linear_operator icon indicating copy to clipboard operation
linear_operator copied to clipboard

A LinearOperator implementation to wrap the numerical nuts and bolts of GPyTorch

Results 31 linear_operator issues
Sort by recently updated
recently updated
newest added

# 🐛 Bug `torch.cat` fails for linear operators. ## To reproduce ** Code snippet to reproduce ** ```python from linear_operator.operators import DiagLinearOperator import torch D = DiagLinearOperator(torch.randn(2, 3, 100)) #...

bug

# 🚀 Feature Request Use a fix random seed for the tests. ## Motivation Hello! In a CI build of linear_operator, we are sometimes experiencing a build failure due to...

enhancement

With this change, the tolerance for diagonalization is also specified in `tolerances` instead of it being hardcoded. This unifies stuff a bit, and allows downstream tests to change the tolerance.

Previously, this method failed if `row_index` and `col_index` have different dtype/device. Now, we first unify these and do the comparisons afterwards. Differing dtype/device combinations can be created for example by...

# 🐛 Bug When using a KeOps kernel in GPyTorch, and making predictions (data larger than `settings.min_preconditioning_size`), the pivoted cholesky decomposition fails. This seems to be due to `covar_func` returning...

bug

Greetings, LinearOperator seems to require `typeguard~=2.13.3`, which is from December 2021 and typeguard is currently at v4.1.5. It would be nice to upgrade this dependency to something more recent. Thank...

This PR proposes an implementation of the non-symmetric Toeplitz linear operator (https://github.com/cornellius-gp/linear_operator/issues/76). It extends the actual ToeplitzLinearOperator with an additional constructor argument "row". If not supplied, the Toeplitz operation is...

Hi all, I've written a function to concatenate 2 square `LinearOperator` objects with different shapes: ``` from linear_operator.operators import ZeroLinearOperator, CatLinearOperator, KroneckerProductLinearOperator def cat_LinearOperators(*linear_operators): oper_A = linear_operators[0] oper_B = linear_operators[1]...

# 🐛 Bug When slicing the CatLinearOperator using a negative index, the final shape of the slice does not match the expected shape and an error is returned. This fails...

bug

### Discussed in https://github.com/cornellius-gp/linear_operator/discussions/78 Originally posted by **AndreaBraschi** September 13, 2023 Hi all, I've written a function to concatenate 2 square `LinearOperator` objects with different shapes: ``` from linear_operator.operators import...