grondilu
searching PlanetScale…
- 1.▲
- 2.▲
- 3.▲
- 4.▲
- 5.▲
- 6.▲
51 ms·
- 481.▲
- 482.▲
- 483.▲
It seems to me that it makes more sense to approximate sinPi(x) = sin(pi x) scaling[x_] := 8/3 (x - x^3)
indices = {0, 2, 4, 6, 8, 10}
inner[g_, h_] := NIntegrate[g h/Sqrt[1 - x^2], {x, -1, 1}, WorkingPrecision-> - 484.▲
- 485.▲
- 486.▲
- 487.▲
- 488.▲
- 489.▲
- 490.▲
- 491.▲
- 492.▲
- 493.▲
- 494.▲
- 495.▲
- 496.▲
- 497.▲
- 498.▲
- 499.▲
- 500.▲
- 501.▲
- 502.▲
- 503.▲
- 504.▲
- 505.▲
- 506.▲
- 507.▲
- 508.▲
- 509.▲
- 510.▲
More ›