@@ -13,38 +13,38 @@ After we get 'runs', we can do the following things:
1313Convert exodus files to sequence files:
1414
1515 make -f runs convert timestepfile=timesteps.txt
16- (using normalized timesteps, time: 20min36s)
16+ (using normalized timesteps, time: 20min36s)
1717
1818 make -f runs convert exodus2seq_output=hdfs://icme-hadoop1.localdomain/user/yangyang/simform/output/data.seq2/
19- (without using normalized timesteps, time: 20min24s)
19+ (without using normalized timesteps, time: 20min24s)
2020
2121
2222Predict without SVD:
2323
2424 make -f runs predict design=design_points.txt points=new_points.txt
25- (design points:99 new points:10, time:16min2s)
25+ (design points:99 new points:10, time:16min2s)
2626
2727 make -f runs seq2exodus output_name=thermal_maze_interpolation
28- (output 10 new interpolation exodus files, time: 4min45s)
28+ (output 10 new interpolation exodus files, time: 4min45s)
2929
3030
3131Predict with SVD (we can get not only interpolation value but also estimated errors):
3232
3333 Convert sequenc files to matrix sequence files:
3434 make -f runs seq2mseq
35- (time:2hr...)
35+ (time:2hr...)
3636
37- Compute the SVD model:
37+ Compute the SVD model:
3838 make -f runs model numExodusfiles=99
39- (step1: full1, 44min46s
40- step2: full2 1min
41- step3: full3 4min6s
42- step4: TSMatMul 3min42s
43- total time: 54min
39+ (step1: full1, 44min46s
40+ step2: full2 1min
41+ step3: full3 4min6s
42+ step4: TSMatMul 3min42s
43+ total time: 54min
4444
45- Predict with SVD singular values and vectors (before this steps, we should create weights.txt first):
45+ Predict with SVD singular values and vectors (before this steps, we should create weights.txt first):
4646 make -f runs predict weights=weights.txt SVD=True
47- (new points:2, time: 22min26s)
47+ (new points:2, time: 22min26s)
4848
4949 make -f runs seq2exodus output_name=thermal_maze_interpolation SVD=True
50- (output 2 new interpolation exodus file, time: 2min40s)
50+ (output 2 new interpolation exodus file, time: 2min40s)
0 commit comments