for any function g C b 1 ( R d ) and h R d , where
N T = 0 T S T s S T h b ( X s ) + h S T 0 s s u S T h f ( X s X u ) d u d W S s .
This completes the proof of Theorem 1.
Now, we check Theorem 2. Taking = S ε and applying the same technique, we obtain
P T ε g ( x ) = E P T S ε g ( x ) E P T S ε g p ( x ) 1 p exp p p 1 C T , K 1 , K 2 h x 2 p 1 p P T ε g p ( x ) 1 p exp p p 1 C T , K 1 , K 2 h x 2 p 1 p
for all p > 1 , T > 0 and x , h R d , and all non-negative g C b 1 ( R d ) . Combining Lemmas 3–5, (13), and letting ε 0 to lead that
P T g ( x ) P T g p ( h ) 1 p exp p p 1 C T , K 1 , K 2 h x 2 p 1 p
for all x , h R d and all non-negative g C b 1 ( R d ) .
It remains to apply the approximation argument in Zhang [15] to finish the proof. Let
b ˜ ( x ) : = b ( x ) K x , f ˜ ( x ) : = f ( x ) K x , x R d .
According to Da Prato et al. [17], because for any n N , the mapping i d n 1 b ˜ : R d R d and i d n 1 f ˜ : R d R d are injective under our condition, the maps
b ( n ) : = n ( i d n 1 b ˜ ) 1 ( x ) x + K x , n N , x R d
and
f ( n ) : = n ( i d n 1 f ˜ ) 1 ( x ) x + K x , n N , x R d
are globally Lipschitz continuous. Denote ( X t ( n ) ( x ) ) t 0 , solve the equation (3) with b, f replaced by b ( n ) , f ( n ) , respectively, and P t ( n ) is the associated operator. Based on the previous proofs, the statements of Theorems 1 and 2 hold for P T ( n ) in place of P T .
On the other hand, we combine the argument in Wang [14] that
lim n X T ( n ) ( x ) = X T ( x ) , a . s . ,
for all x R d . And
lim n P T ( n ) g = P T g , g C b 1 ( R d ) .
Therefore, the claim follows if we let n .

Author Contributions

Conceptualization, Y.S. and L.Y.; Methodology, Y.S. and L.Y.; Formal analysis, Y.S. and L.Y.; Investigation, Y.S. and L.Y.; Writing—original draft, Y.S. and L.Y.; writing—review and editing, Y.S. and L.Y. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by NSFC (11971101, 12171081) and the NSF of Shanghai (No. 24ZR1402900).

Data Availability Statement

Data are contained within the article.

Acknowledgments

We thank the anonymous referees for the careful reading of the manuscript and their suggestions, which helped to improve this paper.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Durrett, R.; Rogers, L.C.G. Asymptotic behavior of Brownian polymer. Probab. Theory Relat. Fields 1992, 92, 337–349. [Google Scholar] [CrossRef]
  2. Benaïm, M.; Ledoux, M.; Raimond, O. Self-interacting diffusions. Probab. Theory Relat. Fields 2002, 122, 1–41. [Google Scholar] [CrossRef]
  3. Benaïm, M.; Ciotir, I.; Gauthier, C.-E. Self-repelling diffusions via an infinite dimensional approach. Stoch PDE Anal. Comp. 2015, 3, 506–530. [Google Scholar] [CrossRef]
  4. He, J.; Guo, Q. An explicit method for the self-interacting diffusion driven by fractional Brownian motion under global Lipschitz conditions. Appl. Math. Lett. 2022, 134, 108379. [Google Scholar] [CrossRef]
  5. Pan, Y.; Jiang, H. Asymptotic properties for quadratic functionals of linear self-repelling diffusion process and applications. Stoch. Anal. Appl. 2022, 40, 691–713. [Google Scholar] [CrossRef]
  6. Forte, B. Functional Equations and Inequalities; Springer: Berlin, Germany, 2010. [Google Scholar]
  7. Vespri, V.; Gianazza, U.; Monticelli, D.D.; Punzo, F.; Andreucci, D. Harnack Inequalities and Nonlinear Operators: Proceedings of the INdAM Conference to Celebrate the 70th Birthday of Emmanuele DiBenedetto; Springer: Berlin, Germany, 2021. [Google Scholar]
  8. Negro, L. Harnack inequality for Ornstein-Uhlenbeck type operators. Arch. Math. 2020, 114, 331–341. [Google Scholar] [CrossRef]
  9. Bao, J.; Wang, F.; Yuan, C. Bismut formulae and applications for functional SPDEs. Bull. Sci. Math. 2013, 137, 509–522. [Google Scholar] [CrossRef]
  10. Bismut, J.M. Large Deviations and the Malliavin Calculus; Birkhäuser: Boston, MA, USA, 1984. [Google Scholar]
  11. Fan, X.; Ren, Y. Bismut formulas and applications for stochastic (functional) differential equations driven by fractional Brownian motions. Stoch. Dyn. 2017, 17, 1750028. [Google Scholar] [CrossRef]
  12. Guillin, A.; Wang, F. Degenerate Fokker-Planck equations: Bismut formula, gradient estimate and Harnack inequality. J. Differ. Equ. 2012, 253, 20–40. [Google Scholar] [CrossRef]
  13. Arnaudon, M.; Thalmaier, A.; Wang, F. Gradient estimates and Harnack inequalities on non-compact Riemannian manifolds. Stoch. Proc. Appl. 2009, 119, 3653–3670. [Google Scholar] [CrossRef]
  14. Wang, F.; Wang, J. Harnack inequalities for stochastic equations driven by Lévy noise. J. Math. Anal. Appl. 2014, 410, 513–523. [Google Scholar] [CrossRef]
  15. Zhang, X. Derivative formulas and gradient estimates for SDEs driven by α-stable processes. Stoch. Proc. Appl. 2013, 123, 1213–1228. [Google Scholar] [CrossRef]
  16. Dragomir, S.S. Some Gronwall Type Inequalities and Applications. 2002. Available online: https://www.researchgate.net/publication/228607920_Some_Gronwall_type_inequalities_and_applications (accessed on 6 April 2025).
  17. Da Prato, G.; Röckner, M.; Wang, F. Singular stochastic equations on Hilbert spaces: Harnack inequalities for their transition semigroups. J. Funct. Anal. 2009, 257, 992–1017. [Google Scholar] [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Sun, Y.; Yan, L. Harnack Inequality for Self-Repelling Diffusions Driven by Subordinated Brownian Motion. Mathematics 2025, 13, 1262. https://doi.org/10.3390/math13081262

AMA Style

Sun Y, Yan L. Harnack Inequality for Self-Repelling Diffusions Driven by Subordinated Brownian Motion. Mathematics. 2025; 13(8):1262. https://doi.org/10.3390/math13081262

Chicago/Turabian Style

Sun, Yaqin, and Litan Yan. 2025. "Harnack Inequality for Self-Repelling Diffusions Driven by Subordinated Brownian Motion" Mathematics 13, no. 8: 1262. https://doi.org/10.3390/math13081262

APA Style

Sun, Y., & Yan, L. (2025). Harnack Inequality for Self-Repelling Diffusions Driven by Subordinated Brownian Motion. Mathematics, 13(8), 1262. https://doi.org/10.3390/math13081262

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop