

Something popped into my head recently, and now I am trying to write an equation
for the cumulative distribution function of a probability density function using
sum ().
I'm using a normal distribution, but when I try to graph the summation of the
values over the range of 0 through 5, I get unexpected results.
I tried using N, then since i is related to N, I tried seeing what that result
looked like, and then I averaged them... It makes no sense that a graph of a
running tally starts to decrease, instead of being nondecreasing and
rightcontinuous like it should be.
#declare NDist = function (N, mu, sigma)
{1/(sigma*sqrt(tau))*exp(0.5*pow((Nmu)/sigma,2))}
#declare NDist_cdf = function (N, mu, sigma) {sum (i, 0, N, NDist (N, mu,
sigma))}
#declare NDist_cdf2 = function (N, mu, sigma) {sum (i, 0, N, NDist (i, mu,
sigma))}
#declare NDist_cdf3 = function (N, mu, sigma) {sum (i, 0, N, NDist ((i+N)/2,
mu, sigma))}
Has anyone ever tried to use sum () to graph a running tally of a function, or
have any idea what I'm doing wrong?
Have I broken POVRay _*AGAIN*_???!
Post a reply to this message
Attachments:
Download 'importancesamplingtest1.png' (67 KB)
Preview of image 'importancesamplingtest1.png'




"Bald Eagle" <cre### [at] netscapenet> wrote:
> Have I broken POVRay _*AGAIN*_???!
Nope, just me glossing over the fact that sum doesn't have a step value, because
its default is one.
Tired brain does not listen to very little inner voice pointing out exactly
that.
"The iteration variable is incremented by one for each step, but its initial and
final value may be any value"
https://wiki.povray.org/content/Reference:Function
So, what I needed to do was:
#declare NDist_cdf2 = function (N, mu, sigma) {sum (i, 0, N*100, NDist (i/100,
mu, sigma))}
So at present, a can take a pdf, get the cdf, and even transpose it to its
inverse function. :)
Now to figure out how to properly weigh the sample points and I think I should
have a semiworkable / partial importance sampling algorithm.
Like ... 7 years after clipka wanted one.
I'm glad I wrote that equation graphing scene way back when, but I also think I
need to debug my axislabeling code.
Post a reply to this message
Attachments:
Download 'importancesamplingtest1.png' (64 KB)
Preview of image 'importancesamplingtest1.png'

