Modify linear regressions to parameters in R

[albcab]

Closes #378

As a side note: If initial parameters for the log_scale are set to $1.0$ in regression_test_cases only test_pathfinder_adaptation fails with samples for log_scale diverging to inf.

[codecov]

Codecov Report

Merging #388 (3a05028) into main (1aaa6f6) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main     #388   +/-   ##
=======================================
  Coverage   89.74%   89.74%           
=======================================
  Files          45       45           
  Lines        2184     2184           
=======================================
  Hits         1960     1960           
  Misses        224      224           

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[rlouf]

Thanks! Why use flat priors?

[junpenglao]

I think it is better to keep the prior and add the forward log jacobian. @rlouf wdyt?

[rlouf]

I think it is better to keep the prior and add the forward log jacobian. @rlouf wdyt?

We had the same thought at the same time 😄

[junpenglao]

Could you also update the one in:

• tests/test_optimizers.py
• tests/test_benchmarks.py
• tests/test_tempered_smc.py

Optional the ones in examples.

[junpenglao]

Looks good, minor comments.

[albcab]

Great. Added a reminder in the introduction notebook to reparametrize before MCMC.

[rlouf]

LGTM, thank you for doing the unsexy work!