calibration.py — Parameter Fitting and MCMC¶
This module implements optimization-based and Bayesian parameter estimation.
Functions¶
loss_function(theta, model, sampled_times, observed_data, fit_compartments)¶
Computes the loss between simulated and observed data for selected compartments.
Parameters
- theta (list[float]) — Parameter vector.
- model (CompartmentalModel) — Model to simulate.
- sampled_times (list[float]) — Time points used for fitting.
- observed_data (np.ndarray) — Observed noisy data.
- fit_compartments (list[str]) — Compartments included in the loss.
fit_with_optimizers(model, optimizers, ...)¶
Runs selected optimizers (e.g., Nelder-Mead, BFGS, L-BFGS-B) to fit parameters.
run_mcmc(model, ...)¶
Uses the emcee sampler to estimate parameter posterior distributions.
Workflow 1. Define prior bounds. 2. Initialize walkers inside bounds. 3. Simulate the model and compute log-likelihood. 4. Save and plot posterior samples.
Example Usage¶
```python
from calibration import fit_with_optimizers
results = fit_with_optimizers(model, ["BFGS", "Nelder-Mead"], ...)
```