|
|
|
|
|
|
by LegionMammal978
92 days ago
|
|
|
Actually, we can improve this a bit further, by also adjusting the "π/2" constant in arcsin(x) = π/2 - P(x)*sqrt(1-x). We take coefficients [1.5707256467180715, -0.21298179775496026, 0.07727939759417458, -0.02132102849918157] for P(x), then take arcsin(x) = 1.570760986756484 - P(x)*sqrt(1-x). This reduces the max error by 6.97%, from 3.80e-5 to 3.53e-5. Adjusting the "1" upward in sqrt(1-x) does not seem to help at all. |
|
|