Mars Pathfinder does in 26 pages, running in 30 seconds on a 486 PC (this was circa 2000) what NASA needs millions of lines of code running on a massively parallel Cray supercomputer running for hours – days!
It solves the problem directly. The solution is as accurate as anything NASA has, given the constraints.
Its accuracy is comparable because every computation by NASA’s code has roundoff error. Millions of calculations accumulate significant inaccuracy. Because Mars Pathfinder is a direct method with minimal iteration, its accumulated error is much less.
Here we are, 25 years later, and I guarantee that these 26 pages of FORTRAN code are the most compact, precise algorithm ever created.
Bar X Software
I suppose everybody in the Aerospace Department thought I was cheating. Impossible for a young engineer to write that kind of code. (Personally, I suspect none of them could even understand FORTRAN.)
If they had asked, I would have shown them my resume and list of published papers. Half a dozen conference papers on dBase applications for engineering design in the construction industry. Commercial apps. I even had my startup. You can visit the little website I had during grad school, where I gave them away to promote my McGraw-Hill textbooks.
But, alas. This was right in the middle of the transition to C++. FORTRAN was old-school, archaic.
I dare anybody to engineer my FORTRAN code into C++ or any other language to do calculations faster, more accurately, or with a smaller footprint.
The DOS version 1.0 was only 145 kb. The full windows program fit, zipped, on a 1.5 MB floppy.