1: \begin{abstract}

2: 

3: We introduce an adaptive output-sensitive

4: Metro\-polis-Hast\-ings algorithm for probabilistic models

5: expressed as programs, Adaptive Lightweight

6: Metro\-polis-Hast\-ings (AdLMH). The algorithm extends

7: Light\-weight Metro\-polis-Hast\-ings (LMH) by adjusting the

8: probabilities of proposing random variables for modification to

9: improve convergence of the program output. We show that AdLMH

10: converges to the correct equilibrium distribution and compare

11: convergence of AdLMH to that of LMH on several test problems to

12: highlight different aspects of the adaptation scheme. We observe

13: consistent improvement in convergence on the test problems.

14: 

15: \keywords{Probabilistic programming, adaptive MCMC}

16: \end{abstract}

17: