Tried the new improved version, which I saved as "clockgearrevised" in a folder on my desktop called Clockgear, with the original and my edit of that.
Here's what I got (took about 1 minute):
I'm guessing I got only 4 answers, all with the same target result, with the same 4 numbers for the wheels and pinions except switched around,
shogun@shadow:~$ cd Desktop/Clockgear
shogun@shadow:~/Desktop/Clockgear$ ls clock*
clockgearoriginal.py clockgearrevised.py clockgear-shogun.py
shogun@shadow:~/Desktop/Clockgear$ python clockgearrevised.py
data: (1.0027376804380288, 2.1956197127082078e-07, 51.0, 79.0, 49.0, 82.0)
data: (1.0027376804380288, 2.1956197127082078e-07, 51.0, 79.0, 82.0, 49.0)
data: (1.0027376804380288, 2.1956197127082078e-07, 79.0, 51.0, 49.0, 82.0)
data: (1.0027376804380288, 2.1956197127082078e-07, 79.0, 51.0, 82.0, 49.0)
because those were the most accurate answers, and less than the error value you assigned, so it posted only those 4 results.
Since that only took 1 minute to finish, I'd be willing to try up to, say, 8 gears and pinions, and let it run for a day. Really, the calculation time doesn't
bother me much. I'm fine with letting it run for a day or a few days if I get more accurate results with wheels and pinions.
For now, would you be able to try a version with 4 wheels and 4 pinions, and see what happens?
I'll try 8 or 10 or 20 later on. I also think I understand completely how the code works and limits now.
Btw, How does it know to move up or down the values though- it realizes automatically which direction to go based on which number is higher to start?
This is quite fun Fred