While attempting to utilize this repository for research, I ran across an issue that I believe to be a very quick fix. Essentially, the standard code lines for Laplace, for example, work perfectly:
using IFGF
targets = rand(3,100)
sources = rand(3,100)
charges = rand(100)
IFGF.laplace3d(targets, sources; charges, tol=1e-4)
However, when I changed the size of the targets to not match the sources, the code threw an error complaining about the output size:
using IFGF
targets = rand(3,99)
sources = rand(3,100)
charges = rand(100)
IFGF.laplace3d(targets, sources; charges, tol=1e-4)
After some work, I tracked the issue to the following line in the individual kernel code files. I know for sure the line exists in both the Laplace and Helmholtz, in particular.
out = grad ? zeros(T, 3, m) : zeros(T, n)
The code runs after changing "n" to "m". I just want to make sure that the necessary changes are uploaded and everything continues to work as intended.
While attempting to utilize this repository for research, I ran across an issue that I believe to be a very quick fix. Essentially, the standard code lines for Laplace, for example, work perfectly:
However, when I changed the size of the targets to not match the sources, the code threw an error complaining about the output size:
After some work, I tracked the issue to the following line in the individual kernel code files. I know for sure the line exists in both the Laplace and Helmholtz, in particular.
out = grad ? zeros(T, 3, m) : zeros(T, n)The code runs after changing "n" to "m". I just want to make sure that the necessary changes are uploaded and everything continues to work as intended.