Skip to content

Read the derivative of a sum of exponentials over another off the integrand over the other's square - #2040

Merged
Rafael-SOWNet merged 1 commit into
masterfrom
the-derivative-of-a-quotient-over-a-sum-of-exponentials
Oct 22, 2026
Merged

Rafael-SOWNet merged 1 commit into
masterfrom
the-derivative-of-a-quotient-over-a-sum-of-exponentials

Conversation

@Rafael-SOWNet

Copy link
Copy Markdown
Member

Part of #233.

  • The derivative of a sum of exponentials over another, read off the integrand over the other's square. Rubi's 6.7.1 x (b - a sinh(x))/(a + b sinh(x))^2 ran past the budget, in three steps:

    1. the split took the bracket's terms apart;
    2. the substitution in e^x answered each over the square of a quadratic, with a piecewise on its discriminant;
    3. parts beside x integrated that piecewise again.

    For a sum D of exponentials of x, the integrand over D^2 is now read as (N/D)', where N is a constant plus a constant times each exponential in D.

  • How N is found.

    • The exponentials' span is closed under the derivative, so N' D - N D' is linear in N's constants. With each exponential a whole power of the first, w, the identity read at six values of w is a linear system.
    • N and N + c D share a derivative over D. The solution with no constant is tried first, and that is -cosh(x)/(a + b sinh(x)).
    • Beside x, by parts, the integral of N/D is written as a constant times ln(D) where N is a constant times D', and asked as a question otherwise.
  • Where it sits. It is the hyperbolic twin of SolveAsTheDerivativeOfASineRaisedFromNothingBesidePowersOfTheCosine, dispatched beside it. There are no hyperbolic nodes, since the parser writes sinh as exponentials.

Rubi row master 78cd31b this branch
6.7.1:643 x (b - a sinh(x))/(a + b sinh(x))^2 past 30 s 123 characters
6.7.1:644 x (b + a cosh(x))/(a + b cosh(x))^2 past 30 s 119 characters
6.7.1:1178 sech(x)^2 tanh(x)^2/(2 + tanh(x)^3)^2 117 characters, 16 s, 9.7 GB -1/3/(2 + tanh(x)^3), 60 characters, 55 ms
(b - a sinh(x))/(a + b sinh(x))^2 7,861 characters, a piecewise 71 characters

Chain against master 78cd31b:

  • The pocket of 2,652 rows with an exponential and a squared quotient went from 2,641 to 2,643 solved, with 0 wrong.
  • Every family's sample is unchanged.
  • The two rows that moved, 643 and 644, run alone, went from timeout to solved.
  • 8,181 rows were compared for allocation, and none grew past 1.3x and 300 MB.

The full unit suite passes locally: 16,554 passed, 0 failed. The BREAKING-CHANGES lengths were measured on the branch. Rebased onto 37e2b1a.

🤖 Generated with Claude Code

https://claude.ai/code/session_012sonx8iAspMiwRwokT1Ura

…egrand over the other's square

Rubi's 6.7.1 x (b - a sinh(x))/(a + b sinh(x))^2 ran past the budget: the split
took the bracket's terms apart, the substitution in e^x answered each over the
square of a quadratic with a piecewise on its discriminant, and parts beside x
integrated that piecewise again. For a sum D of exponentials of x, the
integrand over D^2 is now read as (N/D)', N a constant plus a constant times
each exponential in D. The exponentials' span is closed under the derivative,
so N' D - N D' is linear in N's constants, and with each exponential a whole
power of the first, w, the identity read at six values of w is a linear
system. N and N + c D share a derivative over D; the solution with no
constant is tried first, which is -cosh(x)/(a + b sinh(x)). Beside x, by
parts, the integral of N/D is written as a constant times ln(D) where N is a
constant times D', and asked otherwise. The trigonometric twin is
SolveAsTheDerivativeOfASineRaisedFromNothingBesidePowersOfTheCosine's; there
are no hyperbolic nodes, since sinh is parsed into exponentials.

Part of #233.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012sonx8iAspMiwRwokT1Ura
@Rafael-SOWNet Rafael-SOWNet added this to the 2.6.0 milestone Oct 22, 2026
@Rafael-SOWNet
Rafael-SOWNet merged commit 8d5f693 into master Oct 22, 2026
34 checks passed
@github-project-automation github-project-automation Bot moved this from Todo to Done in AngouriMath Oct 22, 2026
@Rafael-SOWNet
Rafael-SOWNet deleted the the-derivative-of-a-quotient-over-a-sum-of-exponentials branch October 29, 2026 00:44
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