workshop: pdset.preprocess for DPC+rotation; pdset ready for v2 ptycho

notebooks/berk_workshop_v1.ipynb

@@ -248,9 +248,15 @@
    "id": "50bcd0fe",
    "metadata": {},
    "source": [
-    "## Step 3 — DPC via `CenterOfMassOriginModel`\n",
+    "## Step 3 — DPC + scan-detector rotation in one upstream call\n",
     "\n",
-    "Per-scan-position centroid (CoM) on the GPU. Two-cell call, two interactive panels."
+    "`PtychographyDatasetRaster.preprocess(plot_rotation=True, plot_com=True)` runs\n",
+    "CoM, fits the origin background, finds the scan-detector rotation by curl\n",
+    "minimization, **and plots both** — all in one upstream call. Bonus: the `pdset`\n",
+    "this returns is ready for iterative ptycho later (v2).\n",
+    "\n",
+    "Takes ~30 s on T4 (the intensity normalization step is the bulk; we use the\n",
+    "plots + the rotation now)."
    ]
   },
   {
@@ -285,69 +291,17 @@
     }
    ],
    "source": [
-    "from quantem.diffractive_imaging import CenterOfMassOriginModel\n",
-    "\n",
-    "com_model = CenterOfMassOriginModel.from_dataset(dset, device=\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
-    "com_model.calculate_origin()\n",
-    "com = com_model.origin_measured.view(*dset.shape[:2], 2).cpu().numpy()\n",
-    "\n",
-    "quantem.widget.Show2D(\n",
-    "    [com[..., 0], com[..., 1]],\n",
-    "    labels=[\"CoM row (qx)\", \"CoM col (qy)\"],\n",
-    "    sampling=meta[\"sampling\"][:2], units=meta[\"units\"][:2],\n",
-    "    cmap=\"RdBu_r\", link_contrast=False,\n",
-    ")"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "e8050ac2",
-   "metadata": {},
-   "source": [
-    "## Step 3b — Scan-detector rotation from CoM curl\n",
-    "\n",
-    "`DirectPtychography` needs the rotation between the scan axes and the detector\n",
-    "axes. Find it by sweeping rotation angles and minimizing the curl of the\n",
-    "rotated CoM field. (Auto-estimated internally by `from_dataset4d` too — this\n",
-    "cell exposes the curve so you can see WHERE the minimum lands.)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "414e6538",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import matplotlib.pyplot as plt\n",
-    "\n",
-    "# Fit origin (needed by estimate_detector_rotation) + run estimate.\n",
-    "com_model.fit_origin_background()\n",
-    "com_model.estimate_detector_rotation()\n",
-    "best_rotation_deg = com_model.detector_rotation_deg\n",
-    "best_rotation_rad = np.deg2rad(best_rotation_deg)\n",
-    "print(f\"detector rotation = {best_rotation_deg:.2f}°  ({best_rotation_rad:.4f} rad)\")\n",
-    "\n",
-    "# Recompute curl-vs-angle for the plot.\n",
-    "angles_deg = torch.arange(-89, 90, 1, device=com_model.device, dtype=torch.float32)\n",
-    "com_t = com_model.origin_measured.view(*dset.shape[:2], 2)\n",
-    "com_n = com_t - com_t.mean(dim=(0, 1))\n",
-    "rot_rad = torch.deg2rad(angles_deg)[:, None, None]\n",
-    "rx = torch.cos(rot_rad) * com_n[None, ..., 0] - torch.sin(rot_rad) * com_n[None, ..., 1]\n",
-    "ry = torch.sin(rot_rad) * com_n[None, ..., 0] + torch.cos(rot_rad) * com_n[None, ..., 1]\n",
-    "grad_xy = torch.gradient(rx, dim=-1)[0]\n",
-    "grad_yx = torch.gradient(ry, dim=-2)[0]\n",
-    "curl = torch.mean(torch.abs(grad_yx - grad_xy), dim=(-2, -1)).cpu().numpy()\n",
-    "angles = angles_deg.cpu().numpy()\n",
-    "\n",
-    "fig, ax = plt.subplots(figsize=(6, 3.5))\n",
-    "ax.plot(angles, curl, lw=1.5)\n",
-    "ax.axvline(best_rotation_deg, color=\"red\", lw=1, label=f\"{best_rotation_deg:.1f}°\")\n",
-    "ax.set_xlabel(\"rotation angle (deg)\")\n",
-    "ax.set_ylabel(\"mean |curl(CoM)|\")\n",
-    "ax.set_title(\"Scan-detector rotation from CoM curl\")\n",
-    "ax.legend()\n",
-    "plt.tight_layout(); plt.show()"
+    "from quantem.diffractive_imaging import PtychographyDatasetRaster\n",
+    "\n",
+    "pdset = PtychographyDatasetRaster.from_dataset4dstem(dset)\n",
+    "pdset.preprocess(\n",
+    "    com_fit_function=\"constant\",\n",
+    "    plot_rotation=True,\n",
+    "    plot_com=True,\n",
+    "    probe_energy=meta[\"voltage_kV\"] * 1e3,\n",
+    ")\n",
+    "best_rotation_rad = pdset.com_rotation_rad\n",
+    "print(f\"detector rotation = {np.rad2deg(best_rotation_rad):.2f}°  ({best_rotation_rad:.4f} rad)\")"
    ]
   },
   {