REgular Grid Linear Interpolator, capable to deal with spectral library or similar model data.
This package implements the basic bilinear interpolation in multi-dimension. However, as commented in Numerical Recipes:
Bilinear interpolation is frequently “close enough for government work.” As the interpolating point wanders from grid square to grid square, the interpolated func- tion value changes continuously. However, the gradient of the interpolated function changes discontinuously at the boundaries of each grid square.
As a consequence, the interpolation model can not, in general, be used as a forward model and incorporated with Gradient-based optimization algorithms to estimate parameters.
Bo Zhang, bozhang@nao.cas.cn
- for the latest stable version:
pip install -U regli - for the latest github version:
pip install -U git+git://github.com/hypergravity/regli - for the Zenodo version: https://zenodo.org/record/3461514
from regli import test
test()output:
regli.interp3 x 10000: 0.5675415992736816 sec
regli.interpn x 10000: 2.5326197147369385 sec
rgi x 10000: 5.4028871059417725 sec
# import Regli
from regli import Regli
import numpy as np
# construct grid coordinates
x1 = np.linspace(-1, 1, 30)
x2 = np.linspace(-1, 1, 30)
x3 = np.linspace(-1, 1, 30)
# initiate regli using coordinates
regli = Regli(x1, x2, x3)
# an arbitrary function of coordinates (for demo)
f = lambda _x1, _x2, _x3: _x1 + _x2 + _x3
# regli.flats stores flattened coordinates of ND grid
flats = regli.flats
# evaluate your function on flats
values = np.array([f(*_) for _ in flats]).reshape(-1, 1)
# set values for regli
regli.set_values(values)
regli(pos) # use any of the 3 ways to interpolate
regli.interpn(pos) # method 1 is equivalent to 2
regli.interp3(pos) # this is accelerated for 3DBibTex:
@misc{https://doi.org/10.5281/zenodo.3461514,
doi = {10.5281/zenodo.3461514},
url = {https://zenodo.org/record/3461514},
author = {Zhang, Bo},
title = {hypergravity/regli: The Zenodo version},
publisher = {Zenodo},
year = {2019}
}
Please go to https://search.datacite.org/works/10.5281/zenodo.3461514 for other formats.