Skip to content

Quick Start

Install pyself

pySELF can be installed using pip.

pip install pyself==0.0.0

If you'd like to install a non-tagged version, from the main branch for example, see Installation.

Read in SELF data from a file

Suppose that you have run a simulation with SELF using the Model2D class or one of its children (e.g. LinearShallowWater2D). The simulation has generated a set of solution.*.h5 files. In the scenario described here, we're interested in loading in the contents of a file called solution.example.h5 in the current working directory.

#!/usr/bin/env python

import self.model2d as model2d


# Initialize your model
model = model2d.model()

# Load in the model data from file
model.load('solution.example.h5')

Working with model data

The model class in pyself provides many of the same attributes as the equivalent class in SELF.

  • solution - The solution array at element interior (quadrature) points
  • solutionGradient - The gradient of the solution at element interior (quadrature) points
  • flux - The conservative flux vector at element interior (quadrature) points
  • fluxDivergence - The divergence of the flux vector at element interior (quadrature) points

Each of these attributes are stored as Dask arrays. The data stored in these attributes can be accessed in the same way as you would numpy arrays.

The dimension ordering that is used for these arrays (in 2-D) is as follows:

  1. Element dimension
  2. Variable dimension
  3. Quadrature dimension (second computational direction)
  4. Quadrature dimension (first computational direction)
  5. (Optional) Physical Vector direction

For example,

solutionData = model.solution[iel,ivar,:,:]

stores the ivar-th solution variable within the iel-th element in solutionData.

Working with mesh, geometry, and interpolant data

Additionally, the model class has mesh and geometry objects as attributes. In version 0.0.0 (the current version), the mesh object is not filled in; this means that you currently don't have access to connectivity information. The geometry class, on the other hand, contains the following attributes :

  • interp - The Lagrange interpolant object that defines coordinates and operators in computational space for the geometry and model data
  • x - The coordinates in physical space at element interior (quadrature) points

As with the model attributes, the x attribute is stored as a Dask array. To obtain the x and y positions as their own arrays, you can do something like the following,

# Get x and y physical coordinates for all elements and quadrature points
x = model.geom.x[:,:,:,:,0]
y = model.geom.y[:,:,:,:,0]