pyspde 0.0.6

PySPDE is a Python library for performing simulations and kriging of non-stationary spatial gaussian random fields with Matérn covariance, by solving the following Stochastic Partial Differential Equations (SPDEs)

PySPDE

PySPDE is a Python library for performing simulations and kriging of non-stationary spatial gaussian random fields with Matérn covariance, by solving the following Stochastic Partial Differential Equations (SPDEs):

$$ ({\kappa}^2 - {\nabla} {\cdot} H(x) {\nabla} )Z(x) = {\tau}(x)W(x) \quad x \in \mathbb{R}^2 $$

The theory is proposed in Fuglstad (2014).

Instalation

On Debian/Ubuntu systems:

sudo apt-get install libsuitesparse-dev
pip install pyspde

On Windows systems:

conda install -c conda-forge suitesparse
pip install pyspde

Basic Usage

Imports:

from pyspde import anisotropy, Grid, Spde, anisotropy_from_svg

Define the anisotropy and the grid:

atpy = anisotropy_from_svg(r'assets/anticline.svg', beta=50, gamma=0.5)
grid = Grid(nx=400, ny=200, anisotropy=atpy)

Define SPDE and Simulate:

sp = Spde(grid, sigma=1, a=50)
Z = sp.simulate(seed=0)