dolfin team mailing list archive

[Andre Massing <massing@xxxxxxxxx>]

** Attachment added: "demo_strange_l2norm_bug.py"
   https://bugs.launchpad.net/bugs/745130/+attachment/1954095/+files/demo_strange_l2norm_bug.py

-- 
You received this bug notification because you are a member of DOLFIN
Team, which is subscribed to DOLFIN.
https://bugs.launchpad.net/bugs/745130

Title:
  Strange l2 norms in stokes taylor hood demo.

Status in DOLFIN:
  New

Bug description:
  Hi!
  I recently encounter rather strange values when computing the l2 norm of u in the taylor hood demo and using increasingly finer meshes. The norms for the pressure and velocity do not seem to converge. There seems something wrong in the  method u.vector().norm("l2").  Executing the attached script (slight modification of the demo, computes Stokes on a unit square) with UnitSquare(N,N) and N = 10:20:100
  gives norms
  Norm of velocity coefficient vector: 14.3406836573253
  Norm of pressure coefficient vector: 48.8200968180882
  Solving linear variational problem.
  Norm of velocity coefficient vector: 42.3283877532062
  Norm of pressure coefficient vector: 135.906376761598
  Solving linear variational problem.
  Norm of velocity coefficient vector: 70.3139009258995
  Norm of pressure coefficient vector: 223.011978085219
  Solving linear variational problem.
  Norm of velocity coefficient vector: 98.2990847944175
  Norm of pressure coefficient vector: 310.120581527338
  Solving linear variational problem.
  Norm of velocity coefficient vector: 126.284157720703
  Norm of pressure coefficient vector: 397.23021024167