# Julia's SciML and UDE in temporal networks: preprint out

**URL:** <https://discourse.julialang.org/t/julias-sciml-and-ude-in-temporal-networks-preprint-out/136099>\
**Category:** Modelling & Simulations\
**Tags:** sciml, network, paper\
**Created:** [March 9, 2026, 5:11am UTC](https://discourse.julialang.org/t/julias-sciml-and-ude-in-temporal-networks-preprint-out/136099 "2026-03-09T05:11:01Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![gvdr](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/gvdr/32/6387_2.png) [@gvdr](https://discourse.julialang.org/u/gvdr)\
**Post date:** [March 9, 2026, 5:11am UTC](https://discourse.julialang.org/t/julias-sciml-and-ude-in-temporal-networks-preprint-out/136099/1 "2026-03-09T05:11:01Z")

</div>

Hi everyone, I got a new pre-print out on arXiv, that uses the Julia’s SciML ecosystem quite a bit.

> **[Random Dot Product Graphs as Dynamical Systems: Limitations and Opportunities](https://arxiv.org/abs/2603.05703)**
>
> Can we learn the differential equations governing the evolution of a temporal network? We investigate this within Random Dot Product Graphs (RDPGs), where each network snapshot is generated from latent positions evolving under unknown dynamics. We...

The base idea is: if we consider a complex network that changes in time as a dynamical system (and take a nice model for it, such as a Random Dot Product Graphs, that gives in theory smooth trajectories) can we use SciML to recover the differential equations that guide that system?

The idea started out as a thesis project for one of my students some years ago. At that point, I was convinced the problem was easy to solve, and mostly about implementing some clever UDE code. Student worked super hard, very well, on that project, but our results were meh. I parked it for a while. Then I got back to it after some reflection, but the more I worked on it, the harder the problem showed to be, and now I’m quite convinced it’s actually not possible to solve it _as is_. So now it’s mostly a differential-geometry-with-some-Julia code kind of paper.

Yet, maybe I can nerd snipe some of you to come up with something super clever and fix it?

---

<div class="post-metadata">

**Author:** ![gvdr](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/gvdr/32/6387_2.png) [@gvdr](https://discourse.julialang.org/u/gvdr)\
**Post date:** [March 9, 2026, 10:37am UTC](https://discourse.julialang.org/t/julias-sciml-and-ude-in-temporal-networks-preprint-out/136099/2 "2026-03-09T10:37:07Z")

</div>

PS the code itself is here [GitHub - gvdr/RDPG\_in\_differential\_geometry: manuscript and code for Random Dot Product Graphs as Dynamical Systems: Limitations and Opportunities · GitHub](https://github.com/gvdr/RDPG_in_differential_geometry)