(50g) Normal Distribution

[Dieter]

(12-04-2018 08:54 PM)John Keith Wrote:  RPL equivalent:

Code:

\<< DUP IP SQ 2. / NEG EXP OVER IP PICK3 FP * EXP / SWAP FP SQ EXP \v/ / .3989422804028 *
\>>

where the in-line constant is SQRT(1/(2*pi)).

I don't know much about RPL, but 1/sqrt(2pi) is 0,39894228040143...
However, on a 12-digit calculator dividing by 2,50662827463 might be slightly more accurate. ;-)

(12-04-2018 08:54 PM)John Keith Wrote:  I tried both the above code and my original prog with an input of -10 and the results differ by only one ULP. I am clearly not an expert on either statistics or the HP-67 so I may be missing something here but I don't see an advantage for the more involved program above.

An input of x=10 will not make much of a difference as x has merely two significant digits so that x²/2 is easily evaluated exactly. But try a larger value with more digits, say 61/3 = 20,3333333333. The exact PDF for this (12-digit rounded) input is 6,64644887122 E-91. What do you get with the direct method?

Dieter