Running v 3.1.3; if you run a fit for long enough the count of formulas overflows. After 24,894 s it indicated 2.67e9 formulas, but I saw it overflow at about 20,000 s (it was something like 2e8). It seems to be an unsigned integer overflow (4.3e9) that occurred thrice given a fit to previous runs suggesting it should be at about 15e9 formulas by this runtime.
Maybe consider an unsigned 64-bit integer?
