Skip to content

Latest commit

 

History

History
66 lines (38 loc) · 1.73 KB

README.rst

File metadata and controls

66 lines (38 loc) · 1.73 KB

WrightSim

A simulation package for multidimensional spectroscopy.

Installation

$ git clone https://github.com/wright-group/WrightSim
$ cd WrightSim
$ python setup.py develop

Note: This will install all required dependencies. PyCUDA is not a required dependency in general, but is required if GPU simulations are desired.

$ pip install pycuda

PyCUDA requires an Nvida graphics card and drivers, and the CUDA libraries installed.

Usage

An example script is provided at ./scripts/target.py

This script can be modified to suit an individual simulation.

The basic steps are:

  1. Select an experiment
  2. Set up the axes of the scan
  3. Set the time interfal and buffers
  4. Create a Hamiltonian object
  5. Run the scan
  6. (optional) review the results

Level of parallelism is selected by the mp parameter of teh exp.run method.

  • True or "cpu" enables CPU multiprocessing
  • "gpu" enables CUDA
  • False or "" runs in single threaded mode

The script is set up to read the dimensions from arguments. it can be run like so:

$ ./target.py 32 16

This will run a 3D simulation of 32x32x16 Freq-Freq-Delay.

It will print the time it took to compute.

An example of how to visualize results is provided at the end of the script, but is not active, as it only works for 2D scans.

To convert this script to a 2D scan and visualize, uncomment the lines which set exp.d2.points and exp.d2.active and remove the triple quotes around the plotting routine. There it will ignore the second argument, producing a 32x32 2D frequency spectrum. (The second argument is still parsed, so it is required, just ignored.)