Add a quadrature rule for higher degree polynomials #72

btalamini · 2023-12-03T15:12:48Z

Add a 25 point rule for the triangle that exactly integrates polynomials up to degree 10.
Improve error messaging. Before, if the user asked for a quadrature rule of higher precision than we have implemented, they would see UnboundLocalError: local variable 'xi' referenced before assignment. Now they get ValueError: Quadrature of precision this high is not implemented.
Improved the test fixture for the quadrature rules: one of the tests loops over degrees of precision, ensuring that the quadrature rule returned can actually exactly integrate polynomials of that degree. Before, if any case failed, the test reported a single failure and nothing else, leaving you to guess which quadrature rule(s) were broken. Now the subTest facility of unittest (or pytest) is used so that you see which precision checks failed.

… module

… tests

…t matrix log, exp, and sqrt

…actly symmetric

Before: the primal output was computed in-line in the custom jvp function. The advantage is that this avoids a second call to the eigendecomposition. The downside is that this in-line computation doesn't itself have a custom jvp, so its derivative can be wrong. After: I re-compute the primal value through the base function (e.g., log_symm), which has the custom jvp defined on it. The eigendecomposition is repeated. We can refactor to eliminate this later if profiling reveals it to be a performance bottleneck.

Put a leading underscore on functions menat for internal use. Most Python tools will ignore these when reporting contents of a module.

Replace all calls except in the new viscoelastic model. Changes are about to merge there and I want to handle the conflicts separately.

Also: now a ValueError is emitted if a user asks for a quadrature rule precision that is above what we have implemented. Before the error was a bit obscure. Improved the unit testing of the quadrature rules by making subtests for every polynomial degree check. This way, if the test fails, you can figure out which quadrature rules are broken.

btalamini added 16 commits November 19, 2023 07:05

Remove gneral matrix routines from tensor math library and put in own…

ccacb4f

… module

Clean up format and give a few operations better names

046b821

Give another tensor operator a more succinct and common name

144b310

Add high precision determinant plus identity function and determinant…

48169d5

… tests

Add polar decomposition and matrix square root

da454de

Add way to make a tensor function out of any scalar function, implmen…

e257f49

…t matrix log, exp, and sqrt

Write more comments

50e8b7f

Sort eigenvalues in log relative difference functino so that it is ex…

f827374

…actly symmetric

Add test that shows jvp of matrix power function is wrong

70530e9

Add a new matrix power function with a correct gradient

1652627

Remove the broken mtk_power function

46bef82

Hide internal helper functions

fa2f3e1

Put a leading underscore on functions menat for internal use. Most Python tools will ignore these when reporting contents of a module.

Remove mtk_log in favor of new implementation

0c84e55

Replace all calls except in the new viscoelastic model. Changes are about to merge there and I want to handle the conflicts separately.

Comments and formatting

d81daee

btalamini requested a review from tupek2 December 3, 2023 15:12

btalamini closed this Dec 3, 2023

btalamini deleted the feature/quadrature_precision_10_rule branch December 3, 2023 15:19

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add a quadrature rule for higher degree polynomials #72

Add a quadrature rule for higher degree polynomials #72

btalamini commented Dec 3, 2023

Add a quadrature rule for higher degree polynomials #72

Add a quadrature rule for higher degree polynomials #72

Conversation

btalamini commented Dec 3, 2023