Subscribe Bookmark RSS Feed

Is there a way to calculate erfc(x) in JMP?

surfacethought

Community Trekker

Joined:

Aug 8, 2012

Title is pretty self explanatory. I am trying to do non-linear least squares with a transcendental function that includes the error function (complement). However when I try to define a predict column using the function I was sad to find that there is no erf(x) in the transcendental group of functions. Without being able to calculate the error function there is no way that I can perform this nonlinear least squares in JMP!  Does anyone know if it is possible to do in JMP? I can use an approximation using elementary functions, however I would like to avoid adding any error if at all necessary. Thanks!

--

SurfaceThought

1 ACCEPTED SOLUTION

Accepted Solutions
Solution

erf(X) = 2 * Normal Distribution( :X * Root( 2, 2 ) ) - 1

erfc(X) = 1 - erf(X) = 2 * Normal Distribution(-:X * Root( 2, 2 ) )

where "Normal Distribution()" is the cumulative normal distribution function available in JMP for use in either scripts or column formulas.

4 REPLIES
Solution

erf(X) = 2 * Normal Distribution( :X * Root( 2, 2 ) ) - 1

erfc(X) = 1 - erf(X) = 2 * Normal Distribution(-:X * Root( 2, 2 ) )

where "Normal Distribution()" is the cumulative normal distribution function available in JMP for use in either scripts or column formulas.

surfacethought

Community Trekker

Joined:

Aug 8, 2012

Alright I think this is very much on the right track, thank you so much. I just have one question, whats the purpose of the colon inside the Normal Distribution function?

Edit: Yeah that's definitely what I needed. You rock! thanks for the help!

pmroz

Super User

Joined:

Jun 23, 2011

The colon indicates that X is a column name.  If you are defining a column formula in terms of other columns, this is how you would reference it.

surfacethought

Community Trekker

Joined:

Aug 8, 2012

Oh okay great that's easy enough .

Thank You!