The basic space is a 10 by 6 by 3 rectangular room (length units are
irrelevant, but think of them as meters if you like).  At the center
of the ceiling is a single square fixture that measures .5 on a side.
The grey emittances and reflectances are as follows:

	Surface		Emittance (w/m^2)	Reflectance (%)
	=======		=================	===============
	Source		31.416			0
	Ceiling		0			70
	Walls		0			50
	Floor		0			30

The calculation points on the floor and ceiling are separated by .5 in
each dimension, starting .25 from the edge.  Note that this means they
are not symmetrical about the center.  The coordinates are measured
from the origin at one corner of the room.  The longer wall is aligned
with the x-axis, and the z-axis points up.

This room contains two vertical partitions, each full height (3).
The first runs from (x,y) = (3,3) to (3,6).  The second runs from
(x,y) = (7,0) to (7,3).  Both partions are .1 thick in x and have
the same reflectance properties as the other grey walls.

There are three versions of this room.  The first version, room1, has
a single ideally specular wall surface (reflectance = 75%) at y==0.
The second version, room2, has two specular wall surfaces, one at y==0
and the other at x==0.  Since these walls are at right angles, there
can be at most two specular reflections before striking a diffuse surface.
The third version of the room, room3, also has two specular wall surfaces,
one at y==0 and the other at y==6.  Since these surfaces face each other,
there is no real limit to the number of specular interreflections.

Results are stored in three forms.  The first is the raw output of the
Radiance rtrace(1) command, which is stored in a file ending in .res.
This is in the form x y z r g b for each point on the floor and each
point on the ceiling.  The second form is a plot file (ending in .plt)
for bgraph(1) and contains one curve for each Y position.  The third
form is a PostScript version of the plot files, ending in the .ps suffix.

All results are given in units of radiance (watts/sr/m^2).  To convert
to radiosity (w/m^2), simply multiply by pi and divide by the diffuse
reflectance of the surface (.7 for the ceiling, .3 for the floor).  I
prefer to use radiance not because that's the name I chose for my software
but because radiance is consistent with any type of surface reflectance.
It does not make sense to talk about the radiosity of a non-Lambertian
surface, for example, whereas both radiance and irradiance make perfect
sense.
