[
https://issues.apache.org/jira/browse/RNG-50?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16566020#comment-16566020
]
Alex D Herbert commented on RNG-50:
-----------------------------------
{quote}looks like it will silently recompute everything all the time if the
user happened to have mistyped its cache interval
{quote}
Well it won't compute more than what you would have to do if you didn't use a
cache (it's just the initialisation for the algorithm), so no gain but no real
loss either.
{quote}if one creates a cache for values between 0 and 255, then if some caller
requests 345 later, then keep that too in the cache, and so on
{quote}
An expandable cache requires an advanced data structure, e.g. a HashTable, in
order to support the full range of integers that could be requested. And the
speed of a hash table look-up within the cache could become expensive.
I think that the use of the cache is quite specialised so it is OK to leave it
the user to know what they are doing. It's all a bit mute anyway without
knowing it will make a noticeable difference on the speed. So I'll add a
version as described above to my test project JMH benchmark to see it is worth
pursuing.
*Note*
With regard to the possible range for the mean I've noticed that the
computation of {{log(n!)}} does assume the double valued mean can be cast to an
integer. This is also the same in the original algorithm in Commons Math 3.6.1
(which used {{CombinatoricsUtils.factorialLog(int)}}). Perhaps the algorithm
should also check the mean is not above {{Integer.MAX_VALUE}}. This would be a
sensible limit because: (a) the sample() method is limited to returning an
integer and with a mean of Integer.MAX_VALUE half the sample would overflow;
(b) a Poisson distribution is approximated by a Gaussian at high mean so could
be used a substitute anyway. Commons Math 3.6.1 even has
{{PoissonDistribution.normalApproximateProbability(int)}}. Actually support for
this is done by creating a {{NormalDistribution}} each time a
{{PoissonDistribution}} is created which is expensive if it's never used. I've
made a copy implementation that avoids this for my simulations. However the new
Commons RNG {{PoissonSampler}} will be much better and I'll be switching code
over to using that instead when the resolution for this ticket is released.
> PoissonSampler single use speed improvements
> --------------------------------------------
>
> Key: RNG-50
> URL: https://issues.apache.org/jira/browse/RNG-50
> Project: Commons RNG
> Issue Type: Improvement
> Affects Versions: 1.0
> Reporter: Alex D Herbert
> Priority: Minor
> Attachments: PoissonSamplerTest.java, jmh-result.csv
>
>
> The Sampler architecture of {{org.apache.commons.rng.sampling.distribution}}
> is nicely written for fast sampling of small dataset sizes. The constructors
> for the samplers do not check the input parameters are valid for the
> respective distributions (in contrast to the old
> {{org.apache.commons.math3.random.distribution}} classes). I assume this is a
> design choice for speed. Thus most of the samplers can be used within a loop
> to sample just one value with very little overhead.
> The {{PoissonSampler}} precomputes log factorial numbers upon construction if
> the mean is above 40. This is done using the {{InternalUtils.FactorialLog}}
> class. As of version 1.0 this internal class is currently only used in the
> {{PoissonSampler}}.
> The cache size is limited to 2*PIVOT (where PIVOT=40). But it creates and
> precomputes the cache every time a PoissonSampler is constructed if the mean
> is above the PIVOT value.
> Why not create this once in a static block for the PoissonSampler?
> {code:java}
> /** {@code log(n!)}. */
> private static final FactorialLog factorialLog;
>
> static
> {
> factorialLog = FactorialLog.create().withCache((int) (2 *
> PoissonSampler.PIVOT));
> }
> {code}
> This will make the construction cost of a new {{PoissonSampler}} negligible.
> If the table is computed dynamically as a static construction method then the
> overhead will be in the first use. Thus the following call will be much
> faster:
> {code:java}
> UniformRandomProvider rng = ...;
> int value = new PoissonSampler(rng, 50).sample();
> {code}
> I have tested this modification (see attached file) and the results are:
> {noformat}
> Mean 40 Single construction ( 7330792) vs Loop construction
> (24334724) (3.319522.2x faster)
> Mean 40 Single construction ( 7330792) vs Loop construction with static
> FactorialLog ( 7990656) (1.090013.2x faster)
> Mean 50 Single construction ( 6390303) vs Loop construction
> (19389026) (3.034132.2x faster)
> Mean 50 Single construction ( 6390303) vs Loop construction with static
> FactorialLog ( 6146556) (0.961857.2x faster)
> Mean 60 Single construction ( 6041165) vs Loop construction
> (21337678) (3.532047.2x faster)
> Mean 60 Single construction ( 6041165) vs Loop construction with static
> FactorialLog ( 5329129) (0.882136.2x faster)
> Mean 70 Single construction ( 6064003) vs Loop construction
> (23963516) (3.951765.2x faster)
> Mean 70 Single construction ( 6064003) vs Loop construction with static
> FactorialLog ( 5306081) (0.875013.2x faster)
> Mean 80 Single construction ( 6064772) vs Loop construction
> (26381365) (4.349935.2x faster)
> Mean 80 Single construction ( 6064772) vs Loop construction with static
> FactorialLog ( 6341274) (1.045591.2x faster)
> {noformat}
> Thus the speed improvements would be approximately 3-4 fold for single use
> Poisson sampling.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
