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
further numpy type conversions #6506
Comments
This comment has been minimized.
This comment has been minimized.
comment:2
Attachment: 6506-numpy-types.patch.gz Carl, as one of the resident experts on precision issues, could you glance at the last part of this patch, which touches real_mpfr.pyx and seems to deal with precision issues? |
apply on top of previous patch |
comment:3
Attachment: trac-6506-doc-fixes.patch.gz I posted a patch which fixes lots of the documentation errors on #5081. There are still some problems.
|
comment:4
See the documentation patch (the second patch above) for changes to the totallyreal_rel.py file, including a comment about where a bug probably is. |
Reviewer: Jason Grout |
Author: Robert Bradshaw, Jason Grout |
comment:5
I should mention that I'm not qualified to correct the code in totallyreal_rel.py, since I don't know which i is from the comprehension and which i is from the outer loop. |
comment:6
Jason, your code looks good to me. I've fixed totallyreal_rel.py--I don't totally understand the code/error either but did extensive comparison with the original code. I also fixed the remaining precision issues--the missing 0 is due to the fact that the decimal point is no longer considered a significant digit. It'd be nice to get this into 4.1.1. |
comment:7
Attachment: 6506-numpy-types-fixes.patch.gz All doctests pass now. I assume your comments mean that you are giving a positive review to my changes in totallyreal_rel.py (correct me if I'm wrong!). Doctests now pass in totallyreal_rel.py and the other files mentioned above, so I'm changing this to a positive review and marking you as a reviewer too. I agree that it would be very, very nice to get this into 4.1.1, especially considering how many times it's come up on the mailing lists just in the past week. |
Changed reviewer from Jason Grout to Jason Grout, Robert Bradshaw |
comment:8
When I say above that all doctests pass, I mean the doctests pointed out in my comment above. I assume that when this patch is merged, all doctests will be run and any errors will kick this ticket back to "needs work". |
comment:9
I applied patches in this order:
|
comment:10
I believe these are due to the correction of real number precision, not numerical noise. They all look fine to me--I'll try and post a patch later today. |
comment:11
The above issues are resolved. |
comment:12
With the patch at #5081 and the patches on this ticket, I still get a doctest failure:
|
comment:13
Attachment: numpy-fixes-3.patch.gz Oh, somehow I missed that one. I've refreshed the patch. |
comment:14
Merged all four patches. |
Changed reviewer from Jason Grout, Robert Bradshaw to Jason Grout, Robert Bradshaw, Minh Van Nguyen |
Merged: Sage 4.1.1.rc0 |
comment:16
What's this?
The "57" makes no sense to me. How is |
comment:17
The "57" has been fixed in #19758. |
Followup to #5081, depends on Python 2.6
Better handling of ZZ, QQ and CC now supported.
CC: @sagetrac-cwitty
Component: numerical
Author: Robert Bradshaw, Jason Grout
Reviewer: Jason Grout, Robert Bradshaw, Minh Van Nguyen
Merged: Sage 4.1.1.rc0
Issue created by migration from https://trac.sagemath.org/ticket/6506
The text was updated successfully, but these errors were encountered: