DigitalTwin¶

A digital twin is a simulation that has been fitted to a real system and knows how far it can be trusted. These examples use the fitting and twin plugins on a synthetic lab with a hidden truth, so every claim can be checked against the answer.

example	the story
transforms	the original fitting example: a transform (residual dispersion, gain narrowing, post-pulse, ASE pedestal) fitted to a spectrum, d-scan, cross-correlator and power meter; the fitted line on a new seed
build-the-twin	fit spectra at five energies with one set of shared parameters, recover the hidden values, predict two held-out energies with credible bands; a validation report that catches a twin that only fits its training data
explore-the-twin	what-if comparisons with bands, the most informative next measurement, where to put the spectrometer, drift tracking with an ensemble Kalman filter

Run: python docs/examples/make_all.py --group DigitalTwin. Plugin pages: fitting, twin.

Not here yet: a comparison of a spline and a network transform (the network transform exists and is tested in the fitting plugin), a wavefront twin (needs a spatial project template that does not exist yet), and agent-driven design through an MCP server (the MCP server is not built; the design is in the design record).