Updated 0.1.0 version

README.md

@@ -42,6 +42,7 @@ Data generation instructions are available in the [SFNO folder](./fno).
 ## Examples
 - Demos of different simulation setups:
   - [2D simulation with a pseudo-spectral solver](./examples/Kolmogrov2d_rk4_cn_forced_turbulence.ipynb)
+  - [2D simulation with a finite volume solver](./examples/Kolmogrov2d_rk4_fvm_forced_turbulence.ipynb)
 - Demos of Spatiotemporal FNO's training and evaluation using the neural operator-assisted fluid simulation pipelines
   - [Training of SFNO for only 15 epochs for the isotropic turbulence example](./examples/ex2_SFNO_train.ipynb)
   - [Training of SFNO for only ***10*** epochs with 1k samples and reach `1e-2` level of relative error](./examples/ex2_SFNO_train_fnodata.ipynb) using the data in the FNO paper, which to our best knowledge no operator learner can do this in <100 epochs in the small data regime.
@@ -56,6 +57,7 @@ The Apache 2.0 License in the root folder applies to the `torch-cfd` folder of t
 ## Contributions
 PR welcome. Currently, the port of `torch-cfd` currently includes:
 - The pseudospectral method for vorticity uses anti-aliasing filtering techniques for nonlinear terms to maintain stability.
+- The finite volume method on a MAC grid for velocity, and using the projection scheme to impose the divergence free condition.
 - Temporal discretization: Currently only RK4 temporal discretization uses explicit time-stepping for advection and either implicit or explicit time-stepping for diffusion.
 - Boundary conditions: only periodic boundary conditions.
 

examples/ex2_FNO3d_train_normalized.ipynb

@@ -97,11 +97,11 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 17,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
-    "train_dataset = BochnerDatasetFixed(\n",
+    "train_dataset = SpatioTemporalDatasetFixedTime(\n",
     "    datapath=TRAIN_PATH,\n",
     "    fields=[\"vorticity\"],\n",
     "    n_samples=Ntrain,\n",
@@ -112,7 +112,7 @@
     "    out_steps=T,\n",
     "    T_start=T_start,\n",
     ")\n",
-    "test_dataset = BochnerDatasetFixed(\n",
+    "test_dataset = SpatioTemporalDatasetFixedTime(\n",
     "    datapath=VALID_PATH,\n",
     "    fields=[\"vorticity\"],\n",
     "    n_samples=Ntest,\n",
@@ -493,7 +493,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.11.9"
+   "version": "3.10.17"
   }
  },
  "nbformat": 4,

examples/ex2_SFNO_finetune_fnodata.ipynb

@@ -2,7 +2,7 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -14,11 +14,13 @@
     "import torch.nn.functional as F\n",
     "from torch.linalg import norm\n",
     "\n",
-    "from sfno.utils import get_seed\n",
-    "from sfno.pipeline import *\n",
-    "from sfno.visualizations import *\n",
-    "from sfno.sfno import SFNO\n",
-    "from sfno.finetune import OutConvFT\n",
+    "from fno.utils import get_seed\n",
+    "from fno.pipeline import *\n",
+    "from fno.visualizations import *\n",
+    "from fno.datasets import *\n",
+    "from fno.losses import *\n",
+    "from fno.sfno import SFNO\n",
+    "from fno.finetune import OutConvFT\n",
     "from torch.utils.data import DataLoader\n",
     "\n",
     "get_seed(42, printout=False)\n",
@@ -78,11 +80,11 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [
-    "test_dataset = BochnerDataset(datapath=TEST_PATH, \n",
+    "test_dataset = SpatioTemporalDataset(datapath=TEST_PATH, \n",
     "                              fields=fields,\n",
     "                              n_samples=Ntest,\n",
     "                              steps=T,\n",