Merge pull request #178 from eigenvivek/intrinsics

Support autodifferentiability for intrinsic parameters

Author: eigenvivek

diffdrr/detector.py

@@ -57,22 +57,29 @@ def __init__(
 @patch
 def _initialize_carm(self: Detector):
     """Initialize the default position for the source and detector plane."""
-    # Initialize the source on the x-axis
-    source = torch.tensor([[self.sdr, 0.0, 0.0]])
+    try:
+        device = self.sdr.device
+    except AttributeError:
+        device = torch.device("cpu")
 
-    # Initialize the center of the detector plane on the negative x-axis
-    center = torch.tensor([[-self.sdr, 0.0, 0.0]])
+    # Initialize the source on the x-axis and the center of the detector plane on the negative x-axis
+    source = torch.tensor([[1.0, 0.0, 0.0]], device=device) * self.sdr
+    center = torch.tensor([[-1.0, 0.0, 0.0]], device=device) * self.sdr
 
     # Use the standard basis for the detector plane
-    basis = torch.tensor([[0.0, 1.0, 0.0], [0.0, 0.0, 1.0]])
+    basis = torch.tensor([[0.0, 1.0, 0.0], [0.0, 0.0, 1.0]], device=device)
 
     # Construct the detector plane with different offsets for even or odd heights
     h_off = 1.0 if self.height % 2 else 0.5
     w_off = 1.0 if self.width % 2 else 0.5
 
     # Construct equally spaced points along the basis vectors
-    t = (torch.arange(-self.height // 2, self.height // 2) + h_off) * self.delx
-    s = (torch.arange(-self.width // 2, self.width // 2) + w_off) * self.dely
+    t = (
+        torch.arange(-self.height // 2, self.height // 2, device=device) + h_off
+    ) * self.delx
+    s = (
+        torch.arange(-self.width // 2, self.width // 2, device=device) + w_off
+    ) * self.dely
     if self.reverse_x_axis:
         s = -s
     coefs = torch.cartesian_prod(t, s).reshape(-1, 2)

notebooks/api/02_detector.ipynb

@@ -128,22 +128,25 @@
     "@patch\n",
     "def _initialize_carm(self: Detector):\n",
     "    \"\"\"Initialize the default position for the source and detector plane.\"\"\"\n",
-    "    # Initialize the source on the x-axis\n",
-    "    source = torch.tensor([[self.sdr, 0.0, 0.0]])\n",
-    "\n",
-    "    # Initialize the center of the detector plane on the negative x-axis\n",
-    "    center = torch.tensor([[-self.sdr, 0.0, 0.0]])\n",
+    "    try:\n",
+    "        device = self.sdr.device\n",
+    "    except AttributeError:\n",
+    "        device = torch.device(\"cpu\")\n",
+    "    \n",
+    "    # Initialize the source on the x-axis and the center of the detector plane on the negative x-axis\n",
+    "    source = torch.tensor([[1.0, 0.0, 0.0]], device=device) * self.sdr\n",
+    "    center = torch.tensor([[-1.0, 0.0, 0.0]], device=device) * self.sdr\n",
     "\n",
     "    # Use the standard basis for the detector plane\n",
-    "    basis = torch.tensor([[0.0, 1.0, 0.0], [0.0, 0.0, 1.0]])\n",
+    "    basis = torch.tensor([[0.0, 1.0, 0.0], [0.0, 0.0, 1.0]], device=device)\n",
     "\n",
     "    # Construct the detector plane with different offsets for even or odd heights\n",
     "    h_off = 1.0 if self.height % 2 else 0.5\n",
     "    w_off = 1.0 if self.width % 2 else 0.5\n",
     "\n",
     "    # Construct equally spaced points along the basis vectors\n",
-    "    t = (torch.arange(-self.height // 2, self.height // 2) + h_off) * self.delx\n",
-    "    s = (torch.arange(-self.width // 2, self.width // 2) + w_off) * self.dely\n",
+    "    t = (torch.arange(-self.height // 2, self.height // 2, device=device) + h_off) * self.delx\n",
+    "    s = (torch.arange(-self.width // 2, self.width // 2, device=device) + w_off) * self.dely\n",
     "    if self.reverse_x_axis:\n",
     "        s = -s\n",
     "    coefs = torch.cartesian_prod(t, s).reshape(-1, 2)\n",