From 7a3e06da1f629d1186f1f66dc3055f95a121ae40 Mon Sep 17 00:00:00 2001
From: Gard Spreemann <gspr@nonempty.org>
Date: Sat, 4 May 2019 15:24:37 +0200
Subject: Make the Laplacian changeable with each forward call.

---
 scnn/scnn.py | 33 +++++++++++++++++----------------
 1 file changed, 17 insertions(+), 16 deletions(-)

diff --git a/scnn/scnn.py b/scnn/scnn.py
index 91f9706..f46cc56 100644
--- a/scnn/scnn.py
+++ b/scnn/scnn.py
@@ -12,21 +12,17 @@ def coo2tensor(A):
     return torch.sparse_coo_tensor(idxs, vals, size = A.shape, requires_grad = False)
 
 class SimplicialConvolution(nn.Module):
-    def __init__(self, K, L, C_in, C_out, enable_bias = True, variance = 1.0, groups = 1):
+    def __init__(self, K, C_in, C_out, enable_bias = True, variance = 1.0, groups = 1):
         assert groups == 1, "Only groups = 1 is currently supported."
         super().__init__()
 
-        assert(len(L.shape) == 2)
-        assert(L.shape[0] == L.shape[1])
         assert(C_in > 0)
         assert(C_out > 0)
         assert(K > 0)
         
-        self.M = L.shape[0]
         self.C_in = C_in
         self.C_out = C_out
         self.K = K
-        self.L = L # Don't modify afterwards!
         self.enable_bias = enable_bias
 
         self.theta = nn.parameter.Parameter(variance*torch.randn((self.C_out, self.C_in, self.K)))
@@ -35,13 +31,16 @@ class SimplicialConvolution(nn.Module):
         else:
             self.bias = 0.0
             
-    def forward(self, x):
+    def forward(self, L, x):
+        assert(len(L.shape) == 2)
+        assert(L.shape[0] == L.shape[1])
+                
         (B, C_in, M) = x.shape
-       
-        assert(M == self.M)
+     
+        assert(M == L.shape[0])
         assert(C_in == self.C_in)
 
-        X = scnn.chebyshev.assemble(self.K, self.L, x)
+        X = scnn.chebyshev.assemble(self.K, L, x)
         y = torch.einsum("bimk,oik->bom", (X, self.theta))
         assert(y.shape == (B, self.C_out, M))
 
@@ -55,16 +54,14 @@ class SimplicialConvolution(nn.Module):
 # Note: You can use this for a adjoints of coboundaries too. Just feed
 # a transposed D.
 class Coboundary(nn.Module):
-    def __init__(self, D, C_in, C_out, enable_bias = True, variance = 1.0):
+    def __init__(self, C_in, C_out, enable_bias = True, variance = 1.0):
         super().__init__()
 
-        assert(len(D.shape) == 2)
         assert(C_in > 0)
         assert(C_out > 0)
 
         self.C_in = C_in
         self.C_out = C_out
-        self.D = D # Don't modify.
         self.enable_bias = enable_bias
 
         self.theta = nn.parameter.Parameter(variance*torch.randn((self.C_out, self.C_in)))
@@ -73,11 +70,15 @@ class Coboundary(nn.Module):
         else:
             self.bias = 0.0
 
-    def forward(self, x):
+    def forward(self, D, x):
+        assert(len(D.shape) == 2)
+        
         (B, C_in, M) = x.shape
-        assert(self.D.shape[1] == M)
+        
+        assert(D.shape[1] == M)
         assert(C_in == self.C_in)
-        N = self.D.shape[0]
+        
+        N = D.shape[0]
 
         # This is essentially the equivalent of chebyshev.assemble for
         # the convolutional modules.
@@ -85,7 +86,7 @@ class Coboundary(nn.Module):
         for b in range(0, B):
             X12 = []
             for c_in in range(0, self.C_in):
-                X12.append(self.D.mm(x[b, c_in, :].unsqueeze(1)).transpose(0,1)) # D.mm(x[b, c_in, :]) has shape Nx1
+                X12.append(D.mm(x[b, c_in, :].unsqueeze(1)).transpose(0,1)) # D.mm(x[b, c_in, :]) has shape Nx1
             X12 = torch.cat(X12, 0)
 
             assert(X12.shape == (self.C_in, N))
-- 
cgit v1.2.3