iiinova — where your digital self keeps evolving

Reposted by Tim Weiland

Compute! Paris @computeparis.bsky.social · 23/09/2026

Stop Bruteforcing your Posteriors! Exploit known analytical structure! That's the title and pitch of Tim Weiland's talk at Compute! Paris this November.

111

Tim Weiland @timwei.land · 30/06/2026

Validated against R-INLA on matched models, scored by how closely the marginals agree, every figure linked to a runnable script. It's v0.1 and the API's still moving. Feedback and bug reports are very welcome! Check it out: lattejl.org

000

Tim Weiland @timwei.land · 30/06/2026

Write the model once with the `@latte` macro, then pick the engine: INLA (quadrature over hyperparameters, + some tricks), TMB (MAP + Gaussian cov), or NUTS on the Laplace marginal. Same model, language across all three.

110

Tim Weiland @timwei.land · 30/06/2026

Last week, I released Latte.jl 🎉 It's a probabilistic programming framework for latent Gaussian models in Julia: spatial disease maps, time trends, hierarchical GLMMs, GPs on a mesh, and more. Recognizing the structure of these models gives you access to fast inference algorithms. `] add Latte`

210

Tim Weiland @timwei.land · 24/04/2025

Love the skeleton marketing

120

Reposted by Tim Weiland

Frank Schneider @fsschneider.bsky.social · 16/04/2025

Tired of your open-source ML work not getting the academic recognition it deserves? 🤔 Submit to the first-ever CodeML workshop at #ICML2025! It focuses on new libraries, improvements to established ones, best practices, retrospectives, and more. codeml-workshop.github.io/codeml2025/

codeml-workshop.github.io

CODEML Workshop

Championing Open-source Development in Machine Learning.

0346

Reposted by Tim Weiland

Sung Kim @sungkim.bsky.social · 31/03/2025

You’ve probably heard about how AI/LLMs can solve Math Olympiad problems ( deepmind.google/discover/blo... ). So naturally, some people put it to the test — hours after the 2025 US Math Olympiad problems were released. The result: They all sucked!

917350

Tim Weiland @timwei.land · 17/03/2025

Excited to mention that this work was accepted to AISTATS 2025. Shout-out to my amazing collaborators Marvin Pförtner & @philipphennig.bsky.social!

020

Tim Weiland @timwei.land · 17/03/2025

In summary: The SPDE perspective brings us more flexible prior dynamics and highly efficient inference mechanisms through sparsity. 📄 Curious to learn more? Our preprint: arxiv.org/abs/2503.08343 💻 And there's Julia code: github.com/timweiland/G... & github.com/timweiland/D... 8/8

arxiv.org

Flexible and Efficient Probabilistic PDE Solvers through Gaussian Markov Random Fields

Mechanistic knowledge about the physical world is virtually always expressed via partial differential equations (PDEs). Recently, there has been a surge of interest in probabilistic PDE solvers -- Bay...

110

Tim Weiland @timwei.land · 17/03/2025

So here's the full pipeline in a nutshell: Construct a linear stochastic proxy to the PDE you want to solve -> discretize to get a GMRF -> Gauss-Newton + sparse linear algebra to get a posterior which is informed about your data and the PDE. 7/8

A diagram describing our method. You start with some PDE you want to solve. Then, you construct a linearized SPDE that closely resembles the dynamics of this PDE. Discretizing this SPDE (through FEM along space and some e.g. implicit Euler scheme along time) yields a GMRF prior with a sparse precision matrix. Then, you condition this prior on your (noisy) observations, e.g. the initial condition. Afterwards, you perform Gauss-Newton steps until convergence to integrate the (nonlinear) information about the PDE discretization. At convergence, you do Laplace to finally obtain a GMRF posterior over the solution of the PDE.

110

Tim Weiland @timwei.land · 17/03/2025

Turns out that these "physics-informed priors" indeed then converge much faster (in terms of the discretization resolution) to the true solution. 6/8

A plot comparing the two different priors we mentioned before for the task of solving a Burgers' equation. The x axis is the number of collocation points, and the y axis is the relative error in percentage. We observe that the advection-diffusion prior achieves a very low relative error already for a low number of collocation points, while the error drops off much more slowly for the Matérn-based prior.

100

Tim Weiland @timwei.land · 17/03/2025

But wait a sec... With this approach, we express our prior belief through an SPDE. Then why should we use the Whittle-Matérn SPDE? Instead, why don't we construct a linear SPDE that more closely captures the dynamics we care about? 5/8

A diagram comparing the effect of different priors for probabilistic numerical PDE solvers. For the task of solving a Burgers' equation, we compare a product of 1D Matérns with a prior constructed from an advection-diffusion SPDE. Both priors are conditioned only on the initial condition, and then we plot for two different time points how the dynamics evolve. We see that for the Matérn-based prior, the effect of the observations simply "dies out" over time. By contrast, the advection-diffusion prior actually propagates the observations through time in a way that much more closely resembles the Burgers' equation dynamics we want to approximate.

100

Tim Weiland @timwei.land · 17/03/2025

Still with me? So we get a FEM representation of the solution function, with stochastic weights given by a GMRF. Now we just need to "inform" these weights about a discretization of the PDE we want to solve. These computations are highly efficient, due to the magic of ✨sparse linear algebra✨. 4/8

A screenshot of equations that express the posterior of a Gaussian Markov Random Field under affine observations y = Au + b.

110

Tim Weiland @timwei.land · 17/03/2025

Matérn GPs are solutions to the Whittle-Matérn stochastic PDE (SPDE). In 2011, Lindgren et al. used the finite element method (FEM) to discretize this SPDE. This results in a Gaussian Markov Random Field (GMRF), a Gaussian with a sparse precision matrix. 3/8

A screenshot of the equation for the Whittle-Matérn stochastic PDE.

100

Tim Weiland @timwei.land · 17/03/2025

In the context of probabilistic numerics, people have been using Gaussian processes to model the solution of PDEs. Numerically solving the PDE then becomes a task of Bayesian inference. This works well, but the underlying computations involve expensive dense covariance matrices. What can we do? 2/8

100

Tim Weiland @timwei.land · 17/03/2025

⚙️ Want to simulate physics under uncertainty, at FEM accuracy, without much computational overhead? Read on to learn about the exciting interplay of stochastic PDEs, Markov structures and sparse linear algebra that make it possible... 🧵 1/8

A plot comparing the performance of different methods to numerically solve PDEs. The x-axis depicts the compute time in seconds, the y-axis depicts the relative error to the ground-truth solution in percentage. The graph for the "standard" finite element method has the steepest downward slope and thus the best performance. Our methods based on Gaussian Markov Random Fields achieve the same accuracies as the finite element method, but at slight computational overheads, depicted by their graphs having a slightly flatter slope. For instance, the highest accuracy solves require ~8-9 seconds for FEM and ~25 seconds for our GMRF-based method.

1112

Reposted by Tim Weiland

Motonobu Kanagawa @motonobu-kanagawa.bsky.social · 11/02/2025

The submission site for ProbNum 2025 is now open! The deadline is March 5th. We welcome your beautiful work on probabilistic numerics and related areas! probnum25.github.io/submissions

probnum25.github.io

ProbNum25 : Submissions

Coming

0115

Tim Weiland @timwei.land · 30/12/2024

My recs: Doom emacs to get started; org mode + org-roam for notes, org-roam-bibtex + zotero auto-export for reference management; dired for file navigation; tramp mode for remote dev; gptel for LLMs; julia snail for julia, make sure you set up lsp. Start small and expand gradually, see what sticks

120

Tim Weiland @timwei.land · 03/12/2024

Amazing work! A big physics-informed ML dataset with actually relevant problems, created in cooperation with domain experts. It‘s time to finally move on from 1D Burgers‘ equations 🚀

051

Tim Weiland @timwei.land · 19/11/2024

I would also like to be added :) Great idea, thanks for this!

120

Tim Weiland @timwei.land · 19/11/2024

ELLIS PhD student here, I would also appreciate getting added :)

120