sfepy: linear system not solved! (err = 3.217853e-12)

Dec. 20, 2010
6:18 a.m.
Although SfePy is indicating a system is NOT solved, the solution appears valid. Is the err term perhaps too rigid?
-- Andre

December 2010
6:20 a.m.
On 12/20/10 15:18, Andre Smit wrote:
Although SfePy is indicating a system is NOT solved, the solution appears valid. Is the err term perhaps too rigid?
Probably, try tweaking the following option of the Newton solver:
'lin_red' : 1e-2, # Linear system error < (eps_a * lin_red).
Set it e.g. to 1.
r.
5267
Age (days ago)
5267
Last active (days ago)
1 comments
2 participants
participants (2)
-
Andre Smit
-
Robert Cimrman