A Python package for carrying out some analysis of "typical"1 mathematical models, including:
- parameter estimation
- sampling of parameter posterior distribution
- model reduction
- model selection
The frameworks and methods here follow the treatments of information geometry and its extension information topology2, hence the name of the package.
Note 1: The technical definition of "typical" here is any mathematical model whose predictions are differentiable with respect to parameters, which includes most models in physical sciences.
Note 2: See http://doi.org/10.1103/PhysRevE.83.036701 and http://arxiv.org/abs/1409.6203.
mod = Model()
expts = Experiments()
pred = mod.get_pred(expts)
s = pred.get_spectrum()
dat = pred.get_data()
res = residual.Residual(pred, dat)
fit = fitting.leverberg_marquardt(res, p0)
ens = sampling.sampling(res, p0, nstep=100)
ens.scatterplot()
gds = pred.get_geodesic()
gds.integrate()
gds.plot()