andrewcharles / pyticles Goto Github PK

========
Pyticles
========

Pyticles is a set of python modules for physical modelling using particles.

The main conceptual objects are particles, neighbour lists, boxes and forces.

Interfaces to optimised cython modules, and wrapped fortran modules are also
provided. For example:

properties.py -- computes smooth particle properties using only python
c_properties.py -- cython optimisation of properties.py
f_properties.py -- calls wrapped Fortran modules

Dependencies
------------
Numpy
Scipy (ubuntu packages for scipy and numpy are just fine)
Cython (sudo apt-get setuptools, easy_install cython)


For visualisation modules:
Pyglet
pygarrayimage (<BS>http://pypi.python.org/pypi/pygarrayimage)

rsp-fsp-spac



TODO
- write complete sph force and rates of change routine.
- it can even be in fortran to ease the array copying overhead.
- write fortran speed render computation.


FORTRAN INTEGRATION NOTES
=========================


Files
=====

controller.py
-------------
Forces, boundary conditions, anything that mutates particles is
a subclass of controller.


forces.py
--------- 
Each Force class instance contains a reference to two particle systems and a
neighbour list. But a neighbour list should contain references to the particles
it maps. This looks like a design flaw. The force should operate as a one body
or n body force. Given a list of particles, or particle pairs. A force iterates
over a neighbour list.


neighbour_list.py
----------------- 
Neighbour lists contain a reference to one or two particle systems.


particles.py
------------
Instances of the particle class have attributes that are mostly numpy arrays.

The update() method calls the numerical integrator to advance time step.

config is a dictionary of configuration options.

The ParticleSystem class.

The SmoothParticleSystem class.



About
=====

The main driver is my smooth particle research. There are a number of
front end scripts for pyticles:

spana.py
    This is the front end for testing algorithms related to my research.

ospana.py
    This is a front end to the cython and wrapped fortran optimised version of
    the code.

bspana.py
   Batch front end - no graphics.
   Writes output to file.

pyticles.py
    This is where the bulk of the work on the interactive gui is going. This
    doesn't use any SPH code at the moment - any work relevant to any particle
    system goes here.

particle_man.py
    This is to be an effort to produce a simple game using the pyticles
    modules. It is pretty much nothing at the moment.

vspana.py
    Opens an output file and views it.


Road Map
========
Integrate some elements with lepton.



Andrew Charles
[email protected]


1. More scripted approach to execution
2. Visualisation in real time
3. Fast prototyping of new algoritms
4. Integration with the fortran code

Pyticles uses Pyglet by Alex Holkner.



Cython Notes
============ 
After the success of the simple nbody Cython, I have effectively
forked the code by writing particle_module.pyx and forces_module.pyx. At this
stage I just want to get a feel for the amount of difference in the code and
the amount of speed increase. Later I can think about organising it better.

Having trouble with the modules using objects that reference each other.

build.sh rebuilds the cython modules.


Zoom/Pan/Rotate Notes
=====================
Zoom and rotation

glScalef(d,d,d)