[NeurIPS'25] Sequence Modeling with Spectral Mean Flows, in PyTorch

ScaledRelativeGraphs.jl — Computable SRG-based stability certificates for engineering practice. Bridges theory and application with standards compliance, measured-data workflows, and visualization. Julia.

Research project exploring operator-theoretic approaches to the Riemann Hypothesis, combining spectral theory and computational analysis. Includes English (main) and Portuguese versions.

SpectralNet v4: differentiable Chernoff–Remizov trajectories for inverse problems (1D Darcy). Companion code for Zenodo 10.5281/zenodo.19637120.

Publication list

Repository Of The seminar: Computational Dynamics

Operator-geometric framework for metabolic identifiability and metabolite panel design under partial observability using Human-GEM, proteomics-informed reaction weighting, and spectral geometry.

Operator-geometric inference of non-passive transport structure using PDE-constrained Hodge–Laplacian graph neural networks

SpectralNet v5: spectral evaluation of 2D Chernoff–Remizov trajectories for 2D Darcy inverse problems. Companion code for Zenodo 10.5281/zenodo.19664767.

The invariant subspace problem via operator-spectral persistence on the manifold-constrained canonical lane. Reproducible local-to-global theorem package.

A multidimensional jet space engine for high-order mixed partial Log-Tower derivatives. Features a Chronological Step-Operator Engine (Anchor/Web logic) that eliminates Bell polynomial explosion via asymmetric tensor caching. Replaces the quotient rule with an optimized derivative generator for lightning-fast symbolic mapping.

🔍 Verify and certify analytic bounds related to the Yang–Mills mass gap problem using reproducible computer-assisted scripts.