Causal transport on path space

Rama Cont,Fang Rui Lim
2024-12-04
Abstract:We study properties of causal couplings for probability measures on the space of continuous functions. We first provide a characterization of bicausal couplings between weak solutions of stochastic differential equations. We then provide a complete description of all such bicausal Monge couplings. In particular, we show that bicausal Monge couplings of $d$-dimensional Wiener measures are induced by stochastic integrals of rotation-valued integrands. As an application, we give necessary and sufficient conditions for bicausal couplings to be induced by Monge maps and show that such bicausal Monge transports are dense in the set of bicausal couplings between laws of SDEs with regular coefficients.
Probability
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is the properties of causal couplings and bicausal couplings on the path space, especially the bicausal couplings between probability measures of continuous functions. Specifically, the article focuses on: 1. **Bicausal couplings of weak solutions**: The article first provides a characterization of bicausal couplings between weak - solution stochastic differential equations (SDEs). 2. **Complete description of Monge couplings**: The article further details all such bicausal Monge couplings and shows that the bicausal Monge couplings of the d - dimensional Wiener measure are induced by stochastic integrals of rotation - valued integrals. ### Main contributions - **Characterization of bicausal couplings**: The article proves the characteristics of bicausal couplings between weak - solution SDEs and studies the cases where these couplings are induced by mappings. - **Inherited structure of Monge transport**: The article shows that these mappings inherit the differential structures of η and ν, that is, T satisfies a stochastic differential equation (SDE). In particular, the bicausal Monge transport of the d - dimensional Wiener measure is induced by stochastic integrals of Od - valued integrals. - **Application of Ito mapping**: For SDEs with unique strong solutions, the Ito mapping provides an example of causal transport from the Wiener measure to the distribution of solutions on the path space. The article also characterizes all bicausal Monge mappings from Wd to ν as combinations of Ito mappings and integrals adapted to Od - valued processes. ### Applications - **Existence conditions**: Through the above characterizations, the article gives simple conditions for the existence of bicausal Monge mappings. - **Differential representation**: Any bicausal coupling between Wiener measures has a differential representation, that is, it can be represented by a stochastic differential equation. - **Necessary and sufficient conditions**: The necessary and sufficient conditions for bicausal couplings between weak - solution SDEs to be induced by Monge mappings are established. - **Density**: The article proves that under certain conditions, bicausal Monge transports are dense in the set of bicausal couplings between weak - solution SDEs. ### Formula summary - **Differential structure**: For the bicausal Monge mapping \(T\), it satisfies the following SDE: \[ T_t=\tilde{z}+\int_{0}^{t}b(s, T)ds+\int_{0}^{t}\sigma(s, T)Q_s\sigma^{\dagger}(s, X)dM^{\eta}_s\quad\eta - a.s. \] where \(Q_s\) is a \((H^{\eta}_t)\)-progressively measurable process and satisfies: \[ \sigma(t, T)Q_t = \sigma(t, T)Q_t\sigma^{\dagger}(t, X)\sigma(t, X)\quad d\eta\otimes dt - a.e. \] Through these studies, the article provides an important theoretical basis for understanding causal transport on the path space and its applications.