Skip to content

Leave a power of a ± a coth alone to the exponential substitution - #1982

Merged
Rafael-SOWNet merged 1 commit into
masterfrom
leave-a-power-of-a-plus-a-coth-to-the-exponential-substitution
Oct 21, 2026
Merged

Rafael-SOWNet merged 1 commit into
masterfrom
leave-a-power-of-a-plus-a-coth-to-the-exponential-substitution

Conversation

@Rafael-SOWNet

Copy link
Copy Markdown
Member

#1956 integrates a function of the hyperbolic tangent alone under u = tanh. For a power of a + a coth(L) or a - a coth(L) alone, the 1 ± u of the Jacobian cancels into the base in u, which leaves a root over three linears. In W = e^(2L) the same integrand is a binomial differential written in a few terms. Under #1956, (1 + coth(x))^(7/2) became 14,533 characters and 1.3 GB, where the substitution writes 287 characters at 0.13 GB. The rule now declines when the only factor holding u is a power that is not whole of (p u + q)/(r u + s), with p = ±q and q, r not zero.

Each problem alone, before #1956 → master → this branch (allocation, answer length):

Problem Before #1956 Master Branch
6.4.2:99 (1+coth(x))^(7/2) 0.13 GB, 287 1.31 GB, 14,533 0.13 GB, 287
6.4.2:100 0.13, 235 0.48, 10,435 0.13, 235
6.4.2:101 0.13, 180 0.22, 3,785 0.13, 180
6.4.2:102–105 0.13, 140–341 0.05–0.06, 1,342–2,734 as before #1956

On 102–105 the branch allocates about 0.07 GB more than master, for answers 4 to 10 times shorter. Every answer in the branch column is character for character the one before #1956.

Measured on 8270086f, branch against master:

Run Problems Master Branch
Pocket: all of Rubi's family 6 3,690 3,616 solved 3,616 solved
Family 1, 40 a file 1,381 1,368 1,368
Family 0, whole 1,814 1,789 1,789
Families 2 to 8, sampled 2,410 2,380 2,380

The three moves, 6.1.4:14, 6.7.1:653 and 6.7.1:737, come out the same on both builds when re-run alone; they are problems that take 12–20 s against a 5 s budget. #1956's 36 gained rows answer as on master, and the other 50 rows of 6.3 and 6.4 that #1956 reaches are unchanged. 6.4.2:185 looked more expensive in the pocket, but alone it allocates 0.417 GB on both builds and gives the same answer. v2.5.0 declined all of these rows, so BREAKING-CHANGES.md has no entry: the branch restores the unreleased spelling from before #1956.

AFunctionOfTheHyperbolicTangentIntegralTest differentiates the answers back and requires them to be under 1,000 characters; master's are 1,342–14,533.

Part of #233.

🤖 Generated with Claude Code

https://claude.ai/code/session_012sonx8iAspMiwRwokT1Ura

#1956 integrates a function of the hyperbolic tangent alone under u = tanh.
For a power of a + a coth(L) or a - a coth(L) alone, the 1 ± u of the
Jacobian cancels into the base in u and leaves a root over three linears,
where in W = e^(2L) the integrand is a binomial differential written in a
few terms: (1 + coth(x))^(7/2) became 14,533 characters and 1.3 GB, where
the substitution writes 287 characters at 0.13 GB. The rule now declines
when the only factor holding u is a power that is not whole of
(p u + q)/(r u + s) with p = ±q and q, r not zero.

Rubi's 6.4.2:99-105 have their answers from before #1956 again, character
for character. #1956's 36 gained rows answer as on master, and the other 50
rows of 6.3 and 6.4 it reaches are unchanged.

Part of #233.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012sonx8iAspMiwRwokT1Ura
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

Status: Done

Development

Successfully merging this pull request may close these issues.

1 participant