Skip to content

Latest commit

 

History

History
89 lines (66 loc) · 4.88 KB

README.md

File metadata and controls

89 lines (66 loc) · 4.88 KB
Raw

Rescal-snow utilites

This folder contains scripts for configuring, running and analyzing Rescal-snow simulations. These are run through Python3 and bash. They make the Rescal-snow workflow more consistent and reproducible by making it possible to do most of the work in Python and store both input and output in Python classes.

rescal_utilities.py

Utilities to set up Rescal-snow runs quickly and easily. These tools are designed to aid parameter space explorations, sensitivity analyses, and large batches of runs. The following classes manage most Rescal-snow options and inputs (excepting those in the "real_data" file):

Parameters : Class to hold, update, change, read, or write all the parameters that ReSCAL needs to run

RunScript : Class to hold, update, change, read, or write a ReSCAL run script with appropriate flags

DesignRun : Umbrella class to hold, update, change, read or write all parameters, sorting them into Parameters and RunScript

Example usage (python3):

# Create the DesignRun object and (optional but advised) a basic description
my_run     = rescal_utilities.DesignRun()
my_run.set_name("my_run")
my_run.set_header("An example run varying L, D and Lambda_S, and looking for the rescal executable in build")

# Describe values for some subset of Rescal-snow parameterescal.sbatch
# Others will take defaults from rescal_utilities.Parameters._default_parameters() and rescal_utilities.RunScript._default_options()
parameters = {'L':500, 'D':50, 'Lambda_S':0.01, 'rescallocation':'../../build'}
my_run.set_parameters(parameters)
for the output should be created. The default is RESCAL_SNOW_ROOT/data_runs.
# Create a run script named my_run.run and a parameter script named my_run.par
my_run.write()

See also the usage in scripts/utilities/param_space_exploration_example.py

Example: param_space_exploration_example.py

This is an example file showing how to run a parameter space exploration. In this example, most simulation parameters are set to fixed values, then 30 runs are created to vary:

  • five snowfall rates, controlled by parameter Lambda_I
  • six wind speeds, controlled by parameter Tau_min,
  • and all 30 combinations

The script creates 30 new directories, one for each run. Each directory is seeded with a run script, a parameter file, and rescal and genesis executables. This allows all 30 simulations to run in parallel withou t interferance.

Usage is described in example 4 of ../../docs/rescal-snow-tutorial.md.

datarun.py

A python interface to Rescal-snow. This sets Rescal-snow parameters for one or more runs, runs the simulation, and keeps the results in pythonic arrays for ongoing processing. A DataRun object takes in the parameters and meta-parameters required to run Rescal-snow. The DataRun can receive and processthe output of Rescal-snow while Rescal-snow is running. To run Rescal-snow using a DataRun instance, the environment variable RESCAL_SNOW_ROOT should be defined and be the path of a Rescal-snow installation. Also, a directory for the output should be created. The default is RESCAL_SNOW_ROOT/data_runs.

# Describe some subset of Rescal-snow parameters and a directory for this run
parameters = {'L':500, 'D':50, 'stop after':'200_t0', 'output interval':'50_t0'}
dirname    = 'myrundirectory'

# Create and run the run
thisrun    = datarun.DataRun(parameters, dirname)
thisrun.setup()
thisrun.run()

Example usage is shown in example_pyrescal.py

Example: example_pyrescal.py

This example tests the functionality of datarun.py in a high-performance computing environment. Usage is described in ../../docs/rescal-in-python.md

heightmap.py

Utilities to create, read and visualize Rescal-snow height maps, a 2D surface height map which is standard Rescal output. These are managed through a HeightMap class.

Utilities for creating Rescal-snow heightmaps (these may be used as initial conditions for the Rescal-snow simulation; see the INPUT_ELEVATION CSP_TEMPLATE in src/genesis.c).

  • invader_template
  • gaussian_hill
  • make_sinusoid
  • scale

Utilities for analyzing Rescal-snow heightmaps (most of these use fourier transforms, as snow/sand self-organization has strong emergent wavelengths):

  • fft2d_analyze
  • fft2d_analyze_map_pic
  • fft2d_crop_blur
  • fft2d_center_blur

Utilities for visualizing height maps:

  • make_surface

cellspace.py

Utilities to read, write, modify and visualize Rescal-snow cell spaces, the full state of the cellular automaton. Managed through CellSpace class. Used by datarun.py. May be used for visualizing and modifying internal sand/snow structures that are not adequately represented by 2D HeightMaps.