search

alecjacobson / geometry-processing-parameterization

Parameterization assignment for Geometry Processing course
https://github.com/alecjacobson/geometry-processing
Mozilla Public License 2.0
150 stars 67 forks source link

Confused about a sign in L #27

Closed dknyxh closed 6 years ago

dknyxh commented 6 years ago

In the readme, Mass-spring methods section, L_ij is defined to have w_ij when i!=j and edge[i,j] exists. I am wondering why is this not -w_ij since expanding the equation w_ij|u_i - u_j|^2 will gives us w_ij u_i^2 + w_ij u_j^2 - 2w_ij u_i u_j.

alecjacobson commented 6 years ago

You're correct. It's a bit of a confusion of conventions. Often the Laplacian is assumed to have (mostly) non-negative off-diagonal entries. So this should be referring to -L_ij.