-
Hi,
I'm trying out pyviennacl, it sounds great so far. I can run some test code, so the setup seems to be ok. (Py 2.7 on 64bit Windows 7).
But when I try pyviennacl.linalg.solve() I'm getting an erro…
-
@simone-silvestri and I have discussed a number of improvements to solvers (mostly refactoring and cleanup); I'm opening this issue to record our discussion.
Here are some items in no particular or…
-
_first some motivation:_
I have a __hermitian__ linear operator which is using a jvp followed by a vjp of complex (shallow) neural networks which ~ looks like this:
```python
def matvec(apply_fun…
-
Hi, I'm trying to create an hybrid model, that use your FEM-solver and a neural network.
To do that, I need to solve the same equations with different parameters (e.g the heat diffusion, where the di…
-
Offer a discretisation framework that can flexibly use fast direct solvers or FMM methods depending on user needs.
-
I'm using jax to test the AD of dominant eigensolver. As an example, the function `symeig` below returns only the lowest eigenvalue and corresponding eigenvector of a real symmetric matrix.
```pyth…
-
Intending to refactor the various diophantine solvers to put them under a common interface (i.e., one class that handles parameters, permutations, equation properties, solutions etc).
Task list:
…
-
Some nonlinear problems can be very efficiently solved with non-standard methods, possibly abusing the specific structure of the problem. To allow users to efficiently utilize such methods in tandem w…
-
#### Describe the bug
As noted in #22910, `Ridge(alpha=0, fit_intercept=True)` does not give the minimal norm solution for wide data, i.e. `n_features > n_samples`.
Note that we nowhere guarantee …
-
While working on this idea of using an iterative generalised Wiener filter to solve the calibration problem without the need for solution intervals, I realised that there is a shortcut that will get u…