Hey Axel, thanks so much for the tips! They helped me solve the previous problem, but now I've encountered a new, related one. After obtaining some unexpected peaks when I plotted the dihedral histogram, my professor and I realized that they were occurring because my group command was too encompassing:
group DiHe type 3 6 7 8
compute 1 DiHe dihedral/local phi
Whereas I had only wanted the compute command to compute the value of one dihedral per monomer, it had computed others as well because there were multiple atoms of these types in each monomer. It seems that defining a group with types is not specific enough in my case, but with such a big system (6400 atoms), using atomic ID's does not seem realistic either. Maybe there is a way to restrict the compute to the case where all four atoms are bonded? What do you think is the best way to solve this problem?
Skolkovo Institute of Science and Technology