bias delocalization 1hamiltonian monte carlo 1high-dimensional sampling 1underdamped langevin 1Wasserstein distance 1
From the 1 of 3 linked papers with an AI index.
3 papers
math.PR2026
Nearly sharp comparison results for sliced and max-sliced Wasserstein distances
Jonathan Niles-Weed, Jacob Shkrob
We prove new comparison results between the Wasserstein distance and its sliced and max-sliced counterparts. First, we show that the Hölder exponent~ obtained by Bob…
stat.CO2026
Delocalization of bias in unadjusted Hamiltonian Monte Carlo and underdamped Langevin
Yifan Chen, Xiaoou Cheng, Jonathan Niles-Weed +1
The paper analyzes the bias of unadjusted Hamiltonian Monte Carlo and underdamped Langevin samplers, showing that controlling the Wasserstein‑2 bias of any marginal requires only O…
stat.ML2025
Convergence of Unadjusted Langevin in High Dimensions: Delocalization of Bias
Yifan Chen, Xiaoou Cheng, Jonathan Niles-Weed +1
The unadjusted Langevin algorithm is commonly used to sample probability distributions in extremely high-dimensional settings. However, existing analyses of the algorithm for stron…