Installation#
Requirements#
Component |
Minimum version |
|---|---|
Python |
3.10 |
NumPy |
1.24 |
Fortran compiler |
gfortran, Intel ifx, or Intel ifort |
Step 1 — Clone the repository#
git clone https://github.com/your-username/pykriging.git
cd pykriging
Step 2 — Compile the Fortran library#
The shared library (libkriging.so on Linux/macOS, kriging.dll on Windows)
must be compiled before use. Run build_lib.py from the project root:
Linux / macOS (gfortran)
python build_lib.py
# explicit compiler:
python build_lib.py --compiler gfortran
# debug build:
python build_lib.py --opt debug
Windows (Intel ifx)
call "C:\Program Files (x86)\Intel\oneAPI\setvars.bat"
python build_lib.py --compiler ifx
Windows (gfortran via MSYS2 / rtools)
python build_lib.py --compiler gfortran
The script compiles all Fortran sources in src/libkriging/ in dependency
order and places the compiled library inside src/pykriging/.
Note
Pass --no-openmp to disable OpenMP if your compiler or environment does not
support it. Single-threaded performance is unaffected for most workloads.
Step 3 — Install the Python package#
pip install -e . # editable install (recommended for development)
# or:
pip install .
Step 4 — Verify#
pip install -e ".[dev]"
pytest
All tests should pass. If the shared library is missing, you will get an
OSError when importing pykriging.
Docs dependencies (optional)#
To build this documentation locally:
pip install -e ".[docs]"
cd docs
sphinx-build . _build/html
# open _build/html/index.html