Craig,
Just discovered PDL::NiceSlice which explains that slicing *$vec->(...);*
syntax I asked about.
Will
On Sun, Apr 8, 2018 at 12:33 PM, William Schmidt <
[email protected]> wrote:
> Craig,
>
> Thank you, it works but first I had to pick those method calls apart to
> see what each is doing and rtfm about *sever* in PDL::Core.
>
> So, I am left with a set of questions best depicted with some examples. As
> I understand the ::Core pod, that random call and floor are creating a
> virtual piddle that functions to slice or index into $vec. And since the
> virtual piddle is longer than $vec, it loops until exactly the number of
> items in the virtual piddle are pushed from $vec into $rand. So, what is
> the difference in the following invocations?
>
> $rand = $vec->(floor(random(1e6) * 8))->sever; # yours
> $rand = $vec->(floor(random(1e6) * 8));
> $rand = $vec->index(floor(random(1e6) * 8))->sever;
> $rand = $vec->index(floor(random(1e6) * 8));
>
> When I inspect $rand, all are different, since random() probably generates
> a fresh seed, but they all look to be examples of my weighted probability
> distribution.
>
> To be precise, is there an implied or default method call after *$rand =
> $vec->...* in your example, and what is lost when *sever* is not invoked?
>
> I think I can visualize how to generalize this algorithm but it is much
> less intuitive than the way one can do it in R and building $vec can be
> problematic, or am I missing something? A complex set of weights using
> reals, for example: [0.18, 0.31, 0.05, 0.07,...] summing to 1 would require
> a lot of fiddling for it to be used as a slice.
>
> Regards,
> Will
> [email protected]
>
>
>
>
> On Fri, Apr 6, 2018 at 7:58 PM, Craig DeForest <[email protected]>
> wrote:
>
>> Welcome, William!
>>
>> You are probably looking for “random()”, which has the same syntax as
>> “zeroes()” but returns a vector of pseudorandom values on [0,1).
>> To make a vector of a million of those, use “$a = random(1e6)”.
>>
>> To make random integers based on a histogram that you already have
>> in-hand, a simple way is:
>>
>> $vec = pdl(0,1,1,1,2,2,2,3); # note 8 values
>> $rand = $vec->(floor(random(1e6) * 8))->sever;
>>
>> The “random(1e6”) makes a million elements on [0,1). Multiply by 8 and
>> take the floor to get integers on [0,7]. The outermost operation indexes
>> $vec with the corresponding random value. Since there are three 1’s, 1 is
>> three times as likely in the output.
>>
>> Does that work?
>>
>> Best,
>> Craig
>>
>>
>>
>> On Apr 6, 2018, at 5:16 PM, William Schmidt <[email protected]>
>> wrote:
>>
>> Hello Piddlers,
>>
>> I am moving from R to PDL, with tons of experience with Perl, lots with R
>> but zero with PDL,
>> so this is a pretty basic question. I can see from the PDL Book that PDL
>> is very
>> sophisticated, with much more functionality than I will ever use, but I
>> want
>> to master basic PDL to leverage my Perl. My focus is on probability in
>> two dimensions so
>> I will be working mostly with 1-dimensional vectors. Here is an example
>> from R that
>> I would like to learn how to do in PDL. It is a small example but once I
>> master
>> the construction of this data I will extend it to much larger vectors.
>>
>> Suppose I have random variable X whose values and probabilities are as
>> follows:
>>
>> *x* *p(x)*
>> 0 1/8
>> 1 3/8
>> 2 3/8
>> 3 1/8
>>
>> To get a sample of 50 random values drawn from this population with
>> replacement in R I
>> would say:
>>
>> x <- seq.int(0,3) # Concatenate a sequence of ints from 0 to 3.
>> x # print x.
>> [1] 0 1 2 3
>>
>> weights <- c(1/8, 3/8, 3/8, 1/8) # Another form of concatenation.
>> weights
>> [1] 0.125 0.375 0.375 0.125
>>
>> s <- sample(x, 50, replace=TRUE, prob=weights)
>> s
>> [1] 0 1 1 3 2 2 2 3 2 0 0 1 3 1 1 3 0 2 1 2 2 1 3 1 2 2 0 2 2 2 3 2
>> [33] 1 1 3 1 2 2 1 1 0 1 3 2 2 1 3 0 1 1
>>
>> I can now manipulate *s*, calculate its statistical properties and graph
>> its probability distribution. Fifty integer values is not very interesting
>> but the problems I am studying have thousands of values and very different
>> weights. How do I do this in PDL? I have PDL::Stats::Basic and
>> PDL::Stats::Distr installed along with PGPLOT but it's generating this
>> basic data that has me stumped.
>>
>> Thanks and regards,
>>
>> Will Schmidt
>> [email protected]
>>
>>
>>
>> ------------------------------------------------------------
>> ------------------
>> Check out the vibrant tech community on one of the world's most
>> engaging tech sites, Slashdot.org! http://sdm.link/slashdot______
>> _________________________________________
>> pdl-general mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/pdl-general
>>
>>
>>
>
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
pdl-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/pdl-general
