RE: binomial normal cdf

[Ted Harding]

On 27-Jul-98 Dimitrie O. Paun wrote:
> I am looking for a Octave/Matlab function that can return the binomial
> normal cdf. Does anyone has such a beast?

For the binomial probability function try

function P=binompdf(n,p)
# For integer n, real p (0 <= p <= 1),
# returns the entire binomial distribution on (0,1,2,...,n)
# as a column vector.
if (( p<0 ) | ( p>1) )
  error ( "p is out of range" );
endif
if ( p == 0 )
  P = zeros(n+1,1); P(1) = 1;
elseif ( p == 1 )
  P = zeros(n+1,1); P(n+1) = 1;
else
  r = (0:n)';
  P = exp(lgamma(n+1)-lgamma(r+1)-lgamma(n-r+1)+r*log(p)+(n-r)*log(1-p));
  P = P/sum(P);
endif  
endfunction

and for the binomial CDF try

function P=binomCDF(n,p)
# For integer n, real p (0 <= p <= 1),
# returns the entire binomial cumulative distribution on (0,1,2,...,n)
# as a column vector.
  P = cumsum(binompdf(n,p));
endfunction

For the Normal CDF try

function p = normCDF(x);
# function p = normCDF(x) = int from (-inf) to x of normpdf(t) dt
  s=sqrt(2);
  p=0.5*(1 + erf(x/s));
endfunction

I don't know what would be meant by "binomial normal CDF" unless one of the
above.

Best wishes,
Ted.

--------------------------------------------------------------------
E-Mail: (Ted Harding) <address@hidden>
Date: 28-Jul-98                                       Time: 00:31:03
--------------------------------------------------------------------