BrowserUk:
OK, I've added my routine to your benchmark program, wrapped in readmore tags below.
I've had to add and edit a bit of code outside the actual compared routines. The OP mentioned that the ratio of 1 to 0 entries is 1::500, which is a fact I used to come up with my approach. So the first change is the ability to set the probability of 1 bits.
Since the 1 bits are sparse, rather than making an explicit vector, I use a list of the positions of the 1 bits. In order to come up with the same results, I converted the Y vectors list format.
The biggest change outside of the comparison routine is the setup routine that transforms the xArray. The concept was to do something like this: Build an artificial set of vectors each with 1 bit--one vector per bit position. Then we compare each of these artificial vectors against the xArray set, resulting in a list of x vectors for each bit position. Then in our comparison, we aggregate the selected bins. Thus, if y has five bits in it, we add in the five partial products from the eigenset vectors. So the process of building the lists is amortized over the run of comparisons.
Having said all that, here it is. As mentioned previously, I came up with my approach when I saw that the distribution of 1s was very sparse. As the density of 1s increases, the routine gets progressively slower.
$ perl 1067357_mcm.pl -I=1 -N=100 -W=4000 -P=.05
<<< snipped >>>
Rate array strings bits robo
array 2.33e-02/s -- -98% -99% -100%
strings 1.45/s 6122% -- -25% -72%
bits 1.92/s 8156% 33% -- -63%
robo 5.26/s 22495% 263% 174% --
$ perl 1067357_mcm.pl -I=1 -N=100 -W=4000 -P=.5
<<< snipped >>>
s/iter array robo strings bits
array 60.4 -- -87% -99% -99%
robo 7.75 680% -- -91% -93%
strings 0.690 8659% 1023% -- -23%
bits 0.530 11304% 1362% 30% --
#! perl -slw
use strict;
use Benchmark qw[ cmpthese ];
use Data::Dump qw[ pp ]; $Data::Dump::WIDTH = 500;
our $I //= -1;
our $N //= 1000;
our $W //= 15_000;
# OP mentions that 1 bits are sparse (1 :: 500) It doesn't make a dif
+ference for straight
# bit comparison techniques, but it *does* for mine.
our $P //= 0.002;
our @xArrays = map[ map { rand()<$P ? 1 : 0 } 1 .. $W ], 1 .. $N;
our @yArrays = map[ map { rand()<$P ? 1 : 0 } 1 .. $W ], 1 .. $N;
our @xStrings = map{ join '', @$_ } @xArrays;
our @yStrings = map{ join '', @$_ } @yArrays;
our @xBits = map{ pack 'b*', $_ } @xStrings;
our @yBits = map{ pack 'b*', $_ } @yStrings;
# Because the bit vectors are sparse, I'm storing them as lists of bit
+ positions, rather
# than bit vectors, so I'm converting Y
my $start = time;
our @yArrays2;
for my $y (0 .. $#yArrays) {
for my $bitcol (0 .. $W-1) {
push @{$yArrays2[$y]}, $bitcol if $yArrays[$y][$bitcol];
}
}
my $cur = time - $start;
print "Converting yArrays format took $cur sec.\n";
# I bin the X arrays after which I never use them again, so I'll bin t
+hem directly from the
# originally generated vectors.
$start = time;
our @bitcols;
for my $x (0 .. $#xArrays) {
for my $bitcol (0 .. 14_999) {
push @{$bitcols[$bitcol]}, $x if $xArrays[$x][$bitcol];
}
}
$cur = time - $start;
print "Binning xArrays took $cur sec.\n";
cmpthese $I, {
array => q[
my %top10s;
for my $x ( 0 .. $#xArrays ) {
for my $y ( 0 .. $#yArrays ) {
my $count = 0;
$xArrays[$x][$_] == 1 && $yArrays[$y][$_] == 1 and ++$
+count for 0 .. $#{ $xArrays[ 0 ] };
$top10s{"$x:$y"} = $count;
my $discard = ( sort{ $top10s{$a} <=> $top10s{$b} ||
+$a cmp $b} keys %top10s )[ 0 ];
keys( %top10s ) > 10 and delete $top10s{$discard};
}
}
$I == 1 and pp ' arrays: ', %top10s;
],
strings => q[
my %top10s;
for my $x ( 0 .. $#xStrings ) {
for my $y ( 0 .. $#yStrings ) {
my $count = ( $xStrings[$x] & $yStrings[$y] ) =~ tr[1]
+[];
$top10s{"$x:$y"} = $count;
my $discard = ( sort{ $top10s{$a} <=> $top10s{$b} ||
+$a cmp $b} keys %top10s )[ 0 ];
keys( %top10s ) > 10 and delete $top10s{$discard};
}
}
$I == 1 and pp 'strings: ', %top10s;
],
bits => q[
my %top10s;
for my $x ( 0 .. $#xBits ) {
for my $y ( 0 .. $#yBits ) {
my $count = unpack '%32b*', ( $xBits[$x] & $yBits[$y]
+);
$top10s{"$x:$y"} = $count;
my $discard = ( sort{ $top10s{$a} <=> $top10s{$b} ||
+$a cmp $b} keys %top10s )[ 0 ];
keys( %top10s ) > 10 and delete $top10s{$discard};
}
}
$I == 1 and pp ' bits: ', %top10s;
],
robo => sub {
my %top10s;
my $cur_min=0;
for my $y ( 0 .. $#yArrays2 ) {
my %x;
for my $bitcol (@{$yArrays2[$y]}) {
$x{$_}++ for @{$bitcols[$bitcol]};
}
for my $k (keys %x) {
next unless $x{$k} >= $cur_min;
$top10s{"$k:$y"} = $x{$k};
my $discard = ( sort{ $top10s{$a} <=> $top10s{$b} || $
+a cmp $b } keys %top10s )[ 0 ];
if (keys %top10s > 10) {
$cur_min = $top10s{$discard};
delete $top10s{$discard};
}
}
}
$I == 1 and pp ' bits: ', %top10s;
},
};
...roboticus
When your only tool is a hammer, all problems look like your thumb.
