Numpy docs hmc #2405
Numpy docs hmc #2405
Conversation
|
(also, removing attribution to @jsalvatier is not meant as a slight -- I think the git history provides a record of credit, and this is interpreted by numpy as a malformed module docstring.) |
| @@ -3,3 +3,7 @@ testpaths = pymc3/tests | |||
|
|
|||
| [coverage:run] | |||
| omit = *examples* | |||
|
|
|||
| [pydocstyle] | |||
| add-ignore = D100,D104 | |||
There was a problem hiding this comment.
I'll add a comment -- D100 wants a docstring at the top of every file, and D104 wants a docstring in every __init__.py file
pymc3/step_methods/hmc/hmc.py
Outdated
| @@ -22,11 +18,17 @@ def unif(step_size, elow=.85, ehigh=1.15): | |||
|
|
|||
|
|
|||
| class HamiltonianMC(BaseHMC): | |||
| R"""A sampler for continuous variables based on Hamiltonian mechanics. | |||
|
|
|||
| See NUTS sampler for automatically tuned stopping time. | |||
There was a problem hiding this comment.
stopping time and step size adaptation.
pymc3/step_methods/hmc/nuts.py
Outdated
| @@ -90,7 +91,8 @@ class NUTS(BaseHMC): | |||
| def __init__(self, vars=None, Emax=1000, target_accept=0.8, | |||
| gamma=0.05, k=0.75, t0=10, adapt_step_size=True, | |||
| max_treedepth=10, on_error='summary', **kwargs): | |||
| R""" | |||
| R"""Set up the No-U-Turn sampler. | |||
There was a problem hiding this comment.
The standard says that we shouldn't document the arguments in __init__ and in the class doc string.
https://github.com/numpy/numpy/blob/master/doc/HOWTO_DOCUMENT.rst.txt#class-docstring
There was a problem hiding this comment.
It looks like most of the class doc string is documenting the sampler_stats return object -- I hadn't seen that there is an Attribute section for class docstrings though, which is probably where the sampler stats should go...
There was a problem hiding this comment.
Actually, that is a funny thing to document in __init__ since it is an attribute it provides on the trace object. It makes sense to me to leave it where it is, and not adding an Attribute section.
|
Thanks for writing all those doc strings! |
|
Don't we want new line after """? |
| Created on Mar 7, 2011 | ||
|
|
||
| @author: johnsalvatier | ||
| ''' | ||
| from ..arraystep import metrop_select, Competence | ||
| from .base_hmc import BaseHMC | ||
| from pymc3.vartypes import discrete_types | ||
| from pymc3.theanof import floatX |
There was a problem hiding this comment.
Numpy should be imported before local imports by pep8
There was a problem hiding this comment.
I updated this here -- it is actually a pylint check (wrong-import-order) that I can maybe try adding tomorrow, but that is a ton of violations!
ColCarroll
force-pushed
the
numpy_docs_hmc
branch
from
1756260
to
1259063
Compare
ColCarroll
force-pushed
the
numpy_docs_hmc
branch
from
1259063
to
ee199ef
Compare
|
Going to merge these once tests pass, since the rebasing is a pain and just making sure that the docs follow the functions around! |
|
This is great @ColCarroll. |
* Conform to numpy style docstrings in hmc submodule * Accidentally messed with pylintrc * Comments * Rebase * Rebase again
As was foretold in #2403
@aseyboldt you might take particular offense to some of the docstrings -- would love any nitpicks.