PyAMG vs AMGCL

output.txt

PyAMG
----------------

multilevel_solver
Number of Levels:     5
Operator Complexity:  1.338
Grid Complexity:      1.188
Coarse Solver:        'pinv2'
  level   unknowns     nonzeros
    0      1048576      5238784 [74.73%]
    1       175104      1572176 [22.43%]
    2        19537       175673 [ 2.51%]
    3         2154        21246 [ 0.30%]
    4          219         2365 [ 0.03%]

residual norm is 8.29613125832e-07
setup: 4.93 sec
solve: 2.83 sec

AMGCL
----------------

Number of levels:    4
Operator complexity: 1.34
Grid complexity:     1.19

level     unknowns       nonzeros
---------------------------------
    0      1048576        5238784 (74.75%)
    1       175104        1572176 (22.43%)
    2        19579         175863 ( 2.51%)
    3         2259          21465 ( 0.31%)

residual norm is 1.92973750974e-06
setup: 0.47 sec
solve: 1.05 sec

pyamg_vs_pyamgcl.py

#!/usr/bin/python

import numpy as np
from scipy import *
from scipy.linalg import *
from scipy.sparse.linalg import cg
from pyamg import *
from pyamg.gallery import *
import pyamgcl as amg
from time import time

A = poisson((1024,1024), format='csr')
b = rand(A.shape[0])                     # pick a random right hand side

print "PyAMG"
print "----------------\n"

tic = time()
ml = smoothed_aggregation_solver(A)
tm_setup = time() - tic
print ml
tic = time()
x,info = cg(A, b, tol=1e-8, M=ml.aspreconditioner(cycle='V'))
tm_solve = time() - tic
print "residual norm is", norm(b - A*x)
print "setup: %.2f sec" % tm_setup
print "solve: %.2f sec" % tm_solve

print "\nAMGCL"
print "----------------\n"

tic = time()
solve = amg.make_solver(A, solver=amg.solver_type.cg, prm={"tol" : 1e-8})
tm_setup = time() - tic
print solve

tic = time()
x = solve(b)
tm_solve = time() - tic
print "residual norm is", norm(b - A*x)
print "setup: %.2f sec" % tm_setup
print "solve: %.2f sec" % tm_solve