fix import

src/orisys/HHD.py

@@ -30,9 +30,9 @@ from scipy import signal, spatial
 # from .UnstructuredGrid import UnstructuredGrid
 # import matplotlib.pyplot as plt
 
-# Try to import CuPy for GPU acceleration
 try:
     import cupy as cp
+    import cupyx.scipy.signal as cp_signal
     CUPY_AVAILABLE = True
     
     # Try to import CuPy streams for parallel execution
@@ -65,6 +65,7 @@ try:
         
 except ImportError:
     cp = None
+    cp_signal = None
     CUPY_AVAILABLE = False
     STREAMS_AVAILABLE = False
     KERNEL_AVAILABLE = False
@@ -130,14 +131,14 @@ class nHHD(object):
                 print( 'div =', self.div.shape, self.div.min(), self.div.max())
                 print( 'curlw =', self.curlw.shape, self.curlw.min(), self.curlw.max())
 
-            # compute potentials
-            self.nD = self.psolver.solve_gpu(self.div, verbose > 0) #1)
-            if verbose > 0:
-                print( 'nD =', self.nD.shape, self.nD.min(), self.nD.max())
+            # # compute potentials
+            # self.nD = self.psolver.solve_gpu(self.div, verbose > 0) #1)
+            # if verbose > 0:
+            #     print( 'nD =', self.nD.shape, self.nD.min(), self.nD.max())
 
-            self.nRu = self.psolver.solve_gpu(self.curlw, verbose > 0) #1)
-            if verbose > 0:
-                print( 'nR =', self.nRu.shape, self.nRu.min(), self.nRu.max())
+            # self.nRu = self.psolver.solve_gpu(self.curlw, verbose > 0) #1)
+            # if verbose > 0:
+            #     print( 'nR =', self.nRu.shape, self.nRu.min(), self.nRu.max())
 
             # compute fields as gradients of potentials
             self.d = self.mesh.gradient(self.nD)
@@ -398,8 +399,8 @@ class nHHD(object):
         nRu_gpu = self._nRu_gpu
         
         # Use reduction kernels for better performance
-        # Compute Sn using L1 norm (sum of absolute values)
-        Sn = float(cp.linalg.norm(d_gpu.ravel(), ord=1))
+        # Compute Sn as L1 sum directly on GPU (faster than linalg.norm for this case)
+        Sn = float(cp.sum(cp.abs(d_gpu)))
         
         # Compute Stx, Sty
         Stx = float(cp.sum(h_gpu[:, :, 0]))
@@ -881,19 +882,19 @@ class StructuredGrid(object):
     # --------------------------------------------------------------------------
 
 
-import numpy
-from scipy import signal, spatial
-# from .timer import Timer
+# import numpy
+# from scipy import signal, spatial
+# # from .timer import Timer
 
-# Try to import CuPy for GPU acceleration
-try:
-    import cupy as cp
-    import cupyx.scipy.signal as cp_signal
-    CUPY_AVAILABLE = True
-except ImportError:
-    cp = None
-    cp_signal = None
-    CUPY_AVAILABLE = False
+# # Try to import CuPy for GPU acceleration
+# try:
+#     import cupy as cp
+#     import cupyx.scipy.signal as cp_signal
+#     CUPY_AVAILABLE = True
+# except ImportError:
+#     cp = None
+#     cp_signal = None
+#     CUPY_AVAILABLE = False
 
 
 '''
@@ -1322,7 +1323,7 @@ class PoissonSolver(object):
                 p_gpu *= self._gpu_cache['gdx_prod']               
 
                 return p_gpu  # Return CuPy array
-
+    
 
     # def solve_gpu_batch(self, f_list, verbose=False):
     #     """