When tridag_par checks for zero pivot, it uses "if (maxval(abs(b(1:n))) == 0.0)". This returns true only if every single diagonal element is zero. But shouldnt the check be for whether any diagonal element is zero instead (ie, minval instead of maxval)?? E.g., in the lines below, computing piva and pivc, ocurrence of any b=0.0 would cause division by zero. Am I missing something?