Viscous Flow Questions & Answers  
Did you check if your answer is gridconverged? I.e., if changing $N$ by 2 times does not result in a major change in $u_b$?




First, make sure you use $N=5000$ (or more) to test your algorithm properly. Then, is the answer converged? I.e, are you getting the same answer for $N=20000$ then for $N=5000$?






You don't need to take the absolute value but $\tau_w$ should be positive anyway. If it isn't, this means there's a bug somewhere. You should treat $\tau_w$ as a constant within one iteration only (i.e. because find_coeff_and_rhs is called once at each iteration, thus $\tau_w$ should be taken as constant within find_coeff_and_rhs). Of course, it should be updated every time find_coeff_and_rhs is called..




Hm does mut equal 0 for all nodes?





$\pi$ 