Python Systematic Parameter Generator
Python R C++ Other
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
contrib
examples
scripts
skeleton
spg
LICENCE
Makefile
README
README.rst
TODO
VERSION
setup.py
spg.dia

README.rst

PySPG is a set of python classes aimed to those of you that must (or wish to) run programs in which some parameters change.

For example: Let's suppose you want to run simulations in which three parameters are involved, let's say D, k, a. These are some possible scenarios

You want to run a simulation in which D, k are kept fixed while a linearly changes between (eeeer...) 0 and 5 with a step of 0.2. It's really easy: in the language of your choice you write a program that performs the simulation and before and after the important part of a simulation, you just add a loop on the a variable. Now you are done But after that, you want to keep fixed D. Now both k and a vary linearly between 0 and 5 with a step of 0.2. It's easy once again: you just add an external loop on the k variable. And you're done again. But note that you had to recompile your code without changing the important part of your code: The simulation.

After those simulations you realise that the scale relevant for the variable a is not a linear one, but logarithmic. Although the change is easy, you must recompile. And if you want to run a simulation in which the variation is on D variable? Obviously compile the whole thing again... And if the variation must be exponential??? ...

Well, perhaps you do not have to recompile if you program in an interpreted language. But what you are doing is touch on, and on, and on again your source code. The probability of doing something weird increases.

The only relevant information your program returns is the measures for each parameter set. The parameter variation is something subsidiary of the main point of the program, that is performing measures. The values of the variables can be set from outside. And this is the point of PySPG.

With PySPG you can extract one layer of complexity from your compiled code. For long simulations, is obvious that the time your program takes to run is NOT in the loops of changing parameters. In this way, you can avoid the problem of writing boring code and just write a simple text file that will launch the other program for you.

PySPG generates a directory hierarchy that allows you to easily navigate your data.

Is it all?

No. Although not so well documented yet, PySPG also features a generator of plots for your simulations. It can automatically generate 2D-plots in the format of Grace, and for a future version, a 3D utility is planned. Also an automatic report generator (TeX-based) is half-done. And it is GPL'd, so you can extend it as much as you wish.