Neural Cartographer: Recovering Invisible Laws express

Original: "RTS Smoother-Guided Learning of Physics-Based Neural Differential Models"

· Ahmet Demirkaya, Georgios Stratis, Tales Imbiriba, Zachary D. Danziger, Deniz Erdogmus

arXiv:2607.15180 · 2026-07-16 · CC BY 4.0 · 1 min

A hybrid method recovers hidden dynamics, like a cartographer creating an accurate map from fragmentary landmarks.

Many natural processes are described by equations, but we only have fragments of them. The new method acts like a cartographer who reconstructs a complete map of the terrain from a few landmarks. Known laws provide a reliable standard framework — like a coastline, while the neural network draws the hidden dynamics by analyzing data. With each cycle, the algorithm refines the map: first it reconstructs the system's full history, then it corrects the unknown parts. This is reminiscent of how astronomers map invisible dark matter in galaxies by its gravitational influence.

Reducing uncertainty is a fight against entropy. Each pass of the method reduces chaos, revealing the true structure, just as shading on a map reveals the relief.

Tests showed: the method accurately predicts behavior even when only 5% of all factors are visible. Unlike black boxes, it retains physical meaning, which is crucial for reliable models in medicine and ecology.

Whole graph →

A similar principle is at work in self-driving cars: when sensors momentarily lose sight of an object, a physical model plots its likely path, and a neural network refines the details, just as a cartographer draws a map from a few lines.

Scientists

Fritz Zwicky Vera Rubin Roberto D. Peccei Helen Quinn R. Brent Tully

Concepts

neural network regression stochastic differential equations smoothed particle hydrodynamics Stokes' theorem delay differential equation stochastic dynamical system empirical orthogonal function neural network physics modeling optically detected magnetic resonance self-supervised learning regularization methods finite element method Lotka-Volterra equation lattice simulation robust machine learning mathematical model Navier-Stokes equations physics informed neural network ornstein uhlenbeck process Stochastic modeling

I am the author

Neural Cartographer: Recovering Invisible Laws express