Originally posted by: vdesai
Hi all,
I am having difficulty in understanding something from barrier log file.
1) I sometimes get messages of the type:
Refinement - orig 2.45e-07, refined 2.45e-07, target 4.00e-06, 0 iter
From my experience the solution does not seem right when I get too many of these messages.
Below is a small excerpt from the log file.
Tried aggregator 1 time.
LP Presolve eliminated 4000 rows and 26 columns.
Reduced LP has 4054 rows, 32963 columns, and 1004963 nonzeros.
Presolve time = 0.67 sec.
Parallel mode: using up to 8 threads for barrier.
Looking for dense columns with more than 56 nonzeros.
Number of nonzeros in lower triangle of A*A' = 217431
Using Approximate Minimum Degree ordering
Total time for automatic ordering = 0.12 sec.
Summary statistics for Cholesky factor:
Threads = 8
Rows in Factor = 4054
Integer space required = 217786
Total non-zeros in factor = 222346
Total FP ops to factor = 12272486
L has 3959 super-nodes.
Histogram of super-node widths:
SupWidth: 1 96
Number: 3958 1
0.26s for iteration (0.21s, 253 Mflops for lin. solve)
Max number of central corrections is 0
Itn Primal Obj Dual Obj Prim Inf Upper Inf Dual Inf
0 1.8379841e+06 0.0000000e+00 3.56e+05 0.00e+00 3.92e+05
0.23s for iteration (0.20s, 258 Mflops for lin. solve)
1 1.7470795e+06 1.0507424e+04 3.43e+05 0.00e+00 3.42e+05
0.27s for iteration (0.20s, 268 Mflops for lin. solve)
2 1.3886718e+06 6.7192699e+04 2.76e+05 0.00e+00 3.62e-07
0.25s for iteration (0.20s, 259 Mflops for lin. solve)
3 2.2744777e+04 6.5930201e+04 3.37e-07 0.00e+00 3.80e-07
0.29s for iteration (0.22s, 238 Mflops for lin. solve)
Refinement - orig 2.45e-07, refined 2.45e-07, target 4.00e-06, 0 iter
4 2.6365237e+04 4.0174035e+04 2.20e-07 0.00e+00 2.83e-07
0.28s for iteration (0.20s, 259 Mflops for lin. solve)
Refinement - orig 2.33e-06, refined 2.33e-06, target 4.00e-06, 0 iter
5 2.8368577e+04 3.5427188e+04 8.82e-07 0.00e+00 1.86e-07
0.26s for iteration (0.21s, 253 Mflops for lin. solve)
Refinement - orig 7.33e-07, refined 7.33e-07, target 4.00e-06, 0 iter
6 3.0285482e+04 3.4574844e+04 6.70e-07 0.00e+00 1.23e-07
0.27s for iteration (0.22s, 237 Mflops for lin. solve)
Can some
#CPLEXOptimizers#DecisionOptimization