[SciPy-Dev] Simple arithmetic with scipy.stats.distributions

[Padarn Wilson]

Apologies all, I accidentally had digest mode set, so I'm replying manually.

>> I was wondering if anyone else would interested in having some simple
> >> arithmetic functionality for 'distributions'? So that for example you
> could
> >> add distributions, then sample from them, get the pdf/cdf etc.
> How do you want to do that or what do you have in mind here?
> some analytical calculations or Monte Carlo propagation?
> https://pypi.python.org/pypi/soerp looks interesting (based on
> browsing it's documentation)


Well, at the moment I'm just using the PDF (and random samples, but that is
easy), and doing this approximately by taking a discrete convolution. For
something outside my specific use case though, something like Monte Carlo
progpagation is what I had in mind (I wanted to be able to use approximate
densities eventually too).

However soerp does look good. Thanks for pointing that out. Perhaps not so
useful for the approximate cases, but good otherwise.

>> I've hacked something together for my own purposes, but it seems like it
> be
> >> nice general functionality?
> >
> > Have you hosted your code anywhere?


No - it is tied into some private code at the moment. I'll clean it up and
host it on Github, but it isn't anything special.

Thanks,
Padarn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/scipy-dev/attachments/20131125/2a0e3302/attachment.html>