On Thursday, 16 October 2025 at 09:38:30 UTC-7 Georgi Guninski wrote:
This bug appears to be very rare and it passed many testcases. It is not a large number issue: I suspect the error does arise from a large number issue, but apparently maxima doesn't always bump into it, The number n does not fit in a system float: sage: float(n) inf The mpfr implementation of floats allows for larger exponents so there it's fine: sage: RR(n) 5.80960599536996e924 and then the log isn't so large, so there are plenty bits to get the floor correct: sage: RR(n).log()/log(2.0) 3072.00000000000 which is still a bit of a silly way of computing the 2-log. gmp has better tools: sage: n.nbits() 3073 I think at this point the place to report the specific problem is to maxima. Perhaps they can find a subtle issue on their end that can solve this. For sage, the thing to look at is how sage computes floors of symbolic expressions that happen to represent a number. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion visit https://groups.google.com/d/msgid/sage-devel/9f8ba9a5-43e1-45c1-a8f4-2532eab9aa8fn%40googlegroups.com.
