Rational log not working

[rwst]

sage: ZZ(8).log(2)
3

sage: QQ(8/27).log(2/3)
...
AttributeError: 'sage.rings.rational.Rational' object has no attribute 'log'

This can be done fast using mpz_remove (or ZZ.log) on numerator and denominator.

Depends on #21517

Component: numerical

Author: Ralf Stephan

Branch/Commit: 8c2ad23

Reviewer: Travis Scrimshaw

Issue created by migration from https://trac.sagemath.org/ticket/21429

[rwst]

Dependencies: #21518

[rwst]

Changed dependencies from #21518 to #21517

[rwst]

Branch: u/rws/21429-1

[rwst]

Commit: a16e1c4

[rwst]

Author: Ralf Stephan

[rwst]

New commits:

de1acfa	21517: handle ZZ.log(1/n)
4aba8a9	21517: avoid duplicate computations
a16e1c4	21429: implement rational log

[tscrim]

comment:6

Some micro-optimizations (I slightly care because this could be used in tight loops):

• You call Rational(m) twice.

• You call the perfect_power of anum and the like twice.

• an (and other like variables) are integers, correct? IIRC, a faster way to construct rational numbers is Rational(an, bn).

• It is likely faster to extract the numerator and denominator than to use a try/except block:

  sage: def foo(q):
  ....:     try:
  ....:         return ZZ(q)
  ....:     except (ValueError, TypeError):
  ....:         return None
  sage: def bar(q):
  ....:     if q.denom() == ZZ.one():
  ....:         return q.numer()
  ....:     else:
  ....:         return None
  sage: q = 4 / 3
  sage: %timeit foo(q)
  The slowest run took 7.89 times longer than the fastest. This could mean that an intermediate result is being cached.
  1000000 loops, best of 3: 1.24 µs per loop
  sage: %timeit bar(q)
  The slowest run took 42.67 times longer than the fastest. This could mean that an intermediate result is being cached.
  1000000 loops, best of 3: 212 ns per loop
  sage: q = 4 / 1
  sage: %timeit foo(q)
  The slowest run took 35.47 times longer than the fastest. This could mean that an intermediate result is being cached.
  1000000 loops, best of 3: 202 ns per loop
  sage: %timeit bar(q)
  The slowest run took 37.56 times longer than the fastest. This could mean that an intermediate result is being cached.
  1000000 loops, best of 3: 292 ns per loop

  You might be able to get even a bit more out of this by not going through numer and denom, but using the actual mpir data.

A cosmetic change:

-        -  ``m`` - default: natural log base e
+        -  ``m`` -- (default: natural log base `e`) the base
 
-        -  ``prec`` - integer (default: None): if None, return
+        -  ``prec`` -- integer (default: ``None``); if ``None``, return
            symbolic, else to given bits of precision as in RealField

[rwst]

Changed branch from u/rws/21429-1 to u/rws/21429-2

[rwst]

comment:8

This version also looks much better.

---

New commits:

997afa8

21429: rational log

[rwst]

Changed commit from a16e1c4 to 997afa8

[cheuberg]

comment:9

failing doctests (see patchbot).

[sagetrac-git]

Changed commit from 997afa8 to 2ded570

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

95e8048	Merge branch 'develop' into t/21429/21429-2
2ded570	21429: fix doctest fails

[tscrim]

Changed branch from u/rws/21429-2 to u/tscrim/rational_log_fix-21429

[tscrim]

comment:12

Sorry for having this fall off my radar. I made some fixes to the doc and some other small cosmetic changes. If you agree, then you can set a positive review.

---

New commits:

8c2ad23

Docstring and cosmetic changes.

[tscrim]

Reviewer: Travis Scrimshaw

[tscrim]

Changed commit from 2ded570 to 8c2ad23

[rwst]

comment:13

Thanks!

[vbraun]

Changed branch from u/tscrim/rational_log_fix-21429 to 8c2ad23