Simple file format for import and export

[Andreas Ipp]

As more and more feature get added, we should think about a suitable file format for loading and saving simulation states. A new student will work on this, but it is good if we define the file format first. I'm thinking of a variation of the Windows INI file format ( http://en.wikipedia.org/wiki/INI_file ) because of its simplicity. (Maybe Jan would like to convince us that XML would be better, but I think that would be overkill for our purposes. Scientists only know how to edit simple text files :-) )

Here is a suggested file format.

------------------

# OpenPixi file format. 

# Comments start with '#' and are allowed anywhere.

[Simulation]

version=0.4  # version number is always good to know

width=100

height=100

timestep=0.01

solver=BorisRelativistic

grid=YeeGrid

boundary=PeriodicBoundary

[ConstantForceRelativistic]

bx=1 # add some background magnetic field.

# other values are 0 by default.

[Particle] # add a particle

p=20,25  # position

v=0,0  # initial velocity

radius=1

mass=0.1

charge=+1

[Particle]  # create a second particle

p=70,25

v=-5,0 # let it collide with first particle

# if parameters like mass or charge are omitted,

# then parameters from previous particle are used.

-----

Instead of specifying each particle separately, one could also create many random particles at once:

--------------------

[RandomParticles]

n=1000 # number of particles

vmax = 2 # maximum velocity, randomly distributed

radius=0.23

mass=0.1

charge=+1

[RandomParticles]

n=500 # create 500 more particles

charge=-1 # with opposite charge.

--------------------

Additionally, simulation data could be written to a file:

----------------

[Output]

filename=C:\mydata\two-particle-collision1.dat

data=time, p.1.x, p.1.y  # three columns: time, particle1.x, particle2.x

[Output]

# Let's write at the same time the data for particle 2 into another file

filename=C:\mydata\two-particle-collision2.dat

data=time, p.2.x, p.2.y

-------------

What do you think?

[Jan Kis]

I do not really get what for is the section [output] in your example doing but maybe it is not important.

It might be also useful to have a section for referencing further files with simulation data. Then one can have a single file with some testing particles which one can use in several different simulation setups. 

[ref]

file=C:\...

Maybe Jan would like to convince us that XML would be better

Actually, I think it would be less readable and too verbose for our purpose.

This link might be useful later on so that the person implementing this does not have to create its own parser for ini files:
http://stackoverflow.com/questions/190629/what-is-the-easiest-way-to-parse-an-ini-file-in-java