EMBEr
Exploring Moment-Based Equivalence
for probabilistic program

01/09/2026

One year of EMBEr

It's been a year since EMBEr started: a good moment to pause and look back at what happened.
Here we give a short recap of all the publications and activities carried out through this year thanks to EMBEr's funding.

Publications

Two papers funded by EMBEr came out this year:

- "DeGAS: Gradient-Based Optimization of Probabilistic Programs without Sampling", presented at TACAS 2026.

- "P³MC: Towards Probabilistic Programming-Based Model Checking for Linear Temporal Logic", accepted at ISOLA 2026, soon to be presented.

Conferences & Community

I attended two conferences, OOPSLA and ETAPS, plus the UNRAVEL Symposium at RWTH Aachen.
These were great opportunities to see what the community is working on, cultivate ongoing collaborations and get feedback on my own work.

I also gave back some service to the community this year: I was program committee member for ATVA, QEST, WADT, and the OOPSLA Artifact Evaluation Committee. Moreover, I had the pleasure to co-organize the VeriProP workshop co-located with CAV at FLoC.

Collaborations

New collaborations took shape, within the TrustCPS group (leading to two papers currently under submission) and with Viennese groups working on probabilistic programming: Prof. Kovács's FORSYTE group and Prof. Chatterjee's group at IST Austria.
I presented my work at the biweekly IST–FORSYTE seminar, and attended several other talks in the series.
A stimulating environment, thriving with interesting inputs and ideas.

Mentoring & Leadership

I started supervising students at TU Wien: one B.Sc. student so far, with two M.Sc. theses in the pipeline. I also kept (co-)supervising my former Ph.D. student in Trieste. Small but necessary steps to learn the difficult art of supervising.

Soft Skills Training

Finally, I also had the opportunity to attend workshops organized by TU Wien on grant writing and open access publishing. While these topics may seem less relevant than the strictly scientific matters, they are fundamental tools in a researcher's toolkit, especially at my stage.

One year in, EMBEr already has a small but real footprint.
But the best is yet to come: in the next year I hope to reap the seeds I sowed in the past year and keep expanding my horizons.
Stay tuned for EMBEr's year 2! 🔥

15/04/2026

A piece of EMBEr at TACAS26

From April 12th to April 16th, TACAS (International Conference on Tools and Algorithms for the Construction and Analysis of Systems) was held in Turin as part of ETAPS (European Joint Conferences on Theory and Practice of Software).

Among the accepted papers, the first EMBEr-funded work was presented.

EMBEr focuses on testing moment-equivalence for probabilistic programs. A closely related problem is how moment information can be used effectively in core probabilistic programming tasks.

The paper “DeGAS: Gradient-based Optimization of Probabilistic Programs without Sampling” addresses this by extending a moment-matching approach for Gaussian mixtures, previously introduced for inference (see this), to optimization.

DeGAS relies on exact moment computation to build a differentiable, analytical approximation of the output distribution of a probabilistic program. This makes it possible to apply gradient-based optimization without sampling. The resulting closed-form representation also supports more complex objectives, such as constrained reachability probabilities. DeGAS is fully implemented in PyTorch, to leverage automatic differentiation and efficient GPU acceleration.

For a quick overview of DeGAS, you can look at our poster.
In alternative, you can read the full paper here or you can experiment with DeGAS here.