{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# Learned Closure: Training a Neural Viscosity Model Through a PDE Solver (PyTorch)\n", "\n", "In this tutorial, you will learn how to:\n", "\n", "1. **Build a Tesseract** that wraps a single-timestep differentiable PDE solver (a 1D Burgers' equation solver) and exposes a vector-Jacobian product.\n", "2. **Use [Tesseract-Torch](https://github.com/pasteurlabs/tesseract-torch)** to call the served solver as a native PyTorch autograd layer via `apply_tesseract`, so gradients flow through the container automatically.\n", "3. **Train a neural network closure end-to-end** through the containerized solver, differentiating through the entire time-stepping loop.\n", "4. **Compare the learned closure against baselines** (a pure physics model and a pure ML model).\n", "\n", "We will replace the viscosity model of a Burgers' equation solver, normally a hand-tuned constant, with a small neural network, and train it so that it recovers the true (unknown) viscosity profile from solution data alone.\n", "\n", "This is a small end-to-end example of embedding a learned closure into a solver with Tesseract. The closure is called once per solver step, which is the right structure when the physics genuinely needs a per-step update. It is deliberately kept simple so the moving parts are easy to follow; the *What's next* section at the end points to the tricks real closures usually need.\n", "\n", "## Context\n", "\n", "Closure modeling is a recurring problem across computational science: a simulation resolves the large scales but needs a model for the unresolved physics, such as a turbulence closure, a sub-grid scheme, or a constitutive law. These closures are often hand-tuned or empirical, and a natural idea is to *learn* them from data instead ([Duraisamy, Iaccarino & Xiao, 2019](https://doi.org/10.1146/annurev-fluid-010518-040547)). The catch is that a closure only makes sense *inside* the solver: to train it well, gradients have to flow from the simulation output, through the solver, and back into the network ([Um et al., 2020](https://arxiv.org/abs/2007.00016); [Shankar et al., 2023](https://doi.org/10.1088/2632-2153/acb19c)).\n", "\n", "This is hard in practice because the solver is usually not a 30-line function you can `import` into your training script. It is often a heavyweight simulator with its own runtime, dependencies, and adjoint: a Fortran/C++ CFD code, a legacy in-house solver, or something differentiated by [Enzyme](https://enzyme.mit.edu/) at the LLVM IR level.\n", "\n", "**Tesseract is most useful for that hard case:** a big, expensive simulator with a gnarly toolchain (a specific compiler, MPI, a licensed library, a non-Python runtime, a hand-written adjoint) that you want to hide behind a clean differentiable interface. The Burgers' solver used here is the opposite: a few lines of NumPy/Torch, cheap to run, and trivially `import`-able, so on its own it needs no container at all and the HTTP round-trip dominates its runtime. We use it precisely *because* it is cheap, so the whole demo runs in a few minutes. The composition pattern below is identical either way; only the overhead changes.\n", "\n", "With Tesseracts, we wrap the solver in a container that exposes a clean differentiable interface, and call it over HTTP as a single layer inside an otherwise ordinary PyTorch training loop. The neural network lives in native PyTorch; the solver lives in a Docker image. [Tesseract-Torch](https://github.com/pasteurlabs/tesseract-torch) registers the served solver as a PyTorch autograd custom function, so `loss.backward()` dispatches a vector-Jacobian product (VJP) call back through the solver automatically, the analogue of what [Tesseract-JAX](https://github.com/pasteurlabs/tesseract-jax) does for JAX in the other demos in this series.\n", "\n", "### The components\n", "\n", "- **`burgers_solver`** (containerized Tesseract): a single-timestep Burgers' equation solver. Takes the current velocity field `u` and a viscosity field `nu`, returns the velocity field after one explicit Euler step, a pure physics component with the interface $(u, \\nu, dt) \\to u_\\text{next}$. It exposes a VJP, so it slots into PyTorch autograd like any other layer, even though it runs in a separate container.\n", "- **`ViscosityNet`** (`torch.nn.Module`): a small MLP that maps local flow features $(u, \\partial u/\\partial x, x)$ to a viscosity field $\\nu$. An ordinary network, trained with a standard optimizer, in this process.\n", "\n", "At each timestep the outer loop calls the network directly to predict $\\nu$, then calls the solver via `apply_tesseract` to advance one step (shown in full in Step 4).\n", "\n", "To learn more about building and running Tesseracts, please refer to the [Tesseract documentation](https://docs.pasteurlabs.ai/projects/tesseract-core/latest/)." ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "FULL_RUN=False: 3 train / 2 test ICs, 20 steps, 100 epochs, dt=0.0005\n" ] } ], "source": [ "import os\n", "\n", "import matplotlib.pyplot as plt\n", "import numpy as np\n", "import torch\n", "import torch.nn as nn\n", "from tesseract_torch import apply_tesseract\n", "\n", "from tesseract_core import Tesseract\n", "\n", "torch.set_default_dtype(torch.float64)\n", "\n", "# Workload size. Every solver call is an HTTP round-trip to the container, so\n", "# the defaults are modest to keep the demo (and CI) to a few minutes.\n", "# Set FULL_RUN=1 for the publication-quality run used to produce the figures.\n", "FULL_RUN = os.environ.get(\"FULL_RUN\", \"0\") == \"1\"\n", "\n", "# Time step. The spatially-varying viscosity only becomes identifiable once\n", "# diffusion has acted appreciably over the rollout, so we use a dt large enough\n", "# for the closure to actually \"feel\" the viscosity profile (well within the\n", "# explicit-diffusion CFL limit nu*dt/dx^2 < 0.5; here nu_max*dt/dx^2 ≈ 0.40).\n", "# Too small a dt and every viscosity profile fits the data equally well: the\n", "# closure can't recover the shape no matter how long it trains. Using a larger\n", "# dt also lets us keep the rollout short. The closure is called once per step\n", "# and we backprop through the whole chain, so fewer steps means a cheaper,\n", "# better-behaved gradient. A real, longer rollout would reach for gradient\n", "# checkpointing here (see the \"What's next\" section).\n", "DT = 5e-4\n", "LR = 5e-3\n", "\n", "if FULL_RUN:\n", " N_TRAIN, N_TEST = 8, 4\n", " N_STEPS = 40\n", " N_EPOCHS = 300\n", " DIRECT_EPOCHS = 300\n", "else:\n", " # Small but enough to recover the viscosity hump and beat the constant\n", " # baseline.\n", " N_TRAIN, N_TEST = 3, 2\n", " N_STEPS = 20\n", " N_EPOCHS = 100\n", " DIRECT_EPOCHS = 100\n", "\n", "print(\n", " f\"FULL_RUN={FULL_RUN}: {N_TRAIN} train / {N_TEST} test ICs, \"\n", " f\"{N_STEPS} steps, {N_EPOCHS} epochs, dt={DT}\"\n", ")\n", "\n", "# Grid setup (must match the solver)\n", "N = 128\n", "DX = 1.0 / (N - 1)\n", "X = torch.linspace(0.0, 1.0, N)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Step 1: Build and serve the solver Tesseract\n", "\n", "We build the solver into a Docker image with `tesseract build`, then serve it in a container and call it over HTTP, the same way you would deploy a real simulator. The training loop below never imports the solver's code; it only ever talks to the running container.\n", "\n", "First, build the image (this can take a few minutes the first time):" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "\u001b[2K \u001b[1;2m[\u001b[0m\u001b[34mi\u001b[0m\u001b[1;2m]\u001b[0m Building image \u001b[33m...\u001b[0m\n", "\u001b[2K\u001b[37m⠋\u001b[0m \u001b[37mProcessing\u001b[0m\n", "\u001b[1A\u001b[2K \u001b[1;2m[\u001b[0m\u001b[34mi\u001b[0m\u001b[1;2m]\u001b[0m Built image sh\u001b[1;92ma256:270b\u001b[0m9738ccf8, \u001b[1m[\u001b[0m\u001b[32m'burgers-solver:0.1.0'\u001b[0m, \u001b[32m'burgers-solver:latest'\u001b[0m\u001b[1m]\u001b[0m\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "[\"burgers-solver:0.1.0\", \"burgers-solver:latest\"]\n" ] } ], "source": [ "%%bash\n", "tesseract build burgers_solver/" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Now load the built image and start a server container. `serve()` launches the container; we tear it down at the end of the notebook to free resources." ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Solver endpoints: ['apply', 'jacobian', 'jacobian_vector_product', 'vector_jacobian_product', 'health', 'abstract_eval', 'test']\n" ] } ], "source": [ "solver_tess = Tesseract.from_image(\"burgers-solver\")\n", "solver_tess.serve()\n", "print(f\"Solver endpoints: {solver_tess.available_endpoints}\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Step 2: The ground truth --- Burgers' equation with spatially-varying viscosity\n", "\n", "We generate training data from a Burgers' equation with a **known but non-trivial** viscosity profile:\n", "\n", "$$\\nu_{\\text{true}}(x) = \\nu_0 \\left(1 + A \\sin(\\pi x)\\right)$$\n", "\n", "This represents a spatially-varying material property, analogous to a turbulence closure, constitutive law, or sub-grid model that varies across the domain. The neural closure's job is to recover this profile from solution data alone.\n", "\n", "The reference solutions come from the **same served solver**, called forward-only (no gradients); there is no second copy of the physics anywhere in this notebook." ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Training set: 3 initial conditions -> solutions\n", "Test set: 2 initial conditions -> solutions\n", "Grid: 128 points, dt=0.0005, 20 steps\n" ] }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# True viscosity profile\n", "NU_0 = 0.02\n", "A = 0.8\n", "nu_true = NU_0 * (1.0 + A * torch.sin(np.pi * X))\n", "\n", "\n", "def burgers_reference(u0, nu_field, dt, n_steps):\n", " \"\"\"Reference solution with a prescribed viscosity field (no neural network).\n", "\n", " Calls the served solver Tesseract forward-only via `.apply()`. This is just\n", " data generation, so it stays outside autograd. `.apply()` returns decoded\n", " NumPy arrays, which we wrap back into tensors.\n", " \"\"\"\n", " u = u0.clone()\n", " for _ in range(n_steps):\n", " out = solver_tess.apply({\"u\": u, \"nu\": nu_field, \"dt\": dt})\n", " # .apply() returns a (possibly read-only) NumPy array decoded from JSON;\n", " # copy into a fresh tensor so PyTorch is happy.\n", " u = torch.tensor(np.asarray(out[\"u_next\"]))\n", " return u\n", "\n", "\n", "def make_ic(seed):\n", " \"\"\"Random smooth initial condition: sum of low-frequency sinusoids.\"\"\"\n", " rng = torch.Generator().manual_seed(seed)\n", " a1 = 0.5 + 0.5 * torch.rand(1, generator=rng).item()\n", " a2 = 0.3 * torch.rand(1, generator=rng).item()\n", " phase = torch.rand(1, generator=rng).item() * np.pi\n", " u0 = a1 * torch.sin(2 * np.pi * X + phase) + a2 * torch.sin(4 * np.pi * X)\n", " u0[0] = 0.0\n", " u0[-1] = 0.0\n", " return u0\n", "\n", "\n", "train_ics = torch.stack([make_ic(i) for i in range(N_TRAIN)])\n", "test_ics = torch.stack([make_ic(N_TRAIN + i) for i in range(N_TEST)])\n", "\n", "with torch.no_grad():\n", " train_targets = torch.stack(\n", " [burgers_reference(ic, nu_true, DT, N_STEPS) for ic in train_ics]\n", " )\n", " test_targets = torch.stack(\n", " [burgers_reference(ic, nu_true, DT, N_STEPS) for ic in test_ics]\n", " )\n", "\n", "print(f\"Training set: {N_TRAIN} initial conditions -> solutions\")\n", "print(f\"Test set: {N_TEST} initial conditions -> solutions\")\n", "print(f\"Grid: {N} points, dt={DT}, {N_STEPS} steps\")\n", "\n", "# Visualize one example\n", "fig, axes = plt.subplots(1, 2, figsize=(12, 4))\n", "axes[0].plot(X.numpy(), train_ics[0].numpy(), label=\"Initial condition\")\n", "axes[0].plot(\n", " X.numpy(), train_targets[0].numpy(), label=f\"Solution at t={DT * N_STEPS:.3f}\"\n", ")\n", "axes[0].set_xlabel(\"x\")\n", "axes[0].set_ylabel(\"u\")\n", "axes[0].legend()\n", "axes[0].set_title(\"Example: Burgers' equation with true viscosity\")\n", "\n", "axes[1].plot(X.numpy(), nu_true.numpy(), \"k-\", linewidth=2)\n", "axes[1].set_xlabel(\"x\")\n", "axes[1].set_ylabel(r\"$\\nu(x)$\")\n", "axes[1].set_title(\"True viscosity profile (to be learned)\")\n", "axes[1].set_ylim(bottom=0)\n", "\n", "plt.tight_layout()\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Step 3: The neural closure --- a plain PyTorch module\n", "\n", "The closure is an ordinary `torch.nn.Module`: a small MLP with 2 hidden layers of 32 units. It maps the local flow features $(u, \\partial u/\\partial x, x)$ at each grid point to a viscosity value. There is nothing Tesseract-specific about it; it is the network the closure researcher brings, trained with a standard optimizer, entirely in this process.\n", "\n", "A sigmoid keeps the predicted viscosity in a physically reasonable range $[0, \\nu_{\\max}]$, which also prevents CFL violations in the explicit solver." ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Closure parameters: 1217\n", "Initial viscosity range: [0.0204, 0.0310]\n" ] } ], "source": [ "class ViscosityNet(nn.Module):\n", " \"\"\"MLP closure: local flow features (u, du/dx, x) -> viscosity nu.\"\"\"\n", "\n", " def __init__(self, hidden_dim=32, nu_max=0.05):\n", " super().__init__()\n", " self.nu_max = nu_max\n", " self.net = nn.Sequential(\n", " nn.Linear(3, hidden_dim),\n", " nn.Tanh(),\n", " nn.Linear(hidden_dim, hidden_dim),\n", " nn.Tanh(),\n", " nn.Linear(hidden_dim, 1),\n", " )\n", "\n", " def forward(self, u, dudx, x):\n", " features = torch.stack([u, dudx, x], dim=-1) # (N, 3)\n", " out = self.net(features)[:, 0] # (N,)\n", " return self.nu_max * torch.sigmoid(out)\n", "\n", "\n", "torch.manual_seed(1)\n", "closure = ViscosityNet()\n", "n_params = sum(p.numel() for p in closure.parameters())\n", "print(f\"Closure parameters: {n_params}\")\n", "\n", "# Sanity check: the untrained closure produces sensible viscosities\n", "with torch.no_grad():\n", " dudx0 = torch.gradient(train_ics[0], spacing=(DX,))[0]\n", " nu0 = closure(train_ics[0], dudx0, X)\n", "print(f\"Initial viscosity range: [{float(nu0.min()):.4f}, {float(nu0.max()):.4f}]\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Step 4: End-to-end training through the containerized solver\n", "\n", "The training loss is:\n", "\n", "$$\\mathcal{L}(\\theta) = \\frac{1}{M} \\sum_{i=1}^{M} \\left\\| u_{\\text{solver}}(u_0^{(i)}; \\nu_\\theta) - u_{\\text{data}}^{(i)} \\right\\|^2$$\n", "\n", "where $\\nu_\\theta$ is the neural viscosity closure with parameters $\\theta$, and $u_{\\text{solver}}$ runs the full Burgers' equation with $\\nu_\\theta$ called at every timestep.\n", "\n", "The time-stepping loop calls the network directly and the solver via `apply_tesseract`:\n", "\n", "```python\n", "for each timestep:\n", " nu = closure(u, dudx, x) # plain torch\n", " u = apply_tesseract(solver, {u, nu, dt})[\"u_next\"] # HTTP call to container\n", "```\n", "\n", "`loss.backward()` differentiates through the entire loop: through every timestep, through a VJP HTTP request to the container at each step, and into the network weights, all automatically." ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Testing forward pass + gradient...\n", " Initial loss: 1.283973e-04\n", " Gradient norm: 1.046508e-03\n", " Gradients flow: loss -> solver VJP (HTTP) -> network weights.\n" ] } ], "source": [ "def solve_with_closure(u0, closure, dt, n_steps):\n", " \"\"\"Run the full time-stepping loop, calling the served solver each step.\n", "\n", " The closure is plain in-process PyTorch; the solver is the containerized\n", " differentiable layer, reached over HTTP. In production this same loop would\n", " drive a Fortran solver with an adjoint, with only the served image changing.\n", " \"\"\"\n", " u = u0\n", " for _step in range(n_steps):\n", " dudx = torch.zeros_like(u)\n", " dudx[1:-1] = (u[2:] - u[:-2]) / (2 * DX)\n", "\n", " nu = closure(u, dudx, X) # closure: predict viscosity (native torch)\n", "\n", " # Solver: one explicit Euler step, executed in the container\n", " solver_out = apply_tesseract(solver_tess, {\"u\": u, \"nu\": nu, \"dt\": dt})\n", " u = solver_out[\"u_next\"]\n", " return u\n", "\n", "\n", "def loss_batch(closure, ics, targets):\n", " \"\"\"Mean MSE over a batch of initial conditions.\"\"\"\n", " preds = torch.stack(\n", " [solve_with_closure(ics[i], closure, DT, N_STEPS) for i in range(ics.shape[0])]\n", " )\n", " return torch.mean((preds - targets) ** 2)\n", "\n", "\n", "# Verify the forward pass + gradient flow\n", "print(\"Testing forward pass + gradient...\")\n", "l0 = loss_batch(closure, train_ics, train_targets)\n", "l0.backward()\n", "grad_norm = torch.sqrt(\n", " sum(p.grad.pow(2).sum() for p in closure.parameters() if p.grad is not None)\n", ")\n", "print(f\" Initial loss: {float(l0.detach()):.6e}\")\n", "print(f\" Gradient norm: {float(grad_norm):.6e}\")\n", "print(\" Gradients flow: loss -> solver VJP (HTTP) -> network weights.\")\n", "closure.zero_grad()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Gradient validation against finite differences\n", "\n", "Correctness check: the AD gradient (loss → solver VJP over HTTP → network) matches finite differences to high precision." ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ " AD FD Rel. Error\n", " -6.627976e-07 -6.627976e-07 6.68e-10\n" ] } ], "source": [ "# Finite difference check on a single weight element of the first layer.\n", "ics_sub = train_ics[:2]\n", "tgt_sub = train_targets[:2]\n", "\n", "# AD gradient\n", "closure.zero_grad()\n", "loss_val = loss_batch(closure, ics_sub, tgt_sub)\n", "loss_val.backward()\n", "\n", "w = closure.net[0].weight.data # first Linear layer (raw tensor, no autograd)\n", "idx = (0, 0)\n", "ad = float(closure.net[0].weight.grad[idx])\n", "\n", "# Finite difference\n", "eps = 1e-5\n", "orig = w[idx].item()\n", "with torch.no_grad():\n", " w[idx] = orig + eps\n", " l_plus = float(loss_batch(closure, ics_sub, tgt_sub))\n", " w[idx] = orig - eps\n", " l_minus = float(loss_batch(closure, ics_sub, tgt_sub))\n", " w[idx] = orig # restore\n", "\n", "fd = (l_plus - l_minus) / (2 * eps)\n", "rel_err = abs(ad - fd) / (abs(fd) + 1e-30)\n", "print(f\"{'AD':>14s} {'FD':>14s} {'Rel. Error':>12s}\")\n", "print(f\"{ad:14.6e} {fd:14.6e} {rel_err:12.2e}\")\n", "closure.zero_grad()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Training loop" ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Training neural closure through the solver...\n", " Epoch 0: train loss = 1.2840e-04, test loss = 8.8252e-06\n", " Epoch 10: train loss = 1.1664e-05, test loss = 5.9784e-06\n", " Epoch 20: train loss = 1.1159e-05, test loss = 6.0776e-06\n", " Epoch 30: train loss = 9.1783e-06, test loss = 4.2518e-06\n", " Epoch 40: train loss = 6.9523e-06, test loss = 3.9642e-06\n", " Epoch 50: train loss = 5.6923e-06, test loss = 3.4751e-06\n", " Epoch 60: train loss = 4.7374e-06, test loss = 2.9877e-06\n", " Epoch 70: train loss = 4.0577e-06, test loss = 2.7151e-06\n", " Epoch 80: train loss = 3.6290e-06, test loss = 2.7027e-06\n", " Epoch 90: train loss = 3.3559e-06, test loss = 2.6873e-06\n", " Epoch 99: train loss = 3.1723e-06, test loss = 2.7432e-06\n", "\n", "Final train loss: 3.1723e-06\n", "Final test loss: 2.7432e-06\n" ] } ], "source": [ "# Re-initialize for a clean training run\n", "torch.manual_seed(1)\n", "closure = ViscosityNet()\n", "optimizer = torch.optim.Adam(closure.parameters(), lr=LR)\n", "\n", "train_losses = []\n", "test_losses = []\n", "\n", "print(\"Training neural closure through the solver...\")\n", "for epoch in range(N_EPOCHS):\n", " optimizer.zero_grad()\n", " train_loss = loss_batch(closure, train_ics, train_targets)\n", " train_loss.backward()\n", " optimizer.step()\n", " train_losses.append(float(train_loss.detach()))\n", "\n", " if epoch % 10 == 0 or epoch == N_EPOCHS - 1:\n", " with torch.no_grad():\n", " test_loss = float(loss_batch(closure, test_ics, test_targets))\n", " test_losses.append((epoch, test_loss))\n", " print(\n", " f\" Epoch {epoch:4d}: train loss = {train_losses[-1]:.4e}, \"\n", " f\"test loss = {test_loss:.4e}\"\n", " )\n", "\n", "print(f\"\\nFinal train loss: {train_losses[-1]:.4e}\")\n", "print(f\"Final test loss: {test_losses[-1][1]:.4e}\")" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "fig, axes = plt.subplots(1, 2, figsize=(12, 4))\n", "\n", "axes[0].semilogy(train_losses, label=\"Train\")\n", "test_epochs, test_vals = zip(*test_losses, strict=False)\n", "axes[0].semilogy(test_epochs, test_vals, \"o-\", label=\"Test\")\n", "axes[0].set_xlabel(\"Epoch\")\n", "axes[0].set_ylabel(\"MSE Loss\")\n", "axes[0].set_title(\"Training loss\")\n", "axes[0].legend()\n", "axes[0].grid(True, alpha=0.3)\n", "\n", "# Compare learned viscosity to true viscosity, averaged over training ICs.\n", "nu_samples = []\n", "with torch.no_grad():\n", " for ic in train_ics:\n", " dudx = torch.zeros_like(ic)\n", " dudx[1:-1] = (ic[2:] - ic[:-2]) / (2 * DX)\n", " nu_samples.append(closure(ic, dudx, X))\n", "nu_samples = torch.stack(nu_samples)\n", "nu_mean = nu_samples.mean(dim=0)\n", "nu_std = nu_samples.std(dim=0)\n", "\n", "x_np = X.numpy()\n", "axes[1].plot(x_np, nu_true.numpy(), \"k-\", linewidth=2, label=\"True viscosity\")\n", "axes[1].plot(x_np, nu_mean.numpy(), \"r--\", linewidth=2, label=\"Learned (mean over ICs)\")\n", "axes[1].fill_between(\n", " x_np,\n", " (nu_mean - nu_std).numpy(),\n", " (nu_mean + nu_std).numpy(),\n", " color=\"r\",\n", " alpha=0.15,\n", " label=\"Learned (std)\",\n", ")\n", "axes[1].set_xlabel(\"x\")\n", "axes[1].set_ylabel(r\"$\\nu$\")\n", "axes[1].set_title(\"Recovered viscosity profile\")\n", "axes[1].legend(fontsize=9)\n", "axes[1].set_ylim(bottom=0)\n", "axes[1].grid(True, alpha=0.3)\n", "\n", "plt.tight_layout()\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Step 5: Baselines\n", "\n", "We compare three approaches:\n", "\n", "| Model | Description |\n", "|---|---|\n", "| **Constant viscosity** | Burgers' solver with $\\nu = \\nu_0$ (the standard \"wrong\" closure) |\n", "| **Direct ML** | MLP trained to map $u_0 \\to u_\\text{final}$ directly, no physics |\n", "| **Learned closure** | Neural $\\nu(u, \\partial u/\\partial x, x)$ trained through the solver (this demo) |\n", "\n", "The closure here is **not trained to convergence**: the CI defaults keep the run to a few minutes, so the recovered viscosity profile is approximate. As a result the learned closure decisively beats the pure-ML baseline but only edges out the constant-viscosity baseline, which stays competitive at this short budget. Set `FULL_RUN=1` for the longer run that sharpens the profile and widens that gap. The point of this demo is the composition pattern, not a state-of-the-art closure." ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Constant viscosity test MSE: 6.1354e-06\n", "\n", "Training direct ML baseline...\n", " Epoch 0: train = 3.3670e-01, test = 2.0239e-01\n", " Epoch 20: train = 3.7105e-02, test = 1.1341e-01\n", " Epoch 40: train = 4.0102e-03, test = 7.2236e-02\n", " Epoch 60: train = 1.2106e-03, test = 7.4998e-02\n", " Epoch 80: train = 6.3123e-04, test = 7.7559e-02\n", "\n", "Direct ML test MSE: 7.8228e-02\n", "Learned closure test MSE: 2.7432e-06\n", "\n", "Model Test MSE\n", "----------------------------------------\n", "Constant viscosity 6.1354e-06\n", "Direct ML 7.8228e-02\n", "Learned closure 2.7432e-06\n" ] } ], "source": [ "# --- Baseline 1: Constant (wrong) viscosity ---\n", "nu_const = NU_0 * torch.ones(N)\n", "with torch.no_grad():\n", " const_preds_test = torch.stack(\n", " [burgers_reference(ic, nu_const, DT, N_STEPS) for ic in test_ics]\n", " )\n", "const_mse = float(torch.mean((const_preds_test - test_targets) ** 2))\n", "print(f\"Constant viscosity test MSE: {const_mse:.4e}\")\n", "\n", "\n", "# --- Baseline 2: Direct ML (MLP mapping u0 -> u_final, no physics) ---\n", "class DirectNet(nn.Module):\n", " \"\"\"Pure ML: maps the whole field u0 directly to u_final.\"\"\"\n", "\n", " def __init__(self, n=N, hidden=128):\n", " super().__init__()\n", " self.net = nn.Sequential(\n", " nn.Linear(n, hidden),\n", " nn.Tanh(),\n", " nn.Linear(hidden, hidden),\n", " nn.Tanh(),\n", " nn.Linear(hidden, 64),\n", " nn.Tanh(),\n", " nn.Linear(64, n),\n", " )\n", "\n", " def forward(self, u0):\n", " return self.net(u0)\n", "\n", "\n", "def direct_loss(model, ics, targets):\n", " preds = model(ics)\n", " return torch.mean((preds - targets) ** 2)\n", "\n", "\n", "torch.manual_seed(2)\n", "direct = DirectNet()\n", "direct_opt = torch.optim.Adam(direct.parameters(), lr=1e-3)\n", "\n", "print(\"\\nTraining direct ML baseline...\")\n", "for epoch in range(DIRECT_EPOCHS):\n", " direct_opt.zero_grad()\n", " dl = direct_loss(direct, train_ics, train_targets)\n", " dl.backward()\n", " direct_opt.step()\n", " if epoch % 20 == 0:\n", " with torch.no_grad():\n", " test_dl = float(direct_loss(direct, test_ics, test_targets))\n", " print(\n", " f\" Epoch {epoch:4d}: train = {float(dl.detach()):.4e}, test = {test_dl:.4e}\"\n", " )\n", "\n", "with torch.no_grad():\n", " direct_test_mse = float(direct_loss(direct, test_ics, test_targets))\n", "print(f\"\\nDirect ML test MSE: {direct_test_mse:.4e}\")\n", "\n", "# --- Learned closure (already trained above) ---\n", "with torch.no_grad():\n", " learned_test_mse = float(loss_batch(closure, test_ics, test_targets))\n", "print(f\"Learned closure test MSE: {learned_test_mse:.4e}\")\n", "\n", "print(f\"\\n{'Model':<25s} {'Test MSE':>12s}\")\n", "print(\"-\" * 40)\n", "print(f\"{'Constant viscosity':<25s} {const_mse:12.4e}\")\n", "print(f\"{'Direct ML':<25s} {direct_test_mse:12.4e}\")\n", "print(f\"{'Learned closure':<25s} {learned_test_mse:12.4e}\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Step 6: Solution comparison on test data\n", "\n", "For unseen initial conditions, we compare the three models' predictions against the ground truth." ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "n_show = min(4, N_TEST)\n", "ncols = 2 if n_show > 1 else 1\n", "nrows = (n_show + ncols - 1) // ncols\n", "fig, axes = plt.subplots(nrows, ncols, figsize=(6 * ncols, 4.5 * nrows), squeeze=False)\n", "\n", "with torch.no_grad():\n", " direct_preds = direct(test_ics)\n", " for idx in range(n_show):\n", " ax = axes[idx // ncols][idx % ncols]\n", " ic = test_ics[idx]\n", " target = test_targets[idx]\n", "\n", " const_pred = burgers_reference(ic, nu_const, DT, N_STEPS)\n", " direct_pred = direct_preds[idx]\n", " learned_pred = solve_with_closure(ic, closure, DT, N_STEPS)\n", "\n", " ax.plot(x_np, target.numpy(), \"k-\", linewidth=2, label=\"Ground truth\")\n", " ax.plot(\n", " x_np,\n", " const_pred.numpy(),\n", " \"b--\",\n", " linewidth=1.5,\n", " alpha=0.7,\n", " label=\"Constant viscosity\",\n", " )\n", " ax.plot(\n", " x_np, direct_pred.numpy(), \"g:\", linewidth=1.5, alpha=0.7, label=\"Direct ML\"\n", " )\n", " ax.plot(\n", " x_np, learned_pred.numpy(), \"r-\", linewidth=1.5, label=\"Learned closure\"\n", " )\n", " ax.set_xlabel(\"x\")\n", " ax.set_ylabel(\"u\")\n", " ax.set_title(f\"Test case {idx + 1}\")\n", " ax.grid(True, alpha=0.3)\n", " if idx == 0:\n", " ax.legend(fontsize=9)\n", "\n", "plt.suptitle(\n", " \"Solution predictions on unseen initial conditions\", fontsize=13, fontweight=\"bold\"\n", ")\n", "plt.tight_layout()\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Tear down the solver\n", "\n", "Stop the solver container to free resources." ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [], "source": [ "# Tear down Tesseract after use to prevent resource leaks\n", "solver_tess.teardown()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Takeaways\n", "\n", "In this tutorial, we trained a neural viscosity closure end-to-end through a containerized Burgers' equation solver. Here are the key points:\n", "\n", "1. **Containerized solver, native ML.** The solver runs in a Docker container served as a Tesseract; the closure is a plain in-process `torch.nn.Module`. The two compose in an outer time-stepping loop, with `apply_tesseract` bridging the container boundary at each step.\n", "\n", "2. **End-to-end gradients through the container.** Because Tesseract-Torch registers the served solver as a PyTorch autograd function, `loss.backward()` dispatches the solver's VJP over HTTP and flows the gradient into the network, with no hand-coded plumbing. We validated these gradients against finite differences.\n", "\n", "3. **Physics structure helps.** Even with the short CI training budget, the learned closure beats the pure-ML baseline by orders of magnitude and edges out the constant-viscosity baseline, recovering the shape of the true viscosity profile from solution data alone. Training longer with `FULL_RUN=1` widens the margin over the constant baseline.\n", "\n", "4. **Composability across a container boundary.** The closure (plain torch) and the solver (a different image exposing the same $(u, \\nu, dt) \\to u_\\text{next}$ contract) can be swapped independently. The training loop is identical regardless of what language the solver is written in or how its adjoint is produced.\n", "\n", "5. **Mind the overhead.** Every solver call here is an HTTP round-trip, and a single explicit Euler step is cheap, so this demo is overhead-bound. Tesseract pays off when the solver step is expensive relative to the round-trip (a real CFD/FEM step, a remote or cross-language solver), or when you can space closure updates out over several steps. See the [performance guide](../concepts/performance.md) for the overhead breakdown and rules of thumb.\n", "\n", "### What's next\n", "\n", "This demo is the basic recipe. Real-world closures usually need more than a vanilla MLP trained with Adam through a single-step solver. Here are the directions to explore from here:\n", "\n", "- **Wrap a legacy solver.** Replace the Burgers' Tesseract with a Fortran/C++ solver that exposes an adjoint (for example via [Enzyme](https://enzyme.mit.edu/)). The closure training loop above works unchanged; only the image name changes.\n", "- **Train smarter.** For small closures with a least-squares loss, a Gauss--Newton or L-BFGS solve on the point-wise residual Jacobian (forward-mode AD) can converge far faster than Adam and more readily unlearn spurious feature dependence.\n", "- **Get more data for free.** Chain the solver to emit several snapshots per simulation and train on all of them: extra supervision at no extra simulation cost, and a natural reason to chunk the rollout into shorter segments.\n", "- **Go to longer rollouts.** When the closure is called over many steps, use gradient checkpointing to keep memory bounded, and batch/vmap solver calls to amortize HTTP overhead (batching multiple initial conditions per request).\n", "- **Tackle a real closure.** Swap the Burgers' equation for a turbulence model, a climate sub-grid scheme, or a materials constitutive law.\n", "- **Explore other demos.** See the [CFD optimization](cfd-optimization.ipynb), [FEM shape optimization](fem-shape-optimization.ipynb), and [data assimilation](data-assimilation.ipynb) demos for the JAX side of the same composition pattern.\n", "\n", "Built something with these? We'd love to see it. Share your closure on the [Tesseract Community Forum](https://si-tesseract.discourse.group/), and tell us where Tesseract's overhead or UX got in your way so we can improve it." ] } ], "metadata": { "kernelspec": { "display_name": "science (3.12.7)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.12.7" } }, "nbformat": 4, "nbformat_minor": 4 }