-
Notifications
You must be signed in to change notification settings - Fork 3
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
Adding ability to define test space through convex hull #18
Conversation
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.
@andersonw1 , most of them look good to me, except the use of getParam1DSpace
. I think we can simply use the existing getParam1DSpace
functions, not creating a new one. I left some comments in details related to that.
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.
Everything looks good to me. I left a minor comment on type specification for gridSizes
.
src/lasdi/param.py
Outdated
|
||
Returns | ||
------- | ||
gridSizes : :obj:`list(Nx)` |
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.
can we use list(int)
instead of list(Nx)
?
src/lasdi/param.py
Outdated
|
||
Returns | ||
------- | ||
gridSizes : :obj:`list(Nx)` |
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.
Same as above, can we use list(int)
instead of list(Nx)
?
This PR adds two things:
We introduce the ability to define training points on the exterior some region of interest. To define the testing points, we first form a uniform grid over the training parameters and then remove all testing points which are not in the convex hull of the training parameters.
This adds the new 'exterior' option for the test_space_type and requires that all of the training points are given as lists. This is useful if we are not interested in sampling from a rectangular grid, and instead have a specific region of interest for the problem. The post-processing files need to be changed.
We now track the training loss and each term of the loss function. These losses are saved with the final output file after training.