metrop_accept returns whether the sample was accepted #2058
Conversation
pymc3/step_methods/arraystep.py
Outdated
| @@ -6,7 +6,7 @@ | |||
| from numpy.random import uniform | |||
| from enum import IntEnum, unique | |||
|
|
|||
| __all__ = ['ArrayStep', 'ArrayStepShared', 'metrop_select', 'Competence'] | |||
| __all__ = ('ArrayStep', 'ArrayStepShared', 'metrop_select', 'Competence') | |||
There was a problem hiding this comment.
Why this? I'm confused :)
|
Ah shoot, I can remove that, though I think it is generally "better". My linter was complaining that when |
|
No that's a good reason! However for consistency we'd need to change ALL
the `__all__`
…On Wed, Apr 19, 2017 at 3:35 PM, Colin ***@***.***> wrote:
Ah shoot, I can remove that, though I think it is generally "better". My
linter was complaining that when __all__ is a list, it is mutable so it
is harder to tell what functions are exported.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#2058 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AA8DiJA1CfvGei6e-LnOVFBay7V7oFFLks5rxg2YgaJpZM4NBqFV>
.
--
Peadar Coyle
Skype: springcoilarch
www.twitter.com/springcoil
peadarcoyle.wordpress.com
|
|
I don't think it is a big deal either way. Actually, just found a surprisingly contentious issue in |
ColCarroll
force-pushed
the
metrop_accept
branch
from
9c96a75
to
93f33ee
Compare
|
@ColCarroll Nice refactor. Is acceptance already stored in the sampler stats? |
|
It doesn't look like any of the stats returns the acceptance ratio yet, but
that could be added.
…On Wed, Apr 19, 2017 at 9:56 AM Thomas Wiecki ***@***.***> wrote:
@ColCarroll <https://github.com/ColCarroll> Nice refactor. Is acceptance
already stored in the sampler stats?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#2058 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/ACMHEII4YbF3QBqCAUdyqwr_zVskOPE2ks5rxhKFgaJpZM4NBqFV>
.
|
|
Can do that separately and merge once tests pass. |
|
Do you want to add acceptance to the sampler stats in a separate PR before merging this? I think it stands on its own, but can do that. |
With the metropolis acceptance step, we were always (except in
HamiltonianMC) checking whether a proposal had been accepted just after returning. This makesmetrop_acceptreturn a tuple instead, and I think makes the code a little more readable.