-
-
Notifications
You must be signed in to change notification settings - Fork 2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add Differential Equation API #3590
Conversation
Check out this pull request on ReviewNB: https://app.reviewnb.com/pymc-devs/pymc3/pull/3590 You'll be able to see notebook diffs and discuss changes. Powered by ReviewNB. |
Could the test failure be related to this issue? #1640 |
FYI: Test pass now because I have included the |
I think that's a good choice. For reference, this means it might not work on GPUs -- I am not sure if that is an OK tradeoff for ODEs, but I do think such a failure should not block this PR. We should remember to open an issue once it is merged, though. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Left a few nitpicky things, and one musing suggestion. None of them are dealbreakers, and I will be delighted when this gets merged, even if none of the changes get made.
As another musing suggestion, running
python -m pylint --enable=all pymc3/step_methods/hmc/nuts.py
on each file you changed -- many suggestions will be silly, or ones we do not enforce, but some are helpful.
If you want to be really fancy
git diff --name-only HEAD master | xargs python -m pylint --enable=all
will pipe just the changed files through pylint
Thanks @ColCarroll, I think I added everything you suggested. |
Your module is currently not imported by default. Please add this after line 10 of
(just like |
As this is new code, can you just run black on the new files? |
@twiecki Ran black on all the new files. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@Dpananos running black
gave some nice results there! Can you also run it on test_ode.py
?
@twiecki @ColCarroll Are we ready to merge?
There are some improvements regarding the returned shape & the grad
that will follow in the near future, but the functionality is already there with this PR.
It seems like black was run on just ode.py, can you run it on all new files here? |
I ran black on the |
Just some minor comments on some code style. The notebook could also use spaces on the inline comments. I know its picky, but for me its jarring after seeing a bunch of PEP8 style comments. On the notebook the PyMC3 version is shown, just to double check will that be the version of PyMC3 that includes the ODE solver, or will we increment the PyMC3 version to 3.8 once this ODE solver get merged to master? |
Need to add to release-notes and link the new NBs to the docs. |
@twiecki Not sure how to update the docs, but I did find a reference to the other ODE notebook in @canyon289 Changes made and made the notebook more pep-8 friendly wrt spaces (I will never get used to one space after assignment, no space after arguments). If that's all, are we ready to merge? |
Co-Authored-By: Thomas Wiecki <[email protected]>
Co-Authored-By: Thomas Wiecki <[email protected]>
a compiled function which allows for computation of gradients of the | ||
differential equation's solition with repsect to the parameters. | ||
|
||
Args: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
These still have the old doc-format.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Merged and fixed on master.
Congrats again @Dpananos, this was a beast! |
This work is a part of Google Summer of Code 2019.
This PR adds:
ode
submodule, which containsDifferentialEquation
, the main tool for doing inference with ODEs.An
ODEGradop
for computation of the gradients required for HMC.A function for computing an augmented system of ODEs (that is, the original ODE + ODEs for the sensitivity equations) called
augment_system
.Unit tests for ode
A Notebook showcasing how to use
DifferentialEquation
in a couple of examples regarding a free falling object and a spread of an infection.Rerun existing ODE notebook for comparison
Sorry, I was having some issues with the last PR so I thought I would just start again.Here is a link to the old PR should you need to reference it.