I highly doubt any of the reflectors sold commercially in this hobby are computer designed. "Computer designed" usually means that somebody drew a picture of it on a computer. It's the engineer/mathematician that designed the reflector. It's fairly trivial to design a reflector assuming a point source of light in 3d, and even easier for the 2d scenario for, say, a T5 bulb. To first approximation, you'd put the bulb at the focus of a parabola. To get more into it, you'd have to consider that the reflector doesn't cover the whole area to be illuminated (like luminarcs tend to do). Also, for a parabola, even though the reflected light is collimated, it is more intense in the perifery, which, to some degree, couteracts the intensity of the light coming directly from the bulb, which is more intense in the center. You could also consider that the bulb is not a perfect point, and that the desired coverage is usually square, among other considerations. I doubt that the common reflector companies take these considerations into account mathematically. They probably just tweak the bulb's location and the angle of the panels a bit to even things out somewhat. The reflectors are not even curved, which tells me that they are not as good as they could be.
Mylar is the polymer and it doesn't necesarily have to be mirrored. I'd bet you could get it pretty flat by stretching it over some float glass and taping it behind. I'd suggest using mirrored aluminum sheets, like those from analod. More $$ than mirrored mylar, but you'd still save a ton over commercial prices. It's more durable than the coated mylar stuff.
Good luck with the construction....
G1
