Results of SW flows in dimension 100
Time for Batch OT
| Batch Size |
Time computer (s) |
Time PAGODA (s) |
| 128 | 0.02 | 0.06 |
| 2048 | 0.63 | 1.7 |
| 4096 | 2.29 (max iter) | 2.5 (max iter) |
Table of results for measures of 2048 diracs in dimension 100
(Columns can be sorted by clicking on name)
| Number of gradient steps |
Number of projections |
Learning rate |
time (s) |
SW(u_k, v) / SW(u_0, v) |
| 100 | 10 | 1 | 0.164803 | 0.472074 |
| 100 | 10 | 10 | 0.102963 | 0.216941 |
| 100 | 100 | 1 | 0.182423 | 0.452417 |
| 100 | 100 | 10 | 0.124489 | 0.20018 |
| 100 | 100 | 100 | 0.117438 | 0.136546 |
| 100 | 1000 | 1 | 0.833632 | 0.455624 |
| 100 | 1000 | 10 | 0.778648 | 0.201246 |
| 100 | 1000 | 100 | 0.679935 | 0.0928675 |
| 1000 | 10 | 1 | 0.480614 | 0.198615 |
| 1000 | 10 | 10 | 0.477921 | 0.105917 |
| 1000 | 100 | 1 | 0.975446 | 0.199979 |
| 1000 | 100 | 10 | 1.17974 | 0.0930475 |
| 1000 | 100 | 100 | 1.03078 | 0.0835821 |
| 1000 | 1000 | 1 | 6.66314 | 0.199272 |
| 1000 | 1000 | 10 | 6.62657 | 0.0922859 |
| 1000 | 1000 | 100 | 6.71508 | 0.0754507 |
| 10000 | 10 | 1 | 6.31685 | 0.0930368 |
| 10000 | 10 | 10 | 6.53919 | 0.083482 |
| 10000 | 100 | 1 | 10.2198 | 0.0922568 |
| 10000 | 100 | 10 | 10.0876 | 0.0756016 |
| 10000 | 100 | 100 | 10.1813 | 0.0821584 |
| 10000 | 1000 | 1 | 66.9805 | 0.0920823 |
| 10000 | 1000 | 10 | 67.0885 | 0.0746696 |
| 10000 | 1000 | 100 | 66.7669 | 0.0288367 |
| 100000 | 10 | 1 | 58.7417 | 0.0755984 |
| 100000 | 10 | 10 | 63.3239 | 0.0822453 |
| 100000 | 100 | 1 | 102.394 | 0.0746564 |
| 100000 | 100 | 10 | 116.481 | 0.0285297 |
| 100000 | 100 | 100 | 141.55 | 0.0821322 |
| 100000 | 1000 | 1 | 852.506 | 0.0746223 |
| 100000 | 1000 | 10 | 854.832 | 0.0237895 |
| 100000 | 1000 | 100 | 916.687 | 4.15482e-16 |
Measures of 128 Diracs in Dimension 100
| nb_steps |
nb_slices |
tau |
time (s) |
Relative SW |
Relative Wass |
| 500 | 500 | 100 | 0.597305 | 0.16584 | 0.32916 |
| 500 | 500 | 125 | 0.350269 | 0.142817 | 0.261338 |
| 500 | 500 | 150 | 0.383282 | 0.0547908 | 0.0629778 |
| 500 | 500 | 175 | 0.347542 | 0.0982687 | 0.157601 |
| 500 | 500 | 200 | 0.371706 | 0.896394 | 0.306145 |
| 500 | 700 | 100 | 0.335827 | 0.138417 | 0.257317 |
| 500 | 700 | 125 | 0.314984 | 0.0934727 | 0.168946 |
| 500 | 700 | 150 | 0.325517 | 0.0617724 | 0.10658 |
| 500 | 700 | 175 | 0.446476 | 0.0404182 | 0.0409268 |
| 500 | 700 | 200 | 0.326986 | 0.89604 | 0.305091 |
| 500 | 1000 | 100 | 0.341546 | 0.130391 | 0.252959 |
| 500 | 1000 | 125 | 0.333115 | 0.080263 | 0.1268 |
| 500 | 1000 | 150 | 0.354751 | 0.0492413 | 0.056634 |
| 500 | 1000 | 175 | 0.336964 | 3.11361e-05 | 1.0601e-05 |
| 500 | 1000 | 200 | 0.377021 | 0.894774 | 0.304574 |
| 1000 | 500 | 100 | 0.73185 | 6.59249e-16 | 1.10052e-08 |
| 1000 | 500 | 125 | 0.758921 | 8.67042e-16 | 1.10702e-08 |
| 1000 | 500 | 150 | 0.722464 | 1.33436e-15 | 1.08081e-08 |
| 1000 | 500 | 175 | 0.810443 | 0.00257932 | 0.00089292 |
| 1000 | 500 | 200 | 0.716832 | 0.894828 | 0.304597 |
| 1000 | 700 | 100 | 0.665241 | 5.59715e-16 | 1.11025e-08 |
| 1000 | 700 | 125 | 0.611177 | 7.26283e-16 | 1.04027e-08 |
| 1000 | 700 | 150 | 0.641532 | 1.03296e-15 | 1.11989e-08 |
| 1000 | 700 | 175 | 0.683664 | 4.06271e-06 | 1.38294e-06 |
| 1000 | 700 | 200 | 0.667783 | 0.895537 | 0.304876 |
| 1000 | 1000 | 100 | 0.721565 | 4.82387e-16 | 1.09071e-08 |
| 1000 | 1000 | 125 | 0.705463 | 6.25415e-16 | 1.07749e-08 |
| 1000 | 1000 | 150 | 0.676039 | 8.62412e-16 | 1.12309e-08 |
| 1000 | 1000 | 175 | 0.696477 | 1.74153e-15 | 1.19121e-08 |
| 1000 | 1000 | 200 | 0.736952 | 0.894804 | 0.304585 |
Graphs
Same graphs in time
