Skip to content

Integrate a power of an inverse function of c + d x^2 by its recurrence - #1904

Merged
Rafael-SOWNet merged 4 commits into
masterfrom
a-power-of-an-inverse-function-of-a-quadratic-by-the-recurrence
Oct 16, 2026
Merged

Rafael-SOWNet merged 4 commits into
masterfrom
a-power-of-an-inverse-function-of-a-quadratic-by-the-recurrence

Conversation

@Rafael-SOWNet

Copy link
Copy Markdown
Member

Part of #718.

For u the inverse sine or cosine of w = c + d x^2 with c^2 = 1, or its inverse hyperbolic cosine with c^2 = 1, (a + b u)^n for a whole n from 1 to 12 has an elementary antiderivative by a recurrence that closes at the first two powers:

integral (a + b u)^n = x (a + b u)^n + 2 e n b H (a + b u)^(n - 1) - 4 e b^2 n (n - 1) integral (a + b u)^(n - 2)

Here H = 1/(d x u'(w)), and e is 1 for the circular functions and -1 for the hyperbolic one. Each u has u'' = e w u'^3, and with c^2 = 1 the radicand of u' is d x^2 (2 c + d x^2) up to sign. So H u' = 2, and the derivative of the first two terms is (a + b u)^n plus a multiple of (a + b u)^(n - 2). The parser writes acosh(w) as ln(w + sqrt(w^2 - 1)), and the rule reads it from that form (and asinh with c^2 = -1 the same way).

SolveAPowerOfAnInverseFunctionOfAQuadraticByTheRecurrence checks H u' = 2 symbolically, with the derivative the library takes, rather than at sampled points. The sampled check declined the right answer to asin(1 + d x^2)^2. With c = 1 and the positive d it pins, w is past 1, on the cut of the inverse sine.

Checks

  • Exact checks at 50–2000 digits on nine shapes, each with d of both signs, on each real domain and off it: 432/432.
  • APowerOfAnInverseFunctionOfAQuadraticIntegralTest: 8 cases.

Corpus, measured against master 55a8044 (this branch then merged c92dbe0):

master this branch
pocket: every row of 5.1.5, 5.2.5, 7.1.5, 7.2.5 (719 statable) 642 668
families 0–8 samples 5.2.5 and 7.2.5 rows +3 (the same rows)
rows moved +26: 5.1.5:534–536, 541–543, 548, 549; 5.2.5:136–138, 143–145; 7.1.5:415–417, 422–424; 7.2.5:343–345, 350–352

Twelve of the 26 were timeouts on master; on this branch the slowest takes 0.47 s. No losses and no slower rows. Gate passed. The chain's suite aborted with the test host exiting and no exception logged, 14,372 tests in. Re-run alone it passed: 15,421 passed, 0 failed.

acos(1 + d x^2) was answered on master in 106 characters and is 76 here; 2.5.0 declined it.

🤖 Generated with Claude Code

https://claude.ai/code/session_012sonx8iAspMiwRwokT1Ura

Rafael-SOWNet and others added 4 commits October 16, 2026 06:22
For u the inverse sine or cosine of w = c + d x^2 with c^2 = 1, its hyperbolic cosine with
c^2 = 1, or its hyperbolic sine with c^2 = -1, integral (a + b u)^n is
x (a + b u)^n + 2 e n b H (a + b u)^(n - 1) - 4 e b^2 n (n - 1) integral (a + b u)^(n - 2), with
H = 1/(d x u'(w)), and it closes at the first two powers. Rubi's 5.1.5, 5.2.5 and 7.2.5
(a + b acos(1 + d x^2))^4 and the like were declined.

Part of #718

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012sonx8iAspMiwRwokT1Ura
…se-function-of-a-quadratic-by-the-recurrence

# Conflicts:
#	Sources/AngouriMath/Functions/Continuous/Integration/IndefiniteIntegralSolver.cs
#	Sources/AngouriMath/Functions/Continuous/Integration/Integration.Definition.cs
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012sonx8iAspMiwRwokT1Ura
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