SCOREC
phConstraint
PHASTA has the ability to set two independent comp1 constraints. From my reading of core's code, the prep for this seems to be broken. Specifically, it seems to be trying to reconcile them into some way for some reasons I can't quite follow. I can provide a 400 element case that produces NaN's for the BCtmp array. The inputs are rather simple. One plane has a comp1 of mag 0 and vector 0 0 1. A second plane has a constraint of mag 0 and vector -sin(12) cos(12) 0. These vectors are perpendicular so the correct treatment would be 2 distinct constraints on the model edge that is shared by the two planes without any change to to values. If I change 12 degrees to zero (e.g., 0 1 0) NaN's are not produced so it seems that multi-component vectors are not handled properly.
Actually, this looks like an inheritance problem. I just applied both comp1 constraints directly to the model edges and no-NaNs were produced in the BCtmp array that chef prepped for PHASTA.