Incorrect Frobenius Norm?

[JulianKnodt]

Hi, I'm implementing my own exhaustive matching for the orientation of Gaussian Splats, and was looking at your implementation as a reference. I was a bit confused by your implementation of the frobenius norm, and I can't make heads or tails of it. Specifically, it can be negative and doesn't match the mathematical definition. Is there a reason this is how it was written?

gaussian-hierarchy/rotation_aligner.cpp

Line 19 in 677c855

sum += A(i, j) * B(i, j);