iiinova — where your digital self keeps evolving

Thiparat Chotibut @thipchotibut.bsky.social · 16/12/2025

[n/n] Check out the paper for the full derivation and discussion! arxiv.org/abs/2512.12767 Happy to chat more about sparse, non-hermitian random matrices, attractors, or why physics needs to pay more attention to timescale heterogeneity. ⏳⌛️⏲️

arxiv.org

Random matrix theory of sparse neuronal networks with heterogeneous timescales

Training recurrent neuronal networks consisting of excitatory (E) and inhibitory (I) units with additive noise for working memory computation slows and diversifies inhibitory timescales, leading to im...

010

Thiparat Chotibut @thipchotibut.bsky.social · 16/12/2025

[8/n] Takeaway: By adding realistic complexity (sparse, structured interactions and timescale heterogeneity), new functional mechanisms appear that were once invisible to standard RMT results. It’s a messy math problem, but the SUSY tools make it solvable.

100

Thiparat Chotibut @thipchotibut.bsky.social · 16/12/2025

[7/n] With this technique, we can also reproduce the classic Rajan-Abbott distribution in an economical manner, see Appendix C.

100

Thiparat Chotibut @thipchotibut.bsky.social · 16/12/2025

[6/n] To prove this, we had to solve for a sparse, non-Hermitian random matrix problem, modulated by heterogeneous timescales. . Standard RMT tools fail here. . We adapted techniques from High Energy Physics, using a SUSY-based approach to analytically solve the spectral edge.

100

Thiparat Chotibut @thipchotibut.bsky.social · 16/12/2025

[5/n] We found that the emergent "inhibitory core-excitatory periphery" architecture, coupled with a broad distribution of inhibitory timescales, pushes the spectrum closer to criticality. No weight fine-tuning required!

100

Thiparat Chotibut @thipchotibut.bsky.social · 16/12/2025

[4/n] For those working on critical phenomena, this seems impossible: achieving criticality usually requires precise fine-tuning. So, how does the cortical network do this? . Our proposed solution: Sparsity + Structure + Timescale Heterogeneity 💡

100

Thiparat Chotibut @thipchotibut.bsky.social · 16/12/2025

[3/n] In this new preprint, we have some answers. Here’s the physics problem: The working memory computation reported in PNAS relies on a "discrete attractor hopping" mechanism. To encode memory robustly, each discrete attractor needs to operate near the "edge-of-chaos."

100

Thiparat Chotibut @thipchotibut.bsky.social · 16/12/2025

[2/n] It's a follow-up from our PNAS reporting a paradoxical finding: neuronal networks actually perform working memory tasks better under driven noise. Back then, we reported that it worked, but we didn’t really understand why and how. www.pnas.org/doi/10.1073/...

100

Thiparat Chotibut @thipchotibut.bsky.social · 16/12/2025

[1/n] Happy to share exciting results, combining a rare mix of theoretical physics tools to answer a fundamental question in temporal information processing: . How do cortical networks operate near criticality with improved working memory without requiring precise parameter fine-tuning? 🧠⚛️

110

Thiparat Chotibut @thipchotibut.bsky.social · 06/07/2025

Curious about the scalability of analog VQAs that harness particular quantum phases of matter as an ansatz ⚛️? Check out our new study, led by a talented master’s student Kasidit - nice summary thread below. 👇

010

Thiparat Chotibut @thipchotibut.bsky.social · 01/07/2025

🧵 4/4 To stat mech crowd: think of congestion games as out-of-equilibrium many-body active matter. . This is an exactly solvable active system (multi-agent RL) where microscopic chaos coexists with macroscopic ergodic convergence - check it out! www.pnas.org/doi/10.1073/...

pnas.org

PNAS

Proceedings of the National Academy of Sciences (PNAS), a peer reviewed journal of the National Academy of Sciences (NAS) - an authoritative source of high-impact, original research that broadly spans...

000

Thiparat Chotibut @thipchotibut.bsky.social · 01/07/2025

🧵 3/4 ✨ Remarkably, yet the long-run average number of agents on route 1 settles on the social-optimum / Nash equilibrium (bottom right) ⛳️, despite the day-to-day head-count of route 1 being provably chaotic (bottom left)! 🌪️

120

Thiparat Chotibut @thipchotibut.bsky.social · 01/07/2025

🧵 2/4 Results: When some agents learn (adapt) very fast, their individual strategies turn chaotic 🌪️. Top panel - x axis: agent type with different learning rates, y-axis fraction of that agent selecting route 1.

100

Thiparat Chotibut @thipchotibut.bsky.social · 01/07/2025

🧵 1/4 Ever wondered how tools from statistical physics can help understand learning in diverse reinforcement-learning populations? Check out our new PNAS paper (Special Feature: Collective Artificial Intelligence & Evolutionary Dynamics) here pnas.org/doi/10.1073/... #PNASNews

120

Thiparat Chotibut @thipchotibut.bsky.social · 25/06/2025

Check out our latest work spearheaded by a talented master student, Kasidit. See a summary thread below 👇

010

Thiparat Chotibut @thipchotibut.bsky.social · 17/05/2025

Wondering whether simply scaling up system sizes and cranking up chaos really supercharges a quantum reservoir, or if dissipation is really a resource at scale? 🌀 We demystify these QRC myths in our new preprint: [scirate.com/arxiv/2505.10080] 🧵 Key takeaways in a great thread below!👇

021

Thiparat Chotibut @thipchotibut.bsky.social · 18/01/2025

Published in #PNAS 🎉 Noise isn't just disruptive; it can enhance neural computations, especially in working memory tasks! Biologically plausible RNNs harnessing noise also operate near the “edge of chaos,” supporting the critical brain hypothesis 🧠✨ Check out 👇 www.pnas.org/doi/10.1073/...

010

Thiparat Chotibut @thipchotibut.bsky.social · 16/01/2025

Kudos to the team (especially to Tang and Teerachote) for this computational feat powered by thousands of NVIDIA GPU hours and LOADs of trials and errors! But eventually they succeeded at rivaling state-of-the-art models! Feedbacks are welcome! Paper --> arxiv.org/abs/2501.08998

arxiv.org

000

Thiparat Chotibut @thipchotibut.bsky.social · 16/01/2025

We show that CrystalGRW yields stable, unique, and novel structures (S.U.N. materials) close to their DFT ground states. The fun part for me is to revisit the theory of random walks on Riemannian manifolds and make this works for generative modeling.

100

Thiparat Chotibut @thipchotibut.bsky.social · 16/01/2025

If you’re interested in materials discovery or generative modeling, CrystalGRW might cut down the guesswork and skip expensive ab initio calculations and also let you specify, say, a target crystallographic point group or composition right off the bat.

100

Thiparat Chotibut @thipchotibut.bsky.social · 16/01/2025

The coolest part (in my humble opinion) is how it balances crystal symmetry requirements, periodicity, compositional constraints, and training stability in a single, unified generative modeling framework: diffusion models on natural Riemannian manifolds that suitably represent crystal properties

100

Thiparat Chotibut @thipchotibut.bsky.social · 16/01/2025

Hey everyone, just wanted to share our new preprint on crystal structure generation (CrystalGRW) is live on arXiv! We’ve been playing with diffusion-based models that treat crystal structures in their “natural domain” (Riemannian manifolds) (like Torus for coords capturing periodicity) and it works!

100