> From: Ian Clark
> 
> Interestingly the alternative implementation (after Ewart Shaw) of the
> verb: cnd which is introduced later will work correctly with the
> original definition of BlackScholes, not needing the cnd"0 fix.
> Because for Shaw's implementation:
> 
> cnd (d1,d2) <--> (cnd d1),(cnd d2)
> 
> This discrepancy is something the owners of statdist.ijs in J602 might
> like to ponder.

The line in the article that loads the statdist.ijs script looks like it is 
missing a tilde. I suspect that it won't work in its current form. Should 
probably be:
  load '~system\packages\stats\statdist.ijs'

However the scripts in ~system/packages/stats have recently been recreated as 
the stats/base addon. http://www.jsoftware.com/jwiki/Addons/stats/base
As I understand it the plan is for many of the packages currently distributed 
in ~system/packages to be removed from there and made available as addons 
instead.  For that reason I think it would be safer to reference the script in 
its new incarnation as:

  load 'stats/base/distribution'

or if you want to make it more obvious that it is an addon & must be installed:

  load '~addons/stats/base/distribution.ijs'

That script also includes the verbs erf and normalcdf which are basically the 
erf & cnd described in the At Play with J article.
http://www.jsoftware.com/jwiki/Addons/stats/base/distribution

For a more well-behaved cumulative normal distribution verb, I suggest using 
the verb pnorm (or pnorm01_pdistrib_) from the stats/distribs addon.
http://www.jsoftware.com/jwiki/Addons/stats/distribs

pnorm uses Ewart Shaw's algorithm for values between _7 & 7, but a Chebychev 
expansion for values outside that range where normalcdf becomes unstable as 
explained on the addon page.


----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to