DeepGPR 0.0.13__tar.gz → 0.0.14__tar.gz

{deepgpr-0.0.13 → deepgpr-0.0.14}/PKG-INFO

@@ -1,19 +1,25 @@
 Metadata-Version: 2.4
 Name: DeepGPR
-Version: 0.0.13
+Version: 0.0.14
 Summary: PyTorch and CUDA for GPR FWI
 Author-email: Lei Liu <liulei990222@gmail.com>
+License-Expression: MIT
+Project-URL: Homepage, https://github.com/songc0a/DeepGPR
+Project-URL: Repository, https://github.com/songc0a/DeepGPR
 Classifier: Programming Language :: Python :: 3
 Classifier: Operating System :: Microsoft :: Windows
 Classifier: Operating System :: POSIX :: Linux
+Classifier: Operating System :: MacOS :: MacOS X
 Classifier: Topic :: Scientific/Engineering
 Classifier: Topic :: Scientific/Engineering :: Physics
 Requires-Python: >=3.8
 Description-Content-Type: text/markdown
+License-File: license
 Requires-Dist: numpy
 Requires-Dist: scipy
 Requires-Dist: matplotlib
 Requires-Dist: torch
+Dynamic: license-file
 
 # DeepGPR
 
@@ -144,7 +150,8 @@ def compute(device, dx=None, dt=None,
             model_gradient_sampling_interval=1,
             use_async_offload=False,
             fdtd_order=2,
-            mode=2):
+            mode=2,
+            debug=False):
 ```
 ## 📥 Input Parameters
 ### 1. Basic Physics & Grid Parameters
@@ -156,6 +163,7 @@ def compute(device, dx=None, dt=None,
 | **`dt`** | `float` | Time step size. **Note**: Must strictly satisfy the CFL (Courant-Friedrichs-Lewy) stability condition, or an exception will be raised. Typically in seconds (s). |
 | **`fdtd_order`** | `int` | Spatial finite-difference order used by the FDTD field updates. Supported values are `2`, `4`, and `8`; default is `2` for compatibility with earlier versions. |
 | **`mode`** | `int` | FWI gradient mode. `2` keeps the previous Ez-only model-gradient calculation. `3` uses Ex, Ey, and Ez electric-field contributions for relative permittivity and conductivity gradients. |
+| **`debug`** | `bool` | Runs expensive NaN/Inf and zero-field validation checks when `True`. The default `False` keeps these checks disabled for faster production runs. |
 ### 2. Medium Model Parameters
 
 This section defines the electromagnetic properties of the simulation space. For 2D simulations, set `nz=1`.
@@ -212,7 +220,14 @@ cc -std=c99 -O3 -fopenmp -fPIC -shared -o src/DeepGPR/lib/deepgpr_cpu.so src/Dee
 # macOS
 brew install libomp
 LIBOMP_PREFIX="$(brew --prefix libomp)"
-cc -std=c99 -O3 -Xpreprocessor -fopenmp -DDEEPGPR_USE_OPENMP -I"$LIBOMP_PREFIX/include" -Wl,-rpath,"$LIBOMP_PREFIX/lib" -L"$LIBOMP_PREFIX/lib" -fPIC -shared -o src/DeepGPR/lib/deepgpr_cpu.dylib src/DeepGPR/lib/deepgpr_cpu.c -lomp
+LIBOMP_RUNTIME_NAME="/opt/llvm-openmp/lib/libomp.dylib"
+cc -std=c99 -O3 -Xpreprocessor -fopenmp -DDEEPGPR_USE_OPENMP -I"$LIBOMP_PREFIX/include" -L"$LIBOMP_PREFIX/lib" -fPIC -shared -o src/DeepGPR/lib/deepgpr_cpu.dylib src/DeepGPR/lib/deepgpr_cpu.c -lomp
+cp "$LIBOMP_PREFIX/lib/libomp.dylib" src/DeepGPR/lib/libomp.dylib
+install_name_tool -id "$LIBOMP_RUNTIME_NAME" src/DeepGPR/lib/libomp.dylib
+LIBOMP_DEP="$(otool -L src/DeepGPR/lib/deepgpr_cpu.dylib | awk '/libomp\.dylib/ {print $1; exit}')"
+install_name_tool -change "$LIBOMP_DEP" "$LIBOMP_RUNTIME_NAME" src/DeepGPR/lib/deepgpr_cpu.dylib
+codesign --force --sign - src/DeepGPR/lib/libomp.dylib
+codesign --force --sign - src/DeepGPR/lib/deepgpr_cpu.dylib
 ```
 
 On Windows, build `src\DeepGPR\lib\deepgpr_cpu.dll` with MSVC:

{deepgpr-0.0.13 → deepgpr-0.0.14}/README.md

@@ -127,7 +127,8 @@ def compute(device, dx=None, dt=None,
             model_gradient_sampling_interval=1,
             use_async_offload=False,
             fdtd_order=2,
-            mode=2):
+            mode=2,
+            debug=False):
 ```
 ## 📥 Input Parameters
 ### 1. Basic Physics & Grid Parameters
@@ -139,6 +140,7 @@ def compute(device, dx=None, dt=None,
 | **`dt`** | `float` | Time step size. **Note**: Must strictly satisfy the CFL (Courant-Friedrichs-Lewy) stability condition, or an exception will be raised. Typically in seconds (s). |
 | **`fdtd_order`** | `int` | Spatial finite-difference order used by the FDTD field updates. Supported values are `2`, `4`, and `8`; default is `2` for compatibility with earlier versions. |
 | **`mode`** | `int` | FWI gradient mode. `2` keeps the previous Ez-only model-gradient calculation. `3` uses Ex, Ey, and Ez electric-field contributions for relative permittivity and conductivity gradients. |
+| **`debug`** | `bool` | Runs expensive NaN/Inf and zero-field validation checks when `True`. The default `False` keeps these checks disabled for faster production runs. |
 ### 2. Medium Model Parameters
 
 This section defines the electromagnetic properties of the simulation space. For 2D simulations, set `nz=1`.
@@ -195,7 +197,14 @@ cc -std=c99 -O3 -fopenmp -fPIC -shared -o src/DeepGPR/lib/deepgpr_cpu.so src/Dee
 # macOS
 brew install libomp
 LIBOMP_PREFIX="$(brew --prefix libomp)"
-cc -std=c99 -O3 -Xpreprocessor -fopenmp -DDEEPGPR_USE_OPENMP -I"$LIBOMP_PREFIX/include" -Wl,-rpath,"$LIBOMP_PREFIX/lib" -L"$LIBOMP_PREFIX/lib" -fPIC -shared -o src/DeepGPR/lib/deepgpr_cpu.dylib src/DeepGPR/lib/deepgpr_cpu.c -lomp
+LIBOMP_RUNTIME_NAME="/opt/llvm-openmp/lib/libomp.dylib"
+cc -std=c99 -O3 -Xpreprocessor -fopenmp -DDEEPGPR_USE_OPENMP -I"$LIBOMP_PREFIX/include" -L"$LIBOMP_PREFIX/lib" -fPIC -shared -o src/DeepGPR/lib/deepgpr_cpu.dylib src/DeepGPR/lib/deepgpr_cpu.c -lomp
+cp "$LIBOMP_PREFIX/lib/libomp.dylib" src/DeepGPR/lib/libomp.dylib
+install_name_tool -id "$LIBOMP_RUNTIME_NAME" src/DeepGPR/lib/libomp.dylib
+LIBOMP_DEP="$(otool -L src/DeepGPR/lib/deepgpr_cpu.dylib | awk '/libomp\.dylib/ {print $1; exit}')"
+install_name_tool -change "$LIBOMP_DEP" "$LIBOMP_RUNTIME_NAME" src/DeepGPR/lib/deepgpr_cpu.dylib
+codesign --force --sign - src/DeepGPR/lib/libomp.dylib
+codesign --force --sign - src/DeepGPR/lib/deepgpr_cpu.dylib
 ```
 
 On Windows, build `src\DeepGPR\lib\deepgpr_cpu.dll` with MSVC:

deepgpr-0.0.14/license

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Lei Liu
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

{deepgpr-0.0.13 → deepgpr-0.0.14}/pyproject.toml

@@ -1,21 +1,24 @@
 [build-system]
-requires = ["setuptools>=70", "wheel"]
+requires = ["setuptools>=77", "wheel"]
 build-backend = "setuptools.build_meta"
 
 [project]
 name = "DeepGPR"
-version = "0.0.13"
+version = "0.0.14"
 authors = [
     { name = "Lei Liu", email = "liulei990222@gmail.com" }
 ]
 description = "PyTorch and CUDA for GPR FWI"
 readme = "README.md"
+license = "MIT"
+license-files = ["license"]
 requires-python = ">=3.8"
 
 classifiers = [
     "Programming Language :: Python :: 3",
     "Operating System :: Microsoft :: Windows",
     "Operating System :: POSIX :: Linux",
+    "Operating System :: MacOS :: MacOS X",
     "Topic :: Scientific/Engineering",
     "Topic :: Scientific/Engineering :: Physics"
 ]
@@ -27,6 +30,10 @@ dependencies = [
     "torch"
 ]
 
+[project.urls]
+Homepage = "https://github.com/songc0a/DeepGPR"
+Repository = "https://github.com/songc0a/DeepGPR"
+
 [tool.setuptools]
 package-dir = { "" = "src" }
 include-package-data = true
@@ -44,5 +51,13 @@ DeepGPR = [
     "lib/*.c",
     "lib/*.dll",
     "lib/*.so",
+    "lib/*.dylib",
     "lib/*.pyd"
-]
+]
+
+[tool.setuptools.exclude-package-data]
+DeepGPR = [
+    ".DS_Store",
+    "__pycache__/*",
+    "*.pyc"
+]

{deepgpr-0.0.13 → deepgpr-0.0.14}/src/DeepGPR/__init__.py

@@ -13,6 +13,7 @@ _PACKAGE_DIR = Path(__file__).resolve().parent
 _LIB_DIR = _PACKAGE_DIR / "lib"
 _SYSTEM_NAME = platform.system()
 _LOADED_LIBS: dict[str, ctypes.CDLL] = {}
+_OPENMP_RUNTIME: ctypes.CDLL | None = None
 
 
 def _candidate_library_paths(kind: str) -> list[Path]:
@@ -88,6 +89,46 @@ def _add_windows_dll_search_paths() -> None:
             pass
 
 
+def _preload_macos_openmp_runtime() -> None:
+    """Preload the OpenMP runtime used by PyTorch on macOS.
+
+    Args:
+        None.
+    """
+    global _OPENMP_RUNTIME
+
+    if _SYSTEM_NAME != "Darwin" or _OPENMP_RUNTIME is not None:
+        return
+
+    candidates: list[Path] = []
+    try:
+        import torch
+
+        candidates.append(Path(torch.__file__).resolve().parent / "lib" / "libomp.dylib")
+    except Exception:
+        pass
+
+    conda_prefix = os.environ.get("CONDA_PREFIX")
+    if conda_prefix:
+        candidates.append(Path(conda_prefix) / "lib" / "libomp.dylib")
+
+    candidates.append(_LIB_DIR / "libomp.dylib")
+
+    seen = set()
+    for path in candidates:
+        path = Path(path)
+        if path in seen:
+            continue
+        seen.add(path)
+        if not path.is_file():
+            continue
+        try:
+            _OPENMP_RUNTIME = ctypes.CDLL(str(path.resolve()), mode=ctypes.RTLD_GLOBAL)
+            return
+        except OSError:
+            continue
+
+
 def _require_exported_symbols(lib: ctypes.CDLL, symbols: tuple[str, ...], path: Path) -> None:
     """Validate that a loaded native library exports required symbols.
 
@@ -189,6 +230,9 @@ def _load_deepgpr_library(kind: str) -> ctypes.CDLL:
         return _LOADED_LIBS[kind]
 
     _add_windows_dll_search_paths()
+    if kind == "cpu":
+        _preload_macos_openmp_runtime()
+
     candidates = _candidate_library_paths(kind)
     load_errors: list[str] = []
 

{deepgpr-0.0.13 → deepgpr-0.0.14}/src/DeepGPR/compute2.py

@@ -57,7 +57,8 @@ def compute(device, dx=None, dt=None,
             model_gradient_sampling_interval=1,
             use_async_offload=False,
             fdtd_order=2,
-            mode=2):
+            mode=2,
+            debug=False):
     """Run DeepGPR FDTD forward modeling with autograd support.
 
     Args:
@@ -80,6 +81,7 @@ def compute(device, dx=None, dt=None,
         use_async_offload: Whether CUDA should offload saved wavefields to pinned CPU memory.
         fdtd_order: Spatial finite-difference order, supported values are 2, 4, and 8.
         mode: FWI gradient mode; 2 uses Ez only, 3 uses Ex, Ey, and Ez.
+        debug: Whether to run expensive tensor validation checks.
     """
     device = torch.device(device)
     if fdtd_order not in (2, 4, 8):
@@ -97,7 +99,7 @@ def compute(device, dx=None, dt=None,
     x0EPhi1,x0EPhi2,x0HPhi1,x0HPhi2,xmEPhi1,xmEPhi2,xmHPhi1,xmHPhi2,y0EPhi1,y0EPhi2,y0HPhi1,y0HPhi2,ymEPhi1,ymEPhi2,ymHPhi1,ymHPhi2,z0EPhi1,z0EPhi2,z0HPhi1,z0HPhi2,zmEPhi1,zmEPhi2,zmHPhi1,zmHPhi2=build_pml_phi(x0,xm,y0,ym,z0,zm,nstep,PML,device)
 
     Ex,Ey,Ez,Hx,Hy,Hz,x0EPhi1,x0EPhi2,x0HPhi1,x0HPhi2,xmEPhi1,xmEPhi2,xmHPhi1,xmHPhi2,y0EPhi1,y0EPhi2,y0HPhi1,y0HPhi2,ymEPhi1,ymEPhi2,ymHPhi1,ymHPhi2,z0EPhi1,z0EPhi2,z0HPhi1,z0HPhi2,zmEPhi1,zmEPhi2,zmHPhi1,zmHPhi2,Eall,receiver_amplitudes = DeepGPR.apply(
-        er, se,Ex,Ey,Ez,Hx,Hy,Hz,x0EPhi1,x0EPhi2,x0HPhi1,x0HPhi2,xmEPhi1,xmEPhi2,xmHPhi1,xmHPhi2,y0EPhi1,y0EPhi2,y0HPhi1,y0HPhi2,ymEPhi1,ymEPhi2,ymHPhi1,ymHPhi2,z0EPhi1,z0EPhi2,z0HPhi1,z0HPhi2,zmEPhi1,zmEPhi2,zmHPhi1,zmHPhi2, mr,dx,nx,ny,nz,dt,nt,nstep,source_amplitudes,source_location,receiver_location,pmlthick,nsr,nrx,device,dtype,x0,xm,y0,ym,z0,zm,x01,x02,xm1,xm2,y01,y02,ym1,ym2,z01,z02,zm1,zm2,ere,see,source_direction, reciever_direction, model_gradient_sampling_interval, use_async_offload, fdtd_order, mode)
+        er, se,Ex,Ey,Ez,Hx,Hy,Hz,x0EPhi1,x0EPhi2,x0HPhi1,x0HPhi2,xmEPhi1,xmEPhi2,xmHPhi1,xmHPhi2,y0EPhi1,y0EPhi2,y0HPhi1,y0HPhi2,ymEPhi1,ymEPhi2,ymHPhi1,ymHPhi2,z0EPhi1,z0EPhi2,z0HPhi1,z0HPhi2,zmEPhi1,zmEPhi2,zmHPhi1,zmHPhi2, mr,dx,nx,ny,nz,dt,nt,nstep,source_amplitudes,source_location,receiver_location,pmlthick,nsr,nrx,device,dtype,x0,xm,y0,ym,z0,zm,x01,x02,xm1,xm2,y01,y02,ym1,ym2,z01,z02,zm1,zm2,ere,see,source_direction, reciever_direction, model_gradient_sampling_interval, use_async_offload, fdtd_order, mode, debug)
 
     return Eall,(Ex,Ey,Ez),(Hx,Hy,Hz),(x0EPhi1,x0EPhi2,x0HPhi1,x0HPhi2,xmEPhi1,xmEPhi2,xmHPhi1,xmHPhi2,y0EPhi1,y0EPhi2,y0HPhi1,y0HPhi2,ymEPhi1,ymEPhi2,ymHPhi1,ymHPhi2,z0EPhi1,z0EPhi2,z0HPhi1,z0HPhi2,zmEPhi1,zmEPhi2,zmHPhi1,zmHPhi2),receiver_amplitudes
 
@@ -112,7 +114,7 @@ class DeepGPR(torch.autograd.Function):
                 y0,ym,z0,zm,x01,x02,xm1,xm2,
                 y01,y02,ym1,ym2,z01,z02,zm1,zm2,
                 ere,see,source_direction, reciever_direction, 
-                model_gradient_sampling_interval, use_async_offload, fdtd_order, mode):
+                model_gradient_sampling_interval, use_async_offload, fdtd_order, mode, debug):
         """Call the native forward solver and save tensors for backward.
 
         Args:
@@ -155,6 +157,7 @@ class DeepGPR(torch.autograd.Function):
             use_async_offload: Whether CUDA should offload saved wavefields to pinned CPU memory.
             fdtd_order: Spatial finite-difference order.
             mode: FWI gradient mode; 2 uses Ez only, 3 uses Ex, Ey, and Ez.
+            debug: Whether to run expensive tensor validation checks.
         """
         
         source_amplitudes = source_amplitudes.contiguous()
@@ -179,6 +182,7 @@ class DeepGPR(torch.autograd.Function):
         ctx.use_async_offload = bool(use_async_offload and device.type == "cuda")
         ctx.fdtd_order = fdtd_order
         ctx.mode = mode
+        ctx.debug = bool(debug)
 
         nt_saved = (nt + model_gradient_sampling_interval - 1) // model_gradient_sampling_interval
         e_components = 3 if mode == 3 else 1
@@ -252,23 +256,24 @@ class DeepGPR(torch.autograd.Function):
                 mode)
         _sync_cuda_device(device)
 
-        check_tensors_for_nan_inf(d="forward",
-            Ex=Ex, Ey=Ey, Ez=Ez,
-            Hx=Hx, Hy=Hy, Hz=Hz,
-            x0EPhi1=x0EPhi1, x0EPhi2=x0EPhi2,
-            x0HPhi1=x0HPhi1, x0HPhi2=x0HPhi2,
-            xmEPhi1=xmEPhi1, xmEPhi2=xmEPhi2,
-            xmHPhi1=xmHPhi1, xmHPhi2=xmHPhi2,
-            y0EPhi1=y0EPhi1, y0EPhi2=y0EPhi2,
-            y0HPhi1=y0HPhi1, y0HPhi2=y0HPhi2,
-            ymEPhi1=ymEPhi1, ymEPhi2=ymEPhi2,
-            ymHPhi1=ymHPhi1, ymHPhi2=ymHPhi2,
-            z0EPhi1=z0EPhi1, z0EPhi2=z0EPhi2,
-            z0HPhi1=z0HPhi1, z0HPhi2=z0HPhi2,
-            zmEPhi1=zmEPhi1, zmEPhi2=zmEPhi2,
-            zmHPhi1=zmHPhi1, zmHPhi2=zmHPhi2
-        )
-        _check_nonzero_source_created_fields(c_lib, source_amplitudes, Ex, Ey, Ez, Hx, Hy, Hz)
+        if ctx.debug:
+            check_tensors_for_nan_inf(d="forward",
+                Ex=Ex, Ey=Ey, Ez=Ez,
+                Hx=Hx, Hy=Hy, Hz=Hz,
+                x0EPhi1=x0EPhi1, x0EPhi2=x0EPhi2,
+                x0HPhi1=x0HPhi1, x0HPhi2=x0HPhi2,
+                xmEPhi1=xmEPhi1, xmEPhi2=xmEPhi2,
+                xmHPhi1=xmHPhi1, xmHPhi2=xmHPhi2,
+                y0EPhi1=y0EPhi1, y0EPhi2=y0EPhi2,
+                y0HPhi1=y0HPhi1, y0HPhi2=y0HPhi2,
+                ymEPhi1=ymEPhi1, ymEPhi2=ymEPhi2,
+                ymHPhi1=ymHPhi1, ymHPhi2=ymHPhi2,
+                z0EPhi1=z0EPhi1, z0EPhi2=z0EPhi2,
+                z0HPhi1=z0HPhi1, z0HPhi2=z0HPhi2,
+                zmEPhi1=zmEPhi1, zmEPhi2=zmEPhi2,
+                zmHPhi1=zmHPhi1, zmHPhi2=zmHPhi2
+            )
+            _check_nonzero_source_created_fields(c_lib, source_amplitudes, Ex, Ey, Ez, Hx, Hy, Hz)
 
         ctx.Eall = Eall
         return (Ex,Ey,Ez,Hx,Hy,Hz,x0EPhi1,x0EPhi2,x0HPhi1,x0HPhi2,xmEPhi1,xmEPhi2,xmHPhi1,xmHPhi2,y0EPhi1,y0EPhi2,y0HPhi1,y0HPhi2,ymEPhi1,ymEPhi2,ymHPhi1,ymHPhi2,z0EPhi1,z0EPhi2,z0HPhi1,z0HPhi2,zmEPhi1,zmEPhi2,zmHPhi1,zmHPhi2,Eall,receiver_amplitudes[:,reciever_direction,:,:])
@@ -462,7 +467,8 @@ class DeepGPR(torch.autograd.Function):
         if serequiregrad == 1:
             tensors_to_check["grad_se"] = grad_se
 
-        check_tensors_for_nan_inf(d="backward", **tensors_to_check)
+        if ctx.debug:
+            check_tensors_for_nan_inf(d="backward", **tensors_to_check)
 
         ctx.Eall = None
         del Eall,er, se, mr,receiver_location,x0,xm,y0,ym,z0,zm,x01,x02,xm1,xm2,y01,y02,ym1,ym2,z01,z02,zm1,zm2,ere,see, Eupdatecoffs0, Eupdatecoffs1, Eupdatecoffs4, Hupdatecoffs0, Hupdatecoffs1, Hupdatecoffs4
@@ -481,5 +487,5 @@ class DeepGPR(torch.autograd.Function):
                     None, None, None, None, None, None, None, None,
                     None, None, None, None, None, None, None, None, 
                     None, None, None, None, None, None, None, None, 
-                    None, None, None
+                    None, None, None, None
                 )