We show that, under mild assumptions, every distribution on the hypercube $\{0, 1\}^{n}$ that admits a polynomial-time Markov chain approximate sampler also has an exact sampling algorithm with expected running time in poly$(n)$.

翻译：我们证明，在温和的假设下，超立方体 $\{0, 1\}^{n}$ 上每个允许多项式时间马尔可夫链近似采样器的分布，同样存在一个期望运行时间为 poly$(n)$ 的精确采样算法。