ChaosCalib

ChaosCalib is a web app (with optional desktop agent) for calibrating nonlinear dynamical system models to messy real-world time-series. Users upload data, pick a model family (ODEs, discrete maps, delay systems), set constraints, and run parameter estimation with uncertainty quantification. It focuses on the unglamorous but common problem: you have a plausible nonlinear model, but you can’t reliably fit it, compare alternatives, or quantify identifiability. The app provides automated diagnostics (sensitivity, sloppiness, bifurcation proximity warnings), model comparison, and reproducible reports. It is a combination traditional + AI app: traditional numerical optimization and simulation, plus an AI assistant that suggests priors, detects data issues (drift, missingness), and proposes candidate model structures from a library (not “invent physics”). It’s aimed at applied labs and teams who need defensible fits, not pretty phase portraits.