François Pacaud

ExaModels is currently one of the fastest optimization modelers, and can evaluate (sparse) derivatives in parallel on various GPUs

For those interested, we presented yesterday at Powertech 2025 a comprehensive tutorial presenting the GPU-accelerated modeler ExaModels: frapac.github.io/tutorials/po...

Attention #JuliaLang users & scientific computing fans: #JuliaCon Local Paris 2025 is drawing closer 😱
- Talk submissions are still open (for 1️⃣ more week!)
- Tickets are now available (at early 🐦 prices)
Tell your colleagues, your neighbors, your grandma 👵 details at juliacon.org/local/paris2...

Fellow #JuliaLang enthusiasts in Europe, I've got 4 words for you.

JuliaCon Local Paris 2025

Yes, it's happening. Like the global @juliacon.bsky.social event, but with more croissants. Stay tuned!

Incredible: Chinese researchers using *domestic* chips and software optimization techniques get a ten-fold (!) boost over Nvidia's GPUs www.scmp.com/news/china/s...

The winter-school on numerical methods for nonsmooth optimal control is over. It has been a great pleasure to welcome Armin Nurkanovic at Mines Paris for a week!

All the materials are available here: cloud.minesparis.psl.eu/index.php/s/...

We have an open PhD position focused on advancing the interior-point method for stochastic programming. This
opportunity is ideal for students aiming to deepen their expertise in these areas. If you're interested, please don't hesitate to reach out for more details. frapac.github.io/pdf/thesis-i...

Interior Point Methods in Optimal Control Problems of Affine Systems: Convergence Results and Solving Algorithms | SIAM Journal on Control and Optimization Abstract. This paper presents an interior point method for pure state and mixed-constraint optimal control problems for dynamics, mixed constraints, and cost function all affine in the control variabl...

Nice initiative! Just a naive question regarding doi.org/10.1137/23M1.... You don't have to assume the strong Legendre-Clebsch condition to establish the convergence of your algorithm? What's the behavior of the interior-point method when the problem has singular arcs?

Winter school on numerical methods for optimal control of nonsmooth systems Monday, February 3rd to Wednesday, February 5th 2025 Speaker Dr. Armin Nurkanović https://www.syscop.de/people/armin-nurkanovic Place Mines Paris-PSL 60 boulevard Saint-Michel 75006 Paris Schedule...

We are inviting Armin Nurkanović to present a winter school on "Numerical methods for optimal control of nonsmooth systems" in February.

When? February 3rd to February 5th
Where? Mines Paris-PSL (downtown Paris)

Participation is free of charge!

More details in docs.google.com/forms/d/e/1F...

Nw! We are using double precision, we need that to get an accurate convergence in IPM. It turns out the modern GPUs are much performant with double precision than the previous generations.

In our experience we can achieve 1e-8 on medium instances, 1e-6 on large-scale ones. The condensed strategy induces an additional ill-conditioning that renders the method slightly less accurate than HSL solvers.

Yesterday, I had the chance to present our recent work at #CDC2024. The slides are available on my website: frapac.github.io/pdf/CDC2024....

"GPU-accelerated dynamic nonlinear optimization with ExaModels and MadNLP"

Deadline is November 29th. The link to submit a contribution is: roadef2025.org/soumettre-un...

With @gdalle.bsky.social we are organizing a session on "nonlinear optimization in very high dimension" at the French OR conference ROADEF 2025. We put a particular emphasis on optimization solver and software-based works, all contributions ranging in that span are welcome!

For CPU versus GPU, our main message is that you don't have to use an expensive GPU: I can get reasonable speed-up on the GPU I have on my laptop.

Good question. We did some benchmarks a while ago, we found out it's non-trivial to estimate the energy usage on the CPU. As such we just made a benchmark comparing our new method (HyKKT) versus our previous one (Biegler). Solving one OPF problem amounts to at most 500J using the new method.

NVIDIA cuDSS Library Removes Barriers to Optimizing the US Power Grid | NVIDIA Technical Blog In the wake of ever-growing power demands, power systems optimization (PSO) of power grids is crucial for ensuring efficient resource management, sustainability, and energy security.

I am glad to announce that our recent work on solving large-scale optimization problems on the GPU has been featured on NVIDIA's developer blog: developer.nvidia.com/blog/nvidia-...