Thanks Chris and Derek for the optimized method For unique I was going to use squaretotri() and then sort it The rle had went out of my mind...
I have a doubt, is it efficient to use the qsort method in large matrix ( say 5000 x 5000) On Sat, Oct 18, 2014 at 11:21 PM, Chris Marshall <devel.chm...@gmail.com> wrote: > Nice, Derek! Good catch (it is pretty cool to > see how powerful a few lines of PDL can be). > > --Chris > > On Fri, Oct 17, 2014 at 3:19 PM, Derek Lamb <de...@boulder.swri.edu> > wrote: > >> Ronak, >> >> If you are looking for the smallest 3 unique values, and those are "d", >> "e", and "f", do you want all of the locations where your piddle is "d" or >> "e" or "f" ? Or if "d" is the smallest, just 3 "d" locations? If the >> latter, then what Chris has will work as-is. If the former, then you need >> a little bit extra at the beginning: >> >> $matrix = >> pdl([0,1,2,3,4],[1,2,3,4,5],[2,3,4,5,6],[3,4,5,6,7],[4,5,6,7,8]); >> $k = 3; >> ($quant,$uniq_vals)=$matrix->flat->qsort->rle; #I am a big fan of >> ->qsort->rle. ->hist() might work too. >> $num_in_matrix = $quant(0:$k-1)->sum; >> #then Chris's stuff will work >> $smallest_k = zeroes(indx, $num_in_matrix); >> $matrix->flat->minimum_n_ind($smallest_k); #answer gets stuffed into >> $smallest_k >> $coords2D = pdl($matrix->one2nd($smallest_k)); >> $smallest_vals = $matrix->range($coords2D->transpose)->flat; >> >> cheers, >> Derek >> >> On Oct 17, 2014, at 10:29 AM, Chris Marshall <devel.chm...@gmail.com> >> wrote: >> >> Hi- >> >> First off, you can use the PDL on-line documentation to >> find relevant commands to try: >> >> pdldoc -a maximum >> >> or >> >> pdl> apropos maximum >> >> in the pdl2 or perldl shells includes the following items: >> >> maximum Project via maximum to N-1 dimensions >> maximum_ind Like maximum but returns the index rather than the >> value >> maximum_n_ind Returns the index of `m' maximum elements >> >> where maximum_n_ind is what you are looking for. Here is a >> sample pdl2 session showing the calculation. You can use >> the PDL shells and the online documentation to understand >> fully how it works: >> >> >> pdl> floor(random(10,10)*10) >>> pdl> p $m >>> [ >>> [4 7 5 0 6 7 8 4 6 7] >>> [0 7 0 2 5 3 0 4 6 6] >>> [5 4 3 6 9 5 8 4 4 1] >>> [9 6 7 7 4 1 4 2 0 8] >>> [5 1 3 3 7 7 1 2 8 5] >>> [2 0 8 7 3 1 5 7 7 6] >>> [5 9 6 2 7 2 6 7 0 2] >>> [6 3 1 1 8 2 1 9 9 9] >>> [6 7 9 3 5 2 2 5 9 3] >>> [4 5 0 4 7 9 3 3 1 6] >>> ] >>> >>> pdl> $top5 = zeros indx, 5; # size of $top5 determines n >>> pdl> $m->flat->maximum_n_ind($top5) # get top 5 element indexes >>> pdl> p pdl($m->one2nd($top5)) # convert linear to ND index >>> [ >>> [4 0 1 7 8] >>> [2 3 6 7 7] >>> ] >>> >>> >>> >> Hope this helps, >> Chris >> >> >> On Fri, Oct 17, 2014 at 11:54 AM, Ronak Agrawal <ronagra...@gmail.com> >> wrote: >> >>> First I would Thank You for your constant help...it has helped me a lot >>> in improving my skills >>> ------- >>> >>> I have generated a Hankel Matrix by the following operation >>> >>> // $a is a svd and therby this operation will always form Hankel Matrix >>> $matrix = $a x transpose($a); >>> >>> [image: [a b c d e; b c d e f; c d e f g; d e f g h; e f g h i].] >>> >>> I need to find the top K minimum elements ( 2D) with their 2 Dimesional >>> indices .. >>> Can you suggest me some good approach for this >>> >>> I though to do following but it is not optimized >>> Converting the Lower triangular Matrix as Bad Values or 0 >>> Then Finding the mimimum row_wise and column_wise using minimum function >>> >>> Thanks >>> >>> >>> >>> >>> >>> _______________________________________________ >>> Perldl mailing list >>> Perldl@jach.hawaii.edu >>> http://mailman.jach.hawaii.edu/mailman/listinfo/perldl >>> >>> >> _______________________________________________ >> Perldl mailing list >> Perldl@jach.hawaii.edu >> http://mailman.jach.hawaii.edu/mailman/listinfo/perldl >> >> >> >> _______________________________________________ >> Perldl mailing list >> Perldl@jach.hawaii.edu >> http://mailman.jach.hawaii.edu/mailman/listinfo/perldl >> >> >
_______________________________________________ Perldl mailing list Perldl@jach.hawaii.edu http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
