Metadata-Version: 2.1
Name: phs_poly
Version: 0.2.3
Summary: PHS+poly approximation procedures
Keywords: approximation,RBF-FD,meshfree
Author-Email: Ivan Pribec <ivan.pribec@gmail.com>
License: MIT
Classifier: Development Status :: 1 - Planning
Classifier: Intended Audience :: Science/Research
Classifier: Operating System :: MacOS
Classifier: Topic :: Scientific/Engineering
Classifier: Topic :: Scientific/Engineering :: Mathematics
Classifier: Programming Language :: Fortran
Classifier: Programming Language :: Python :: 3.9
Requires-Python: >=3.9
Requires-Dist: numpy
Description-Content-Type: text/markdown

# phs_poly

Procedures for generating **RBF-FD weights** for derivative calculations.  
Currently only **PHS + poly** approximations in 2-d are supported.

---

## Installation

To install use,

```
pip install phs_poly
```

Requirements_
- A Fortran compiler (e.g. `gfortran` or `flang`)
- A LAPACK library (e.g. OpenBLAS, Accelerate, ArmPL)
- NumPy

## Quickstart

```python
import phs_poly

# ... set x- and y-coordinates, initialize work and coeff arrays ...

ierr = phs_poly.phs3_poly2(n,x,y,coeffs,ldc,wrk,iwrk)

# First derivatives weights
wx = coeffs[0:n,0] 
wy = coeffs[0:n,1]

# Second derivative weights
wxx = coeffs[0:n,2] 
wxy = coeffs[0:n,3]
wyy = coeffs[0:n,4]
```

## Features

* RBF-FD weights for derivative operators
* PHS + poly based approximation
* Designed for 2-d problems

![Approximation of scattered function](./taylor_mesh.png)
‚ 
## License

This project is licensed under the MIT License. See [LICENSE](./LICENSE) for details.
