torchref 0.3.0__tar.gz

torchref-0.3.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Hans Peter Seidel
+
+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.

torchref-0.3.0/PKG-INFO

@@ -0,0 +1,116 @@
+Metadata-Version: 2.4
+Name: torchref
+Version: 0.3.0
+Summary: Tools for multicopy refinement of crystallographic models
+Author: HansPeterSeidel
+License-Expression: MIT
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: numpy<2.4.0,>=1.24.0
+Requires-Dist: pandas<2.4.0,>=2.0.0
+Requires-Dist: torch<2.10.0,>=2.0.0
+Requires-Dist: tqdm<4.68.0,>=4.61.0
+Requires-Dist: numba<0.64.0,>=0.59.0
+Requires-Dist: gemmi<0.8.0,>=0.5.0
+Requires-Dist: scipy<1.18.0,>=1.10.0
+Requires-Dist: matplotlib<3.11.0,>=3.7.0
+Requires-Dist: reciprocalspaceship<1.1.0,>=0.9.18
+Requires-Dist: pyarrow<23.0.0,>=12.0.0
+Provides-Extra: dev
+Requires-Dist: pytest>=6.0.0; extra == "dev"
+Requires-Dist: pytest-cov>=2.12.0; extra == "dev"
+Requires-Dist: black>=21.5b2; extra == "dev"
+Requires-Dist: isort>=5.9.0; extra == "dev"
+Requires-Dist: flake8>=3.9.0; extra == "dev"
+Provides-Extra: docs
+Requires-Dist: sphinx>=4.0.0; extra == "docs"
+Requires-Dist: sphinx-rtd-theme>=1.0.0; extra == "docs"
+Requires-Dist: numpydoc>=1.1.0; extra == "docs"
+Dynamic: license-file
+
+# TorchRef
+
+**A PyTorch-based crystallographic refinement library**
+
+[![Python 3.8+](https://img.shields.io/badge/python-3.8+-blue.svg)](https://www.python.org/downloads/)
+[![PyTorch](https://img.shields.io/badge/PyTorch-1.9+-ee4c2c.svg)](https://pytorch.org/)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+
+TorchRef is a crystallographic refinement package built entirely on PyTorch. By leveraging PyTorch's automatic differentiation and GPU acceleration, TorchRef enables seamless integration with machine learning workflows and provides a flexible, extensible framework for crystallographic structure refinement.
+
+## Key Features
+
+- **Native PyTorch Integration**: Built on PyTorch's `nn.Module` architecture, TorchRef integrates naturally with the PyTorch ecosystem, including machine learning models, optimizers, and GPU acceleration.
+
+- **Automatic Differentiation**: Dynamic computational graphs eliminate the need for manually implemented gradient calculations. Define new refinement targets directly—PyTorch handles the derivatives automatically.
+
+- **Modular Architecture**: Following PyTorch's module pattern, components are easily composable and extensible. Add custom targets, restraints, or optimizers without modifying core code.
+
+- **GPU Acceleration**: Leverage CUDA for structure factor calculations, scaling, and optimization—achieving significant speedups for large structures.
+
+- **FFT-based Structure Factors**: Efficient structure factor calculation using Fast Fourier Transform (FFT) methods, enabling rapid F_calc computation even for large unit cells.
+
+- **State Management**: Full `state_dict` support enables saving and loading complete refinement states, including model parameters, scaler settings, and restraints.
+
+## Installation
+
+```bash
+# Clone the repository
+git clone https://github.com/HatPdotS/TorchRef.git
+cd torchref
+
+# Install with pip
+pip install -e .
+
+# Or install with development dependencies
+pip install -e ".[dev]"
+```
+
+### Dependencies
+
+- Python ≥ 3.8
+- PyTorch ≥ 1.9
+- NumPy ≥ 1.20
+- Gemmi ≥ 0.5
+- reciprocalspaceship ≥ 0.9
+- SciPy ≥ 1.7
+
+## Getting Started
+
+For demonstrations and usage examples, see the example notebooks in [`example_notebooks/`](example_notebooks/):
+
+- [`basic_usage.ipynb`](example_notebooks/basic_usage.ipynb) - Getting started tutorial
+- [`code_examples.ipynb`](example_notebooks/code_examples.ipynb) - Code examples and patterns
+- [`target_exploration.ipynb`](example_notebooks/target_exploration.ipynb) - Exploring refinement targets
+
+## Testing
+
+```bash
+# Run all tests
+pytest tests/
+
+# Run with coverage
+pytest tests/ --cov=torchref
+
+# Run specific test categories
+pytest tests/unit/           # Fast unit tests
+pytest tests/integration/    # Integration tests
+pytest tests/functional/     # Full workflow tests
+```
+
+## Contributing
+
+Contributions are welcome! Please follow these guidelines:
+
+1. Follow the [NumPy docstring style](https://numpydoc.readthedocs.io/en/latest/format.html)
+2. Add tests for new functionality
+3. Ensure all tests pass before submitting
+
+## License
+
+This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
+
+
+
+

torchref-0.3.0/README.md

@@ -0,0 +1,85 @@
+# TorchRef
+
+**A PyTorch-based crystallographic refinement library**
+
+[![Python 3.8+](https://img.shields.io/badge/python-3.8+-blue.svg)](https://www.python.org/downloads/)
+[![PyTorch](https://img.shields.io/badge/PyTorch-1.9+-ee4c2c.svg)](https://pytorch.org/)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+
+TorchRef is a crystallographic refinement package built entirely on PyTorch. By leveraging PyTorch's automatic differentiation and GPU acceleration, TorchRef enables seamless integration with machine learning workflows and provides a flexible, extensible framework for crystallographic structure refinement.
+
+## Key Features
+
+- **Native PyTorch Integration**: Built on PyTorch's `nn.Module` architecture, TorchRef integrates naturally with the PyTorch ecosystem, including machine learning models, optimizers, and GPU acceleration.
+
+- **Automatic Differentiation**: Dynamic computational graphs eliminate the need for manually implemented gradient calculations. Define new refinement targets directly—PyTorch handles the derivatives automatically.
+
+- **Modular Architecture**: Following PyTorch's module pattern, components are easily composable and extensible. Add custom targets, restraints, or optimizers without modifying core code.
+
+- **GPU Acceleration**: Leverage CUDA for structure factor calculations, scaling, and optimization—achieving significant speedups for large structures.
+
+- **FFT-based Structure Factors**: Efficient structure factor calculation using Fast Fourier Transform (FFT) methods, enabling rapid F_calc computation even for large unit cells.
+
+- **State Management**: Full `state_dict` support enables saving and loading complete refinement states, including model parameters, scaler settings, and restraints.
+
+## Installation
+
+```bash
+# Clone the repository
+git clone https://github.com/HatPdotS/TorchRef.git
+cd torchref
+
+# Install with pip
+pip install -e .
+
+# Or install with development dependencies
+pip install -e ".[dev]"
+```
+
+### Dependencies
+
+- Python ≥ 3.8
+- PyTorch ≥ 1.9
+- NumPy ≥ 1.20
+- Gemmi ≥ 0.5
+- reciprocalspaceship ≥ 0.9
+- SciPy ≥ 1.7
+
+## Getting Started
+
+For demonstrations and usage examples, see the example notebooks in [`example_notebooks/`](example_notebooks/):
+
+- [`basic_usage.ipynb`](example_notebooks/basic_usage.ipynb) - Getting started tutorial
+- [`code_examples.ipynb`](example_notebooks/code_examples.ipynb) - Code examples and patterns
+- [`target_exploration.ipynb`](example_notebooks/target_exploration.ipynb) - Exploring refinement targets
+
+## Testing
+
+```bash
+# Run all tests
+pytest tests/
+
+# Run with coverage
+pytest tests/ --cov=torchref
+
+# Run specific test categories
+pytest tests/unit/           # Fast unit tests
+pytest tests/integration/    # Integration tests
+pytest tests/functional/     # Full workflow tests
+```
+
+## Contributing
+
+Contributions are welcome! Please follow these guidelines:
+
+1. Follow the [NumPy docstring style](https://numpydoc.readthedocs.io/en/latest/format.html)
+2. Add tests for new functionality
+3. Ensure all tests pass before submitting
+
+## License
+
+This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
+
+
+
+

torchref-0.3.0/pyproject.toml

@@ -0,0 +1,86 @@
+[build-system]
+requires = ["setuptools>=42", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "torchref"
+version = "0.3.0"
+description = "Tools for multicopy refinement of crystallographic models"
+readme = "README.md"
+requires-python = ">=3.9"
+license = "MIT"
+authors = [
+    {name = "HansPeterSeidel"},
+]
+# Version requirements validated via tox compatibility testing (2026-01-13)
+# Tested combinations:
+#   - Python 3.9:  numpy 1.24-1.26, pandas 2.0.x, torch 2.0-2.4, numba 0.59-0.60, scipy 1.10-1.13
+#   - Python 3.11: numpy 1.24-2.3,  pandas 2.0-2.3, torch 2.0-2.9, numba 0.59-0.63, scipy 1.10-1.17
+#   - Python 3.12: numpy 2.0-2.3,   pandas 2.2-2.3, torch 2.4-2.9, numba 0.61-0.63, scipy 1.13-1.17
+# Upper bounds set to next minor version above tested maximum
+dependencies = [
+    "numpy>=1.24.0,<2.4.0",
+    "pandas>=2.0.0,<2.4.0",
+    "torch>=2.0.0,<2.10.0",
+    "tqdm>=4.61.0,<4.68.0",
+    "numba>=0.59.0,<0.64.0",
+    "gemmi>=0.5.0,<0.8.0",
+    "scipy>=1.10.0,<1.18.0",
+    "matplotlib>=3.7.0,<3.11.0",
+    "reciprocalspaceship>=0.9.18,<1.1.0",
+    "pyarrow>=12.0.0,<23.0.0",
+]
+
+[project.scripts]
+torchref-refine = "torchref.cli.refine:main"
+torchref-refine-screened = "torchref.cli.refine_screened:main"
+torchref-refine-everything = "torchref.cli.refine_everything:main"
+torchref-refine-random-weights = "torchref.cli.refine_everything_random_weights:main"
+torchref-refine-policy = 'torchref.cli.refine_everything_policy:main'
+torchref-refine-static = 'torchref.cli.refine_everything_static:main'
+torchref-refine-hyperparameters = 'torchref.cli.refine_everything_hyperparameters:main'
+torchref-download-data = 'torchref.cli.download_data:main'
+
+[project.optional-dependencies]
+
+dev = [
+    "pytest>=6.0.0",
+    "pytest-cov>=2.12.0",
+    "black>=21.5b2",
+    "isort>=5.9.0",
+    "flake8>=3.9.0",
+]
+
+# Optional CCTBX/iotbx functionality (requires conda, not pip)
+# Install with: conda install -c conda-forge cctbx-base
+# The iotbx module is used in torchref/math_functions/CCTBX_related.py
+
+docs = [
+    "sphinx>=4.0.0",
+    "sphinx-rtd-theme>=1.0.0",
+    "numpydoc>=1.1.0",
+]
+
+[tool.setuptools]
+packages = ["torchref"]
+
+[tool.setuptools.package-data]
+torchref = ["*.py"]
+
+[tool.setuptools.exclude-package-data]
+torchref = ["*.py"]
+
+[tool.black]
+line-length = 88
+target-version = ["py39"]
+
+[tool.isort]
+profile = "black"
+line_length = 88
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+python_files = "test_*.py"
+
+[tool.ruff.lint]
+ignore = ["F841","E741","F841","E402","E722"]

torchref-0.3.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

torchref-0.3.0/torchref/__init__.py

@@ -0,0 +1,134 @@
+"""
+TorchRef - A PyTorch-based crystallographic refinement library.
+
+TorchRef provides GPU-accelerated crystallographic structure refinement
+using PyTorch's automatic differentiation and nn.Module architecture.
+
+Key Features
+------------
+- Native PyTorch integration with nn.Module architecture
+- Automatic differentiation for custom target functions
+- GPU acceleration for structure factor calculations
+- Modular design for easy extension
+
+Quick Start
+-----------
+::
+
+    from torchref import Refinement, ReflectionData, Model
+
+    # Load data and model
+    data = ReflectionData().load_mtz('data.mtz')
+    model = Model().load_pdb('structure.pdb')
+
+    # Run refinement
+    refinement = Refinement(data=data, model=model, device='cuda')
+    refinement.run_refinement(macro_cycles=10)
+
+Modules
+-------
+io
+    File I/O for MTZ, PDB, CIF formats.
+model
+    Atomic structure models (coordinates, B-factors, occupancies).
+refinement
+    Core refinement framework with targets and weighting schemes.
+restraints
+    Geometry restraints (bonds, angles, torsions, planes).
+scaling
+    Structure factor scaling and bulk solvent models.
+symmetry
+    Crystallographic symmetry operations.
+alignment
+    Patterson-based structure alignment.
+math_functions
+    Mathematical utilities for crystallography.
+utils
+    General utilities and debugging tools.
+"""
+
+__version__ = "0.3.0"
+
+
+import os
+import warnings
+from pathlib import Path
+
+from torchref._bootstrap import configure_threading, detect_available_cpus
+
+# Configure threading before importing torch
+if "TORCHREF_NUM_THREADS" in os.environ:
+    N_CPUS = int(os.environ["TORCHREF_NUM_THREADS"])
+    warnings.warn(
+        f"TorchRef using user-specified {N_CPUS} threads from TORCHREF_NUM_THREADS.",
+        stacklevel=2,
+    )
+else:
+    N_CPUS = detect_available_cpus()
+    os.environ["TORCHREF_NUM_THREADS"] = str(N_CPUS)
+    warnings.warn(
+        f"TorchRef auto-configured {N_CPUS} threads. Set TORCHREF_NUM_THREADS to override.",
+        stacklevel=2,
+    )
+
+configure_threading(N_CPUS)
+
+import torch
+
+torch.set_num_threads(N_CPUS)
+
+# Dtype configuration (must be imported after torch)
+from torchref.config import dtypes
+
+
+# Project root path for referencing package files
+ROOT_TORCHREF = Path(__file__).parent.parent.resolve()
+
+# Package path for referencing internal files
+PATH_TORCHREF = Path(__file__).parent.resolve()
+
+PATH_TORCHREF_DATA = PATH_TORCHREF / "data"
+
+# =============================================================================
+# Convenience imports for common classes
+# =============================================================================
+
+
+# Data I/O
+from torchref.io import DatasetCollection, ReflectionData
+
+# Model
+from torchref.model import Model, ModelFT
+
+# Refinement
+from torchref.refinement import LBFGSRefinement, Refinement
+
+# Restraints
+from torchref.restraints import Restraints
+
+# Scaling
+from torchref.scaling import Scaler, SolventModel
+
+__all__ = [
+    # Version and paths
+    "__version__",
+    "ROOT_TORCHREF",
+    "PATH_TORCHREF",
+    "N_CPUS",
+    # Dtype configuration
+    "dtypes",
+    # Data I/O
+    "ReflectionData",
+    "DatasetCollection",
+    # Model
+    "Model",
+    "ModelFT",
+    # Refinement
+    "Refinement",
+    "LBFGSRefinement",
+    # Restraints
+    "Restraints",
+    # Scaling
+    "Scaler",
+    "SolventModel",
+]

torchref-0.3.0/torchref/_bootstrap.py

@@ -0,0 +1,49 @@
+"""
+
+Bootstrap module to configure threading before importing heavy libraries.
+
+Is imported automatically when torchref is imported.
+If you are not on a slurm node and want to customize threading,
+call `configure_threading()` before importing torchref.
+
+"""
+
+import os
+
+
+def detect_available_cpus(max_if_not_slurm=4) -> int:
+    """Detect actual available CPUs respecting cgroups, affinity, and SLURM."""
+
+    # 1. Check SLURM first (most reliable on HPC)
+    slurm_cpus = os.environ.get("SLURM_CPUS_PER_TASK")
+    if slurm_cpus:
+        return int(slurm_cpus)
+
+    # 4. Check CPU affinity
+    try:
+        return min(len(os.sched_getaffinity(0)), 4)
+    except (AttributeError, OSError):
+        pass
+
+    # 5. Fallback to os.cpu_count() but cap it sensibly
+    return min(os.cpu_count() or 1, 4)
+
+
+def configure_threading(num_threads: int = None, pin_threads=False) -> int:
+    """Configure all threading libraries. Call BEFORE importing torch/numpy."""
+
+    if num_threads is None:
+        num_threads = detect_available_cpus()
+
+    n = str(num_threads)
+
+    os.environ["OMP_NUM_THREADS"] = n
+    os.environ["MKL_NUM_THREADS"] = n
+    os.environ["OPENBLAS_NUM_THREADS"] = n
+
+    if pin_threads:
+        # Optional: enable thread pinning
+        os.environ["OMP_PROC_BIND"] = "TRUE"
+        os.environ["OMP_PLACES"] = "cores"
+
+    return num_threads

torchref-0.3.0/torchref/config.py

@@ -0,0 +1,147 @@
+"""
+Centralized configuration for TorchRef.
+
+Default dtypes can be set via environment variables at import time:
+- TORCHREF_DTYPE_FLOAT: float32 (default) or float64
+- TORCHREF_DTYPE_INT: int32 (default) or int64
+- TORCHREF_DTYPE_COMPLEX: complex64 (default) or complex128
+
+Users can also change dtypes at runtime via attribute assignment:
+    import torchref
+    torchref.dtypes.float = torch.float64
+    torchref.dtypes.int = torch.int64
+    torchref.dtypes.complex = torch.complex128
+
+Or read current values:
+    torchref.dtypes.float   # torch.float32
+    torchref.dtypes.int     # torch.int32
+    torchref.dtypes.complex # torch.complex64
+"""
+
+import os
+
+import torch
+
+# Map strings to torch dtypes
+_FLOAT_DTYPE_MAP = {
+    "float32": torch.float32,
+    "float64": torch.float64,
+}
+
+_INT_DTYPE_MAP = {
+    "int32": torch.int32,
+    "int64": torch.int64,
+}
+
+_COMPLEX_DTYPE_MAP = {
+    "complex64": torch.complex64,
+    "complex128": torch.complex128,
+}
+
+
+class DtypeConfig:
+    """
+    Dtype configuration with property-based access.
+
+    Access dtypes as attributes:
+        dtypes.float    # get current float dtype
+        dtypes.int      # get current int dtype
+        dtypes.complex  # get current complex dtype
+
+    Set dtypes via assignment:
+        dtypes.float = torch.float64
+        dtypes.int = torch.int64
+        dtypes.complex = torch.complex128
+    """
+
+    def __init__(self):
+        # Parse environment variables with defaults
+        float_str = os.environ.get("TORCHREF_DTYPE_FLOAT", "float32").lower()
+        int_str = os.environ.get("TORCHREF_DTYPE_INT", "int32").lower()
+        complex_str = os.environ.get("TORCHREF_DTYPE_COMPLEX", "complex64").lower()
+
+        # Validate and set
+        if float_str not in _FLOAT_DTYPE_MAP:
+            raise ValueError(
+                f"Invalid TORCHREF_DTYPE_FLOAT: {float_str}. "
+                f"Valid values: {list(_FLOAT_DTYPE_MAP.keys())}"
+            )
+        if int_str not in _INT_DTYPE_MAP:
+            raise ValueError(
+                f"Invalid TORCHREF_DTYPE_INT: {int_str}. "
+                f"Valid values: {list(_INT_DTYPE_MAP.keys())}"
+            )
+        if complex_str not in _COMPLEX_DTYPE_MAP:
+            raise ValueError(
+                f"Invalid TORCHREF_DTYPE_COMPLEX: {complex_str}. "
+                f"Valid values: {list(_COMPLEX_DTYPE_MAP.keys())}"
+            )
+
+        self._float = _FLOAT_DTYPE_MAP[float_str]
+        self._int = _INT_DTYPE_MAP[int_str]
+        self._complex = _COMPLEX_DTYPE_MAP[complex_str]
+
+    @property
+    def float(self) -> torch.dtype:
+        """Get the current default float dtype."""
+        return self._float
+
+    @float.setter
+    def float(self, dtype: torch.dtype) -> None:
+        """Set the default float dtype for all future operations."""
+        if dtype not in (torch.float32, torch.float64):
+            raise ValueError(
+                f"Invalid float dtype: {dtype}. Use torch.float32 or torch.float64."
+            )
+        self._float = dtype
+
+    @property
+    def int(self) -> torch.dtype:
+        """Get the current default int dtype."""
+        return self._int
+
+    @int.setter
+    def int(self, dtype: torch.dtype) -> None:
+        """Set the default int dtype for all future operations."""
+        if dtype not in (torch.int32, torch.int64):
+            raise ValueError(
+                f"Invalid int dtype: {dtype}. Use torch.int32 or torch.int64."
+            )
+        self._int = dtype
+
+    @property
+    def complex(self) -> torch.dtype:
+        """Get the current default complex dtype."""
+        return self._complex
+
+    @complex.setter
+    def complex(self, dtype: torch.dtype) -> None:
+        """Set the default complex dtype for all future operations."""
+        if dtype not in (torch.complex64, torch.complex128):
+            raise ValueError(
+                f"Invalid complex dtype: {dtype}. Use torch.complex64 or torch.complex128."
+            )
+        self._complex = dtype
+
+    def __repr__(self) -> str:
+        return f"DtypeConfig(float={self._float}, int={self._int}, complex={self._complex})"
+
+
+# Global singleton instance
+dtypes = DtypeConfig()
+
+
+# Convenience functions for internal use (avoid repeated attribute lookups)
+def get_float_dtype() -> torch.dtype:
+    """Get the current default float dtype."""
+    return dtypes.float
+
+
+def get_int_dtype() -> torch.dtype:
+    """Get the current default int dtype."""
+    return dtypes.int
+
+
+def get_complex_dtype() -> torch.dtype:
+    """Get the current default complex dtype."""
+    return dtypes.complex

torchref-0.3.0/torchref.egg-info/PKG-INFO

@@ -0,0 +1,116 @@
+Metadata-Version: 2.4
+Name: torchref
+Version: 0.3.0
+Summary: Tools for multicopy refinement of crystallographic models
+Author: HansPeterSeidel
+License-Expression: MIT
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: numpy<2.4.0,>=1.24.0
+Requires-Dist: pandas<2.4.0,>=2.0.0
+Requires-Dist: torch<2.10.0,>=2.0.0
+Requires-Dist: tqdm<4.68.0,>=4.61.0
+Requires-Dist: numba<0.64.0,>=0.59.0
+Requires-Dist: gemmi<0.8.0,>=0.5.0
+Requires-Dist: scipy<1.18.0,>=1.10.0
+Requires-Dist: matplotlib<3.11.0,>=3.7.0
+Requires-Dist: reciprocalspaceship<1.1.0,>=0.9.18
+Requires-Dist: pyarrow<23.0.0,>=12.0.0
+Provides-Extra: dev
+Requires-Dist: pytest>=6.0.0; extra == "dev"
+Requires-Dist: pytest-cov>=2.12.0; extra == "dev"
+Requires-Dist: black>=21.5b2; extra == "dev"
+Requires-Dist: isort>=5.9.0; extra == "dev"
+Requires-Dist: flake8>=3.9.0; extra == "dev"
+Provides-Extra: docs
+Requires-Dist: sphinx>=4.0.0; extra == "docs"
+Requires-Dist: sphinx-rtd-theme>=1.0.0; extra == "docs"
+Requires-Dist: numpydoc>=1.1.0; extra == "docs"
+Dynamic: license-file
+
+# TorchRef
+
+**A PyTorch-based crystallographic refinement library**
+
+[![Python 3.8+](https://img.shields.io/badge/python-3.8+-blue.svg)](https://www.python.org/downloads/)
+[![PyTorch](https://img.shields.io/badge/PyTorch-1.9+-ee4c2c.svg)](https://pytorch.org/)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+
+TorchRef is a crystallographic refinement package built entirely on PyTorch. By leveraging PyTorch's automatic differentiation and GPU acceleration, TorchRef enables seamless integration with machine learning workflows and provides a flexible, extensible framework for crystallographic structure refinement.
+
+## Key Features
+
+- **Native PyTorch Integration**: Built on PyTorch's `nn.Module` architecture, TorchRef integrates naturally with the PyTorch ecosystem, including machine learning models, optimizers, and GPU acceleration.
+
+- **Automatic Differentiation**: Dynamic computational graphs eliminate the need for manually implemented gradient calculations. Define new refinement targets directly—PyTorch handles the derivatives automatically.
+
+- **Modular Architecture**: Following PyTorch's module pattern, components are easily composable and extensible. Add custom targets, restraints, or optimizers without modifying core code.
+
+- **GPU Acceleration**: Leverage CUDA for structure factor calculations, scaling, and optimization—achieving significant speedups for large structures.
+
+- **FFT-based Structure Factors**: Efficient structure factor calculation using Fast Fourier Transform (FFT) methods, enabling rapid F_calc computation even for large unit cells.
+
+- **State Management**: Full `state_dict` support enables saving and loading complete refinement states, including model parameters, scaler settings, and restraints.
+
+## Installation
+
+```bash
+# Clone the repository
+git clone https://github.com/HatPdotS/TorchRef.git
+cd torchref
+
+# Install with pip
+pip install -e .
+
+# Or install with development dependencies
+pip install -e ".[dev]"
+```
+
+### Dependencies
+
+- Python ≥ 3.8
+- PyTorch ≥ 1.9
+- NumPy ≥ 1.20
+- Gemmi ≥ 0.5
+- reciprocalspaceship ≥ 0.9
+- SciPy ≥ 1.7
+
+## Getting Started
+
+For demonstrations and usage examples, see the example notebooks in [`example_notebooks/`](example_notebooks/):
+
+- [`basic_usage.ipynb`](example_notebooks/basic_usage.ipynb) - Getting started tutorial
+- [`code_examples.ipynb`](example_notebooks/code_examples.ipynb) - Code examples and patterns
+- [`target_exploration.ipynb`](example_notebooks/target_exploration.ipynb) - Exploring refinement targets
+
+## Testing
+
+```bash
+# Run all tests
+pytest tests/
+
+# Run with coverage
+pytest tests/ --cov=torchref
+
+# Run specific test categories
+pytest tests/unit/           # Fast unit tests
+pytest tests/integration/    # Integration tests
+pytest tests/functional/     # Full workflow tests
+```
+
+## Contributing
+
+Contributions are welcome! Please follow these guidelines:
+
+1. Follow the [NumPy docstring style](https://numpydoc.readthedocs.io/en/latest/format.html)
+2. Add tests for new functionality
+3. Ensure all tests pass before submitting
+
+## License
+
+This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
+
+
+
+

torchref-0.3.0/torchref.egg-info/SOURCES.txt

@@ -0,0 +1,12 @@
+LICENSE
+README.md
+pyproject.toml
+torchref/__init__.py
+torchref/_bootstrap.py
+torchref/config.py
+torchref.egg-info/PKG-INFO
+torchref.egg-info/SOURCES.txt
+torchref.egg-info/dependency_links.txt
+torchref.egg-info/entry_points.txt
+torchref.egg-info/requires.txt
+torchref.egg-info/top_level.txt

torchref-0.3.0/torchref.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

torchref-0.3.0/torchref.egg-info/entry_points.txt

@@ -0,0 +1,9 @@
+[console_scripts]
+torchref-download-data = torchref.cli.download_data:main
+torchref-refine = torchref.cli.refine:main
+torchref-refine-everything = torchref.cli.refine_everything:main
+torchref-refine-hyperparameters = torchref.cli.refine_everything_hyperparameters:main
+torchref-refine-policy = torchref.cli.refine_everything_policy:main
+torchref-refine-random-weights = torchref.cli.refine_everything_random_weights:main
+torchref-refine-screened = torchref.cli.refine_screened:main
+torchref-refine-static = torchref.cli.refine_everything_static:main

torchref-0.3.0/torchref.egg-info/requires.txt

@@ -0,0 +1,22 @@
+numpy<2.4.0,>=1.24.0
+pandas<2.4.0,>=2.0.0
+torch<2.10.0,>=2.0.0
+tqdm<4.68.0,>=4.61.0
+numba<0.64.0,>=0.59.0
+gemmi<0.8.0,>=0.5.0
+scipy<1.18.0,>=1.10.0
+matplotlib<3.11.0,>=3.7.0
+reciprocalspaceship<1.1.0,>=0.9.18
+pyarrow<23.0.0,>=12.0.0
+
+[dev]
+pytest>=6.0.0
+pytest-cov>=2.12.0
+black>=21.5b2
+isort>=5.9.0
+flake8>=3.9.0
+
+[docs]
+sphinx>=4.0.0
+sphinx-rtd-theme>=1.0.0
+numpydoc>=1.1.0

torchref-0.3.0/torchref.egg-info/top_level.txt

@@ -0,0 +1 @@
+torchref