# Inferring marginal distributions on graphs

**URL:** <https://discourse.julialang.org/t/inferring-marginal-distributions-on-graphs/88159>\
**Category:** Probabilistic Programming\
**Tags:** gen\
**Created:** [October 3, 2022, 1:36pm UTC](https://discourse.julialang.org/t/inferring-marginal-distributions-on-graphs/88159 "2022-10-03T13:36:59Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![hbooth](https://avatars.discourse-cdn.com/v4/letter/h/e19b73/32.png) [@hbooth](https://discourse.julialang.org/u/hbooth)\
**Post date:** [October 3, 2022, 1:36pm UTC](https://discourse.julialang.org/t/inferring-marginal-distributions-on-graphs/88159/1 "2022-10-03T13:36:59Z")

</div>

Hi all,

I am looking for some guidance on how to approach a computational problem I have, hopefully using some of the amazing julia packages! This is early stages, and more conceptual at this point. I want to get some high level opinions on the best computational framework and associated packages.

In essence, I am performing inference on the end points of a random walk on a graph. In other words, suppose that I have some graph G = (V,E) and a stochastic process X(t) representing a random walk on the vertex set V.

I want to estimate P(X(T) = v | X(0) = v0) in the most efficient way possible. Computationally, I have a method of drawing samples, i.e. realised walks. This is a black box calculation - it is the output of a complicated julia function.

Any advice, pointers, recommendations for this probabilistic programming newbie? The graph has around 10,000 vertices, and so I feel like I need an efficient sampling algorithm in order to do anything meaningful. However I am unsure if this is possible given that samples come from a black box calculation.

Thanks in advance for taking the time!