(12C Platinum) Normal Distribution

[Dieter]

(12-01-2018 11:16 PM)Albert Chan Wrote:  Can you post the exact coefficients too ?

According to my results the following set has a largest error of 1,0064 E-6:

a1 = 0,4335629429
a2 = 0,0841440309
a3 = -9,636444 E-5
b1 = 2,462958275
b2 = 1,132887108
b4 = 0,20590614476

(12-01-2018 11:16 PM)Albert Chan Wrote:  Edit: playing with the coefficients, I managed to lower max rel. errors to 1.12e-6

Fine, even with less digits. But it can still be tweaked a bit:

a1 = 0,433563
a2 = 0,084144052
a3 = -9,6366 E-5
b1 = 2,4629583
b2 = 1,1328872
b3 = 0,2059062

This way the error seems to stay a tiny bit below 1,12E-6. And for calculators every digit counts. For such applications I would now recommend this set of coefficients.

Maybe someone with access to Mathematica, Maple or similar software can post the exact largest errors.

Dieter