Closed fritzo closed 3 years ago
great task list. i'm going to try to put together something so we can try larger enumeration windows
Re: splines, I think we will need different spline weights for differently sized windows. Cubic splines require at least four points. Each point added beyond four gives one new degree of freedom on the coefficients to be solved (the other three degrees of freedom are used by the boundary conditions). One natural way to constrain these extra degrees of freedom in spline coefficients is to minimize the sum of squares of cubic coefficients, i.e. a simple notion of maximally smooth.
Closing in favor of finer-grained issues.
This issue tracks tasks related to discrete compartmental models. See design doc for motivation.
Modeling
pyro.contrib.epidemiology
BetaBinomial
@fritzoCoalescentTimes
distributionComparmentalModel.transition_bwd()
#2468Incorporate phylogeny observations into the likelihood (see OverdispersedSEIRModel)Inference
SplitReparam
to split off low-frequency components of Haar reparamsBinomial
distributions overdispersedEngineering
--dct
.with_cache(1)
inDiscreteCosineTransform
(see https://github.com/pytorch/pytorch/pull/36882 )SafeLog
jit-script-compatible (see comment in hmm.py)Tidying up for first release DONE
#2518Add a notebook tutorial on regional models