DiffEqBayes.jl

DiffEqBayes.jl copied to clipboard

Running the code in the [Readme](https://github.com/SciML/DiffEqBayes.jl) gives me a Stan model with two sigma parameters, but a Turing model with a single sigma parameter.

nsiccha

Compare https://github.com/SciML/DiffEqBayes.jl/blob/c983590f6a8e70346dbc52ff6f83a81e0019fafe/src/stan_inference.jl#L58 and https://github.com/SciML/DiffEqBayes.jl/blob/c983590f6a8e70346dbc52ff6f83a81e0019fafe/src/turing_inference.jl#L8 which implies either an InverseGamma(2, 3) or an InverseGamma(3, 3) prior.

nsiccha

CompatHelper: bump compat for DocStringExtensions to 0.9, (keep existing compat)

1

This pull request changes the compat entry for the `DocStringExtensions` package from `0.8` to `0.8, 0.9`. This keeps the compat entries for earlier versions. Note: I have not tested your...

github-actions[bot]

abc_inference is not included but exported

1

Hello, it seems that new version doesn't support approximate bayesian computation any more. However, in main `DiffEqBayes.jl`, `abc_inference` is exported. Also, codes of example of README implies that `abc-inference` depends...

Song921012

Since this is very visible, we should show people towards the Turing one which is less visible from the SciML docs.

ChrisRackauckas

Change ABC inference engine?

2

Since, [ApproxBayes.jl](https://github.com/marcjwilliams1/ApproxBayes.jl) was integrated here there are a few more ABC engines available that are more fully-featured and likely better supported at present. I'm not using so much ABC in...

marcjwilliams1

https://github.com/SciML/DiffEqBayes.jl/blob/ea7fa7c623eb94b8851332bc09d38499f2e83fe2/src/DiffEqBayes.jl#L21-L25 Similar to the LinearSolve.jl formulation.

ChrisRackauckas

Turing inference BoundsError when save_idxs is single value

4

I get a BoundsError when calling `turing_inference` when `save_idxs` is a single digit. A minimum code for this from the tutorial: ```julia using ParameterizedFunctions, OrdinaryDiffEq, RecursiveArrayTools, Distributions using DiffEqBayes f1...

brendandaisy

Possible regression in Turing integration

2

The CI seems to be taking unreasonably long to run through the turing_inference tests, probably something changed in newer releases. Should get bisected and reported

Vaibhavdixit02

Use updated `build_function`

3

Looks like Symbolics has deprecated use of `build_function(

Vaibhavdixit02