CodeUniverse · package
OpenPNM
many-throat-conductance pressure objective gradient
AD speed-up on the same differentiation workload (against finite differences) Speed-up: 188.5×
High-dimensional demo
many-throat-conductance pressure objective gradient
Supplied demo scenario wording.
FD components requested 1 – 256 · AD components 565
View demo measurements
| Measure | Value |
|---|---|
| AD median (s) | 0.0022207500878721476 |
| FD median (s) | 0.4186607499141246 |
| FD / AD | 188.52222598143499 |
| Max scaled error | 4.7196623050206423e-11 |
| AD components | 565 |
| FD components requested | 256 |
| AD parameters | 256 |
| Repeats | 9 |
| FD components requested | AD median (s) | FD median (s) | FD / AD | Max scaled error | AD components |
|---|---|---|---|---|---|
| 1 | 565 | 0.002270875032991171 | 0.0016312500229105353 | 0.7183354430392725 | 2.6236283111324976e-13 |
| 2 | 565 | 0.002174625056795776 | 0.0035997909726575017 | 1.6553616732263958 | 4.0204598547935455e-12 |
| 4 | 565 | 0.0024195839650928974 | 0.006503708893433213 | 2.6879451125737273 | 1.3155767715546258e-11 |
| 8 | 565 | 0.0022125410614535213 | 0.012802000041119754 | 5.786107324358322 | 3.113331922995066e-11 |
| 16 | 565 | 0.003038667025975883 | 0.02539979200810194 | 8.358859918172401 | 3.113331922995066e-11 |
| 32 | 565 | 0.00232812506146729 | 0.05112104199361056 | 21.958030880605598 | 3.113331922995066e-11 |
| 64 | 565 | 0.002329500042833388 | 0.1035289999563247 | 44.44258340961506 | 3.113331922995066e-11 |
| 128 | 565 | 0.002259708009660244 | 0.20453758398070931 | 90.51505022167105 | 4.7196623050206423e-11 |
| 256 | 565 | 0.0022207500878721476 | 0.4186607499141246 | 188.52222598143499 | 4.7196623050206423e-11 |
One-direction performance audit
View audit measurements
| Demo scenario | Problem size | Repeats | AD median (s) | FD median (s) | FD / AD | Relative error | Supplied verdict |
|---|---|---|---|---|---|---|---|
| case001_full_model_and_setup | case001 | 21 | 0.0090011249995 | 0.0079604999628 | 0.884389447 | 3.125e-14 | modest AD slowdown |
FD/AD > 1 means automatic differentiation is faster; FD/AD < 1 means the finite-difference comparison is faster. One direction only, measured inside the package's own case.