Hi Carl, Thanks for the info. I will take a look. I am currently using heavily the histogram class and its impact on the overall performance of the simulation can be large (with my own kernel density estimator) or low (with simple histogram structs like gsl_histogram). I will test your code for the post-processing, and it seems that you have a good criteria for solving the problem of the bandwidth. Thanks again!
Best regards / Cordialmente, -- William Oquendo Phd Candidate Simulation Of Physical Systems Group Universidad Nacional de Colombia Linux User # 321481 ********************* Este correo puede carecer de tildes o eñes ya que el teclado no contiene estos caracteres. Presento excusas por eso. ********************* On Wed, Aug 25, 2010 at 12:06 PM, Carl Boettiger <[email protected]> wrote: > Hi William, > > I wrote a basic kernel density estimator using the gsl library that mimics > the behavior of the default kernel density estimator in the R language: > http://gist.github.com/549857 which uses Silverman's rule of thumb to > estimate the bandwidth. > > -Carl > > > > On Wed, Aug 25, 2010 at 8:16 AM, William Oquendo <[email protected]>wrote: > >> Dear all, >> >> For histogram with uniform sizes, I just need to scale the original >> histogram h by (explicit expresion writen): >> gsl_histogram_bins(h)/(gsl_histogram_sum(h)*(gsl_histogram_max(h) - >> gsl_histogram_min(h))); >> >> For non-uniform bins, the scaling should be performed bin by bin, with the >> particular width. >> >> Any hope for fast kernel density estimators? >> >> Thanks in advance. >> >> Best regards / Cordialmente, >> >> -- >> William Oquendo >> Phd Candidate >> Simulation Of Physical Systems Group >> Universidad Nacional de Colombia >> Linux User # 321481 >> ********************* >> Este correo puede carecer de tildes o eñes ya que el teclado no contiene >> estos caracteres. Presento excusas por eso. >> >> ********************* >> >> >> >> On Wed, Aug 25, 2010 at 9:24 AM, William Oquendo <[email protected]> >> wrote: >> >> > Dear all, >> > >> > I am using the histogram gsl structs. I would like to know if is there a >> > way to compute the Probability Density Function (not the Cummulative >> > Probability Distribution function offered by the struct >> gsl_histogram_pdf). >> > Moreover, is there any option for fast kernel density estimators? >> > >> > Thanks in advance. >> > >> > Best regards / Cordialmente, >> > >> > -- >> > William Oquendo >> > Phd Candidate >> > Simulation Of Physical Systems Group >> > Universidad Nacional de Colombia >> > Linux User # 321481 >> > ********************* >> > Este correo puede carecer de tildes o eñes ya que el teclado no contiene >> > estos caracteres. Presento excusas por eso. >> > >> > ********************* >> > >> > >> _______________________________________________ >> Help-gsl mailing list >> [email protected] >> http://lists.gnu.org/mailman/listinfo/help-gsl >> > > > > -- > Carl Boettiger > Population Biology, UC Davis > http://two.ucdavis.edu/~cboettig <http://two.ucdavis.edu/%7Ecboettig> > _______________________________________________ Help-gsl mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-gsl
