1: \begin{abstract}

2: 		In this note, we establish a descent lemma for the population limit Mirrored Stein Variational Gradient Method~(MSVGD). This descent lemma does not rely on the path information of MSVGD but rather on a simple assumption for the mirrored distribution $\nabla\Psi_{\#}\pi\propto\exp(-V)$. Our analysis demonstrates that MSVGD can be applied to a broader class of constrained sampling problems with non-smooth $V$. We also investigate the complexity of the population limit MSVGD in terms of dimension $d$.

3: 		

4: 		

5: 		

6: %		Stein Variational Gradient Descent (SVGD) is an important alternative to the Langevin-type algorithms for sampling from a probability distribution of the form $\pi(x) \propto \exp(-V(x))$. 

7: %		In the existing theory of Langevin-type algorithms and SVGD,  the potential function $V$ is often assumed to be $L$-smooth. 

8: %		However, this restrictive condition excludes a large class of functions such as polynomials of degree greater than $2$. 

9: %		Our paper studies the convergence of the SVGD algorithm in population limit for distributions with $(L_0,L_1)$-smooth potentials. 

10: %		This relaxed smoothness assumption was introduced by \cite{zhang2019gradient} for gradient clipping algorithms in optimization.

11: %		With the help of trajectory-independent auxiliary conditions, we provide a descent lemma establishing that the algorithm decreases the KL-divergence at each iteration and prove a complexity bound for SVGD in population limit in terms of the Stein  Fisher information.

12: 		

13: 	\end{abstract}

14: