Update NS results.

README.md

@@ -43,9 +43,8 @@ For more details, see
 
 <div style="text-align: center;">
 <a href="https://aai-institute.github.io/continuiti/benchmarks/#navierstokes">
-<img alt="navierstokes" src="https://aai-institute.github.io/continuiti/img/ns.png" width="100"><br>
-Fourier Neural Operator (FNO) for Navier-Stokes flow.<br>
-Ground truth vs. prediction.
+<img alt="navierstokes" src="https://aai-institute.github.io/continuiti/img/ns.png" width="80%"><br>
+Fourier Neural Operator (FNO) for Navier-Stokes flow
 </a>
 </div>
 

benchmarks/navierstokes/results.txt

@@ -1,10 +1,25 @@
+
+Depth 4 (201M parameters)
+=========================
+
 rel. error train = 4.9422e-03
 min loss = 2.3351e-03 at index 587
 max loss = 1.0090e-02 at index 648
 rel. error test =  1.8622e-01
 min loss = 1.0219e-01 at index 144
 max loss = 4.4294e-01 at index 179
 
+
+Depth 16 (805M parameters)
+==========================
+
+rel. error train = 1.8400e-04
+min loss = 1.0007e-04 at index 2
+max loss = 4.5649e-04 at index 707
+rel. error test =  1.4862e-01
+min loss = 7.2528e-02 at index 59
+max loss = 4.2331e-01 at index 179
+
 --
 
 Reference: _Li, Zongyi, et al. "Fourier neural operator for parametric partial

docs/benchmarks/index.md

@@ -7,13 +7,20 @@ this directory for detailed information on how the benchmarks are run.
 ## [NavierStokes](../api/continuiti/benchmarks/#continuiti.benchmarks.NavierStokes)
 
 Reference: _Li, Zongyi, et al. "Fourier neural operator for parametric partial
-differential equations." arXiv preprint arXiv:2010.08895 (2020)_ _Table 1 ($\nu$ = 1e−5  T=20  N=1000)_
+differential equations." arXiv preprint [arXiv:2010.08895](https://arxiv.org/abs/2010.08895) (2020)_
+_Table 1_ ($\nu$ = 1e−5  T=20  N=1000)
 
 _reported for_ FNO-3D: __0.1893__ (rel. test error)
 
-[FourierNeuralOperator](../api/continuiti/operators/#continuiti.operators.FourierNeuralOperator):
-0.0185 (rel. train error)  __0.1841__ (rel. test error)
+[FourierNeuralOperator](../api/continuiti/operators/#continuiti.operators.FourierNeuralOperator)
 
+| Depth | Parameters | Training Time (V100) | rel. train error | rel. test error |
+|-------|------------|----------------------|------------------|-----------------|
+|     4 |       201M |                91min |         4.94e-03 |      __0.1862__ |
+|    16 |       805M |               181min |         1.84e-04 |          0.1486 |
+
+<div style="text-align:center;">
+Visualization of best and worst training and test samples for FNO with depth 4 (as in paper).
 <table>
 <tr>
 <td>
@@ -38,5 +45,6 @@ rel. error = 4.4294e-01
 </td>
 </tr>
 </table>
+</div>
 
 {% include 'benchmarks/tables.html' %}