Mathematics > Optimization and Control
[Submitted on 18 Sep 2020 (v1), last revised 9 Apr 2021 (this version, v2)]
Title:A Low-rank Approximation for MDPs via Moment Coupling
View PDFAbstract:We introduce a framework to approximate a Markov Decision Process that stands on two pillars: state aggregation -- as the algorithmic infrastructure; and central-limit-theorem-type approximations -- as the mathematical underpinning of optimality guarantees. The theory is grounded in recent work Braverman et al (2020} that relates the solution of the Bellman equation to that of a PDE where, in the spirit of the central limit theorem, the transition matrix is reduced to its local first and second moments. Solving the PDE is $\textit{not}$ required by our method. Instead, we construct a "sister" (controlled) Markov chain whose two local transition moments are approximately identical with those of the focal chain. Because of this $\textit{moment matching}$, the original chain and its "sister" are coupled through the PDE, a coupling that facilitates optimality guarantees. Embedded into standard soft aggregation algorithms, moment matching provided a disciplined mechanism to tune the aggregation and disaggregation probabilities. The computational gains arise from the reduction of the effective state space from $N$ to $N^{\frac{1}{2}+\epsilon}$ is as one might intuitively expect from approximations grounded in the central limit theorem.
Submission history
From: Amy B.Z. Zhang [view email][v1] Fri, 18 Sep 2020 17:53:17 UTC (18,188 KB)
[v2] Fri, 9 Apr 2021 19:53:36 UTC (36,630 KB)
Current browse context:
math.OC
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.