Local refinement scales poorly
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
DOLFIN |
Confirmed
|
Medium
|
Anders Logg |
Bug Description
Bug reported by Glenn Terje Lines.
Local refinement seems to be n^3 in the number of cells.
Can be reproduced with the following code:
from dolfin import *
import numpy
import time
import random
mesh = UnitCube(10,10,10)
for i in range(10):
mf = MeshFunction(
k = 0
for cell in mesh.cells():
k += 1
t0 = time.time();
mesh = refine(mesh,mf);
dt = time.time() - t0;
print i, dt, k, k/dt
Results:
# results
P = numpy.array([
[0,0.1449220180
[1,0.3982179164
[2,0.8847539424
[3,2.4663689136
[4,10.550200939
[5,39.347328901
[6,302.
[7,2436.
n = P[:,2];
import pylab
pylab.loglog(
pylab.show()
Changed in dolfin: | |
status: | New → Confirmed |
assignee: | nobody → Anders Logg (logg) |
importance: | Undecided → Medium |
Changed in dolfin: | |
milestone: | none → 0.9.11 |
This bug is "superseded" by https:/ /bugs.launchpad .net/dolfin/ +bug/735588 and will (most likely) not be fixed until we reimplement the mesh refinement.