MDPN

A Unified notation for Markov Decision Processes (MDPs)

Building on the efforts of Philip Thomas in arXiv:1512.09075, we build a common Latex style for writing PO(MDP) related texts for publication, teaching, etc.

Why?

• Have a standard notation, to save everyone time (readers, writers) of anything related to MDPs and friends
• It is annoying to write the same paragraphs in every paper while trying to avoid self plagiarizing at the same time. This effort replaces such a paragraphs with a single phrase, e.g "we use the notation from X et.al".

Usage

% ...
\usepackage[alpha]{mdpn} % Select verbosity level

% In text
MDP is given by $\MDP$, with $\sset$ being the set of states, $\aset$ is the set of actions, $\Rfun$ is the reward function...
Let $\T{s}{a}{s'}$ be the probability of transitioning from ....

% ...

See the sample file for a concrete example. To compile the sample, simple run make if you have latexmk already installed.

Contributing

Simply create an issue, open a pull request, fork away

But wont this lead to https://xkcd.com/927/?

Hopefully not!