-
Hi, I was attempting to run Euclid through PETSc but ran into an error message that Euclid does not support 64 bit indices, which I need for a large application. How much work would it be to add 64 bi…
-
Hi All,
I am trying to perform adjoint calculation for a stokes problem with Schur complement preconditioner. When I use the default scipy optimizer it works as expected, however when I switch to ROL…
-
Maybe this issue should be directed towards LinearAlgebra.jl.
Iterative solvers require lots of linear solves Ax=b, but these are often quite useless when A is ill-conditioned. Looking at some pack…
-
See https://github.com/exadg/exadg/blob/3bc6d4178089288615107f36d74e25872251d9fc/include/exadg/grid/grid.h#L71-L75
It is not clear from the name why this variable should include the finest triangul…
-
### Description
We have O(50) new diffs showing up in our Nalu-nightly gcc-testing. One of the tests in our suite is singular (this is not uncommon in a subset of low-Mach flows where the number of e…
-
**Required Info:**
- Operating System:
Ubuntu 22.04) -->
- ROS Version
ROS2 - humble
- Version or commit hash:
- Laser unit:
rplidar
#### Steps to reproduce issue
-run: ro…
-
### 🐛 Describe the bug
`torch.linalg.eigh` fails on some large low-rank float32 matrices on GPU, but succeeds on CPU or when cast to float64. (See similar issue at https://github.com/pytorch/pytorch/…
-
Recently, I have observed a slowdown in computation speed in our models (@mibillen ). I looked into this and found the " Solve Stokes system" causing the slowdown, where the difference is about 1/3. …
-
I run a program with gpytorch that needed more than 30GB GPU memory, so I decide to run it with multiple GPUs.
```
class MultitaskGPModel(gpytorch.models.ExactGP):
def __init__(self, x_train,…
-
**Title:** Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training
**Abstract:** Given the massive cost of language model pre-training, a non-trivial improvement of the…