Skip to content

fix: correct Putnam-like Set_2/B6 integral identity - #10

Open
sylvesterkaczmarek wants to merge 1 commit into
google-deepmind:masterfrom
sylvesterkaczmarek:fix/putnam-set2-b6-integral
Open

fix: correct Putnam-like Set_2/B6 integral identity#10
sylvesterkaczmarek wants to merge 1 commit into
google-deepmind:masterfrom
sylvesterkaczmarek:fix/putnam-set2-b6-integral

Conversation

@sylvesterkaczmarek

Copy link
Copy Markdown

Summary

Fixes #7.

putnam_like/Set_2/B6 currently asks models to prove a false identity with cos x on the right-hand side. The corrected integrand is sin x.

This change:

  • corrects the problem statement from cos x to sin x;
  • repairs the grading scheme so the comparison function uses the sine integral consistently;
  • fixes the differential-equation step to (f-g)'' + (f-g) = 0, matching the resulting trigonometric general solution;
  • adds ERRATUM.md to preserve provenance for the released samples.

The existing samples/ directory is intentionally unchanged. Those files remain historical model outputs and grades produced against the original released cos x question.

Validation

I numerically cross-checked the three integrals with high-precision mpmath evaluation:

∫_0^∞ e^(-2x)/(1+x^2) dx  = 0.39902098859418384689
∫_0^∞ sin(x)/(2+x) dx     = 0.39902098859418384689
∫_0^∞ cos(x)/(2+x) dx     = 0.14454530303733242046

This confirms the sine version and rejects the published cosine identity. The corrected integration-by-parts derivation in the rubric yields g''(t) = 1/t - g(t), matching the corresponding equation for f.

No historical sample or grade files are modified.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

1 participant