Ryan Thompson › Perl Weekly Review #051

Tuesday, Mar 24, 2020| Tags: perl

Continues from previous week.

Welcome to the Perl review for Week 051 of the Weekly Challenge! For a quick overview, go through the original tasks and recap of the weekly challenge.

Getting in Touch

Email › Email me (Ryan) with any feedback about this review.

GitHub › Submit a pull request for any issues you may find with this page.

Twitter › Join the discussion on Twitter!

We’d greatly appreciate any feedback you’d like to give.

Table of Contents

Task 1

[ Adam Russell | Alicia Bielsa | Andrezgz | Cheok-Yin Fung | Colin Crain | Cristina Heredia | Dave Cross | Dave Jacoby | Duncan C. White | E. Choroba | Jaldhar H. Vyas | Javier Luque | Mohammad S Anwar | Phillip Harris | Roger Bell West | Ruben Westerberg | Ryan Thompson | Saif Ahmed | Ulrich Rieke | User Person | Walt Mankowski | Wanderdoc | Yet Ebreo ]

Task 2

[ Adam Russell | Alicia Bielsa | Andrezgz | Cheok-Yin Fung | Colin Crain | Cristina Heredia | Dave Cross | Dave Jacoby | Duncan C. White | E. Choroba | Jaldhar H. Vyas | Javier Luque | Mohammad S Anwar | Phillip Harris | Roger Bell West | Ruben Westerberg | Ryan Thompson | Saif Ahmed | Ulrich Rieke | User Person | Walt Mankowski | Wanderdoc | Yet Ebreo ]

Blogs



Task #1 - 3 Sum

Here is the original task description:

Given an array @L of integers, write a script to find all unique triplets such that a + b + c is same as the given target T. Also make sure a <= b <= c.

Example:

@L = (-25, -10, -7, -3, 2, 4, 8, 10);

One such triplet for target 0 i.e. -10 + 2 + 8 = 0.


Solution types

O(n³) brute force

By simply using a 3-nested loop, one for each of a, b and c, you get a straightforward O(n³) solution.

There are some optimized variations of this approach that prune early or choose better upper and lower bounds. Fundamentally these are still n-cubed complexity, but in practice they can reduce the number of comparisons by as much as ~83% if done carefully.

O(n²) with a “defined” hash

You can reduce the degree of the function by storing all values of @L in a hash (or array, but a hash is more space-efficient while still giving constant-time lookups). You then have a and b loops, but where you would have a c loop, you can instead simply check whether Target - a - b exists in your hash.

In either case, you may as well sort if it helps you, as even an n² term will dominate an n log n sort.

O(n²) solution from Wikipedia

The Wikipedia link given in the task description has pseudocode for a quadratic time or O(n²) algorithm. Some opted to implement that in Perl (or perhaps arrived at it independently, I have no way to know). It works by having a nested loop, with the outer loop iterating through the entire list. The inner while loop walks start and end indices toward each other until they meet. The decision to increment start, decrement end, or increment/decrement both depends on what the current a + b + c sums to:

  • a + b + c == 0 : Increment/decrement both, assuming numbers are distinct
  • a + b + c > 0 : Decrement end
  • a + b + c <= 0 : Increment start

Adam Russell

Adam is back with another Prolog and Perl solution! The Perl setup code is as follows, using AI::Prolog:

use AI::Prolog;
use constant TARGET => 0;
use constant NUMBERS => "-25, -10, -7, -3, 2, 4, 8, 10";
my $prolog = do{
    local $/;
    <DATA>;
};
my $numbers = NUMBERS;
$prolog =~ s/NUMBER_LIST/$numbers/;
$prolog = new AI::Prolog($prolog);
$prolog->query("unique_triplets(X, Y, Z, " . TARGET . ").");
my $result = $prolog->results;
my($x, $y, $z) = @{$result}[1 .. @{$result} - 1];
print "X: $x\nY: $y\nZ: $z\n";

The Prolog code is in the __DATA__ section:

member(X,[X|T]).
member(X,[H|T]) :- member(X,T).
numbers([NUMBER_LIST]).
unique_triplets(X, Y, Z, T) :-
    numbers(L),
    member(X, L),
    member(Y, L),
    member(Z, L),
    X <= Y,
    Y <= Z,
    X <= Z,
    T is X + Y + Z.

I’m glad to see Adam submitting solutions again after a brief hiatus.

BlogMore Perl & Prolog

Alicia Bielsa

Alicia Bielsa’s solution is a straightforward brute force implementation:

my $target = 0;
my  @L = sort(-25,  -7, -3, 2, 4, -10,8, 10);
foreach my $a (0..$#L){
    foreach my $b ($a+1..$#L){
        foreach my $c ($b+1..$#L){
            if ( ($L[$a] + $L[$b] + $L[$c]) == $target){
                print $L[$a]."  + ".$L[$b]." + ".$L[$c]." = $target.\n";
            }
        }
    }
}

Andrezgz

Andrezgz’s solution is also brute force:

my $integers = join ',', @ARGV;
my @L = sort { $a <=> $b } grep {/-?\d+/} split /,/, $integers;
die "At least 3 integers are needed" if @L < 3;
my $triplets;
for my $i (0 .. $#L - 2) {
    for my $j ($i+1 .. $#L - 1) {
        for my $k ($j+1 .. $#L) {
            next unless $L[$i] + $L[$j] + $L[$k] == TARGET;
            my $key = join '#', $L[$i], $L[$j], $L[$k];
            $triplets->{$key}++;
        }
    }
}
print 'Triplets for target '.TARGET."\n";
printf "(%s,%s,%s)\n", split /#/ foreach keys %$triplets;

Cheok-Yin Fung

Cheok-Yin Fung’s solution uses Math::Combinatorics to simplify obtaining the list of combinations. The loop body is a simple $target == sum @tsum check:

use Math::Combinatorics;
use List::Util qw{sum};
#Usage: ch-1.pl $target $L[0] $L[1] ... $L[$#L];
my $target = shift @ARGV;
my @L = @ARGV;
my $tripets = Math::Combinatorics->new( count => 3 , data => [@L] );
while (my @tsum = $tripets->next_combination) {
    print join(" ", sort {$a<=>$b} @tsum)."\n" if $target == sum @tsum;
}

Colin Crain

Colin Crain’s solution reads integers from a large __DATA__ block (not shown) to populate @L:

my @L;
while (my $line = <DATA>) {
    chomp $line;
    push @L, split /, /, $line;
};

What follows is Colin’s O(n²) solution, which maintains $idx, $high, and $low indexes into the sorted @list:

## nominally 0, but this can be changed easily here
my $TARGET  = 0;
my @list    = sort {$a <=> $b} @L;
my $length  = scalar @L;
my @output;
for my $idx ( 0..$length - 2) {
    ## if a, the smallest value, is greater than the target value, no more
    ## solutions are possible and we are done
    last if $list[$idx] > $TARGET;
    ## if a is a duplicate of the previous search, short-circuit to the next value
    next if ($idx > 0 && $list[$idx] == $list[$idx-1]);
    my $a     = $list[$idx];
    my $low   = $idx + 1;
    my $high  = $length - 1;
    while ( $low < $high ) {
        ## if b is a duplicate of the previous search, increment the index and short-circuit
        if ($low > $idx+1  &&  $list[$low] == $list[$low-1]){
            $low++;
            next;
        }
        ## if c is a duplicate of the previous search, decrement the index and short-circuit
        if ($high < $length - 1  &&  $list[$high] == $list[$high+1]) {
            $high--;
            next;
        }
        my $b = $list[$low];
        my $c = $list[$high];
        ## on success note to output, increment the start index and decrement the end
        ## so as not to duplicate searches
        if ($a + $b + $c == $TARGET) {
            push @output, [$a, $b, $c];
            $low++;
            $high--;
        }
        ## if we are already above target shift the end element down and start again
        elsif ($a + $b + $c > $TARGET) {
            $high--;
        }
        ## else try the next internal candidate
        else {
            $low++;
        }
    }
}
say join ', ', $_->@* for @output;

For reference, an unoptimized O(n³) solution on 1000 integers would take 1000³ = 1,000,000,000 iterations, but Colin’s manages it in just 364,890, completing in about 0.3 seconds.

Cristina Heredia

Cristina Heredia’s solution is an O(n³) loop:

sub checkNumber {
    for (my $j=0; $j<$digits; $j++) {
        for (my $k=0; $k<$digits; $k++) {
            if ($j != $k){
                for (my $l=0; $l<$digits; $l++) {
                    if ($l != $k and $j != $l){
                        $total = $aNumber[$j] +$aNumber[$k] +$aNumber[$l];
                        if ($total == $target and $aNumber[$j] < $aNumber[$k] and $aNumber[$k] < $aNumber[$l]) {
                            $message = $message."Result that match the criteria are the numbers: $aNumber[$j], $aNumber[$k], $aNumber[$l]\n";
                            print "$message\n";
                        }
                    }
                }
            }
        }
    }
    if ($message eq '') {
        print "There are no results\n";
    }
}

Dave Cross

Dave Cross’s solution also manages $start and $end indices, similar to the Wikipedia solution:

my $target  = 0;
my @numbers = sort { $a <=> $b } grep { /^-?\d+$/ } @ARGV;
die "Usage: $0 [list of integers]\n" unless @numbers;
for my $i ( 0 .. $#numbers - 1 ) {
    my $x     = $numbers[$i];
    my $start = $i + 1;
    my $end   = $#numbers;
    while ( $start < $end ) {
        my $y = $numbers[$start];
        my $z = $numbers[$end];
        if ( $x + $y + $z == $target ) {
            say "($x, $y, $z)";
            $start++;
            $end--;
        }
        else {
            if ( $x + $y + $z > $target ) {
                $end--;
            }
            else {
                $start++;
            }
        }
    }
}

It’s quite concise for a solution of this type, and Dave’s solution is similarly performant, requiring just 498,501 iterations to handle a 1000-integer input list.

Dave Jacoby

Dave Jacoby’s solution is brute force, optimized with dynamic lower and upper bounds on each loop:

my @L = ( -25, -10, -7, -3, 2, 4, 8, 10 );
my $target = 0;

for my $i ( 0 .. scalar @L - 1 ) {
    for my $j ( $i .. scalar @L - 1 ) {
        for my $k ( $j .. scalar @L - 1 ) {
            my $l = $L[$i] + $L[$j] + $L[$k];
            next unless $l == $target;
            say qq{  $L[$i] + $L[$j] + $L[$k] = $l };
        }
    }
}

The intelligent loop bounds do help quite a lot, reducing the 1000-integer test case to 167,167,000 comparisons: 16.7% of O(n³). The cubed term still dominates, though, of course.

Duncan C. White

Duncan C. White’s solution uses the hash optimization approach:

die "Usage: 3-sum T LIST\n" if @ARGV < 4;
my $t = shift;
my @l = @ARGV;
@l = sort @l;
my $n   = @l;
my %v2p = map { $l[$_] => $_ } 0 .. $n - 1;
foreach my $i ( 0 .. $n - 3 ) {
    my $a = $l[$i];
    foreach my $j ( $i + 1 .. $n - 2 ) {
        my $b    = $l[$j];
        my $sum2 = $a + $b;
        my $left = $t - $sum2;
        if ( defined $v2p{$left} ) {
            say "found a=$a, b=$b, c=$left (target=$t)";
        }
    }
}

This uses just under half of the iterations its O(n²) complexity would suggest, thanks to the careful bounds on the inner loop.

E. Choroba

E. Choroba’s solution is a lightly-optimized O(n³) brute force implementation:

#!/usr/bin/perl
use warnings;
use strict;
use feature qw{ say };
my @L = (-25, -10, -7, -3, 2, 4, 8, 10);
my $target = 0;
for my $i (0 .. $#L - 2) {
    for my $j ($i + 1 .. $#L - 1) {
        for my $k ( $j + 1 .. $#L) {
            say join ' + ', sort { $a <=> $b } @L[$i, $j, $k]
                if $target == $L[$i] + $L[$j] + $L[$k];
        }
    }
}

Choroba performed bounds optimizations, reducing the number of comparisons by ~83% compared to a naïve n³ solution.

Blog3 Sum and Colourful Numbers

Jaldhar H. Vyas

Jaldhar H. Vyas’s solution starts with a combinations sub that recursively generates combinations of numbers from the input array ref, of a specified length:

sub combinations {
    my @list = @{$_[0]};
    my $length = $_[1];
    if ($length <= 1) {
        return map [$_], @list;
    }
    my @combos;
    for (my $i = 0; $i + $length <= scalar @list; $i++) {
        my $val  = $list[$i];
        my @rest = @list[$i + 1 .. $#list];
        for my $c (combinations(\@rest, $length - 1)) {
            push @combos, [$val, @{$c}] ;
        }
    }
    return @combos;
}

The main logic then simply iterates over the result from combinations and outputs the tuples whose sum matches the target.

my $T = shift;
my @L = @ARGV;
for my $combo (combinations(\@L, 3)) {
    my @triplet = sort{ $a <=> $b } @{$combo};
    my $total = 0;
    for my $elem (@triplet) {
        $total += $elem;
    }
    if ($total == $T) {
        say join q{ }, @triplet;
    }
}

Javier Luque

Javier Luque’s solution also resembles the Wikipedia solution. The inner loop is a while loop that walks $start and $end indices toward each other until they meet:

my @L = (-25, -10, -7, -3, 2, 4, 8, 10);
@L = sort { $a <=> $b } @L;
for (my $i = 0; $i < scalar(@L) - 2; $i++) {
    my $start = $i + 1;
    my $end   = scalar(@L) - 1;
    while ($start < $end) {
        if ($L[$i] + $L[$start] + $L[$end] == 0) {
            say "$L[$i] + $L[$start] + $L[$end] = 0";
            $start = $start + 1;
            $end = $end -1
        } elsif ($L[$i] + $L[$start] + $L[$end] > 0) {
            $end = $end - 1;
        } else {
            $start = $start + 1;
        }
    }
}

This is another very capable O(n²) approach.

Javier mentions on his blog that he did copy the algorithm from Wikipedia, which I appreciate and respect.

Blog051 – Perl Weekly Challenge

Mohammad S Anwar

Mohammad S Anwar’s solution is pure, CPU-warming n³:

my @L = (-25, -10, -7, -3, 2, 4, 8, 10);
my $S = $#L;
my $T = 0;
foreach my $i (0 .. $S) {
    foreach my $j (0 .. $S) {
        next if ($i == $j);
        foreach my $k (0 .. $S) {
            next if (($k == $i) || ($k == $j));
            print sprintf("[%d, %d, %d]\n", $L[$i], $L[$j], $L[$k])
                if ($L[$i] + $L[$j] + $L[$k] == $T);
        }
    }
}

Mohammad knows his solution wasn’t very efficient, but he enjoyed the aesthetic of this code, and I respect that. The joy of coding has always been at the heart of my own career.

BlogBLOG: The Weekly Challenge #051

Phillip Harris

Phillip Harris’s solution implements the wiki O(n²) solution:

my @L = ( -25, -10, -7, -3, 2, 4, 8, 10 );
my $target = 0;
@L = sort { $a <=> $b } @L;
print "INPUT:\n";
print join ",", @L;
print "\n";
my %triplets;
for ( my $x = 0 ; $x <= $#L ; $x++ ) {
    my $start   = $x + 1;
    my $end     = $#L;
    my $current = $x;
    while ( ( $end - $start ) > 0 ) {
        my $result = $L[$current] + $L[$start] + $L[$end];
        if ( $result == $target ) {
            $triplets{"$L[$current],$L[$start],$L[$end]"} = 1;
            $start++;
            $end--;
        }
        elsif ( $result < $target ) { $start++ }
        else                        { $end-- }
    }
}
print "\nOUTPUT:\n";
print join "\n", keys(%triplets);

Roger Bell West

Roger Bell West’s solution uses an O(n³) nested loop, putting all of the results in a result (%r) hash of hashes (HoH):

my @l = ( -25, -10, -7, -3, 2, 4, 8, 10 );
my $t = 0;
@l = sort { $a <=> $b } @l;
my %r;
foreach my $a ( 0 .. $#l - 2 ) {
    foreach my $b ( $a + 1 .. $#l - 1 ) {
        foreach my $c ( $b + 1 .. $#l ) {
            if ( $l[$a] + $l[$b] + $l[$c] == $t ) {
                $r{ $l[$a] }{ $l[$b] }{ $l[$c] } = 1;
            }
        }
    }
}

Roger then iterates through the keys in %r and the sub-hashes to print the results in order:

foreach my $d ( sort { $a <=> $b } keys %r ) {
    foreach my $e ( sort { $a <=> $b } keys %{ $r{$d} } ) {
        foreach my $f ( sort { $a <=> $b } keys %{ $r{$d}{$e} } ) {
            print "$d + $e + $f\n";
        }
    }
}

Ruben Westerberg

Ruben Westerberg’s solution implements a combinations sub that generates all possible 3-tuples and greps for those tuples whose values sum to the $target:

use List::Util;
my @list=(-25, -10, -7, -3, 2, 4, 8, 10);
my $target=0;
#my @combinations=combinations(\@list,3);
my @triplets=grep {my @s=sort(@$_); @s ~~ @$_ }
grep { List::Util::sum(@$_)==$target}
combinations(\@list,3);
printf "3 Sum triplet: %s+%s+%s=%s\n", @$_,$target for (@triplets);
sub combinations {
    my @combinations=();
    my ($data,$size)=@_;
    my @indexes=(0) x ($size+1);;
    my $i=0;
    until ($indexes[$size]) {
        my $count=List::Util::uniq(@indexes[0..$size-1]);
        #print $count,"\n";;
        push @combinations, [@$data[@indexes[0..$size-1]]] if $count == $size;
        $indexes[0]++;
        for (0..$size-1) {
            if ($indexes[$_] != 0 and 0 == ($indexes[$_] % @$data)) {
                $indexes[$_]=0;
                $indexes[$_+1]++;
            }
        }
    }
    @combinations;
}

Ryan Thompson

My solution eliminates the inner loop by building a hash of numbers that are greater than a given number, with some help from a temporary copy of the array:

    my @a = @_; my %Lh = map { shift @a => { map { $_ => 1 } @a } } 1..$#a;

After this line, the keys of %Lh are the numbers, and the values are hashes of numbers in @_ that are greater. For example, if @_ = (1, 5, 10, 12), then:

%Lh = (
    1 => { 5 => 1, 10 => 1, 12 => 1 },
    5 => {         10 => 1, 12 => 1 },
   10 => {                  12 => 1 },
   12 => {                          },
);

Here is the full subroutine:

sub sum3 {
    my $T = shift;
    my @a = @_; my %Lh = map { shift @a => { map { $_ => 1 } @a } } 1..$#a;
    my @r;
    while (my $x = shift) {
        $Lh{$_}{ $T-$x-$_ } and push @r, [$x, $_, $T-$x-$_] for @_;
    }
    @r;
}

This turns an O(n) loop into an O(1) lookup, and given that loop was the inner loop in an n³ algorithm, it’s now n². It returns a result for a 1000-number input list in just 498501 iterations in about 0.3 seconds.

Blog3Sum and Colourful Numbers

Saif Ahmed

Saif Ahmed’s solution gives us the findTriplet sub, an O(n³) traversal of all possible combinations, with some bounds checking:

sub findTriplet {
    my ( $target, @list ) = @_;
    @list = sort { $a <=> $b } @list;             # the list may not be sorted, so sort first
    my @found = ();                               # initialise list triplets found
    foreach my $i ( 0 .. $#list - 2 ) {           # first number cannot be the last two numbers in the list
        foreach my $j ( $i + 1 .. $#list - 1 ) {  # second number bigger than first but can not be last number
            foreach my $k ( $j + 1 .. $#list ) {  # third number bigger than second in list
                                                  # check for triplet and if found add triplet to @found
                unshift @found, [ $list[$i], $list[$j], $list[$k] ]
                    if $list[$i] + $list[$j] + $list[$k] == $target;
                # dump duplicates using smartmatch (only check if 2 or more triplets found)
                shift @found if ( @found >= 2 and @{ $found[0] } ~~ @{ $found[1] } );
            }
        }
    }
    if (@found) {
        print scalar @found, " Triplets found\n";
        foreach my $triple (@found) {
            print "[ $$triple[0], $$triple[1], $$triple[2] ] ";
        }
    }
    else {
        print "No Triplet found that add to $target\n";
    }
}

The bounds optimizations bring the iterations down by around 83% for a 1000-item list, at around 166 million.

Ulrich Rieke

Ulrich Rieke’s solution is the O(n²) wiki solution:

sub findTriplets {
    my $sum   = shift;
    my $array = shift;
    my $len   = scalar @{$array};
    my @result;
    if ( $len < 3 ) {
        return ();
    }
    foreach my $i ( 0 .. $len - 2 ) {
        my $a     = ${$array}[$i];
        my $start = $i + 1;
        my $end   = $len - 1;
        while ( $start < $end ) {
            my $b = ${$array}[$start];
            my $c = ${$array}[$end];
            if ( $a + $b + $c == $sum ) {
                my @innerresult;
                push( @innerresult, $a, $b, $c );
                push( @result, \@innerresult );
                $start++;
                $end--;
            }
            elsif ( $a + $b + $c > $sum ) {
                $end--;
            }
            else {
                $start++;
            }
        }
    }
    return @result;
}

User Person

User Person’s solution is an O(n³) implementation with the usual bounds optimizations for the usual 83% savings compared to pure brute force:

my %seen = ();
for         ( my $i =    0; $i <= $#L-2; ++$i ) {
    for     ( my $j = $i+1; $j <= $#L-1; ++$j ) {
      INNER:
        for ( my $k = $j+1; $k <= $#L  ; ++$k ) {
            my $sum = $L[$i] + $L[$j] + $L[$k];
            if ($sum == $T) {
                my $string = "$L[$i] + $L[$j] + $L[$k] = $T";
                if (exists($seen{$string})) {
                    next INNER;
                } else {
                    print $string . "\n" if $sum == $T;
                    $seen{$string}++;
                }
            }
        }
    }
}

Walt Mankowski

Walt Mankowski’s solution uses Algorithm::Combinatorics

use Algorithm::Combinatorics qw(combinations);
use List::Util qw(sum);
my @L = (-25, -10, -7, -3, 2, 4, 8, 10);
@L = sort {$a <=> $b} @L;
my $target = 0;
my $iter = combinations(\@L, 3);
while (my $p = $iter->next) {
    say prettyprint($p, $target) if sum(@$p) == $target;
}

On my 1000-item test run, while the number of combinations was 166 million (83% less than the 1 billion of pure n³), it took over 4 minutes to run. Algorithm::Combinatorics is known to be somewhat slower than Algorithm::Permute, so that is the likely reason.

The prettyprint sub is a thoughtful touch:

sub prettyprint($p, $target) {
    my $s = $p->[0];
    for my $i (1..$#$p) {
        $s .= $p->[$i] >= 0 ? ' + ' : ' - ';
        $s .= abs($p->[$i]);
    }
    $s .= " = $target";
    return $s;
}

It takes care of the signs, so instead of seeing things like -7 + -3 + 10, you instead see -7 - 3 + 10, which is more natural.

Wanderdoc

Wanderdoc’s solution is a slightly faster variant of the O(n²) algorithm:

sub find_triplets {
    my @arr = sort { $a <=> $b } @{ $_[0] };
    my $t   = $_[1];
    my @result;
    for my $pt1 ( 0 .. $#arr ) {
        next if $pt1 > 0 and $arr[$pt1] == $arr[ $pt1 - 1 ];
        my $pt2 = $pt1 + 1;
        my $pt3 = $#arr;
        while ( $pt2 < $pt3 ) {
            if ( $arr[$pt1] + $arr[$pt2] + $arr[$pt3] == $t ) {
                push @result, [ $arr[$pt1], $arr[$pt2], $arr[$pt3] ];
                $pt2++;
            }
            elsif ( $arr[$pt1] + $arr[$pt2] + $arr[$pt3] < $t ) {
                $pt2++;
            }
            else {
                $pt3--;
            }
        }
    }
    return @result;
}

The key to the speedup is this line:

        next if $pt1 > 0 and $arr[$pt1] == $arr[ $pt1 - 1 ];

If the list contains duplicates, skipping them early can result in an improvement. But then I wondered whether this was better, worse, or the same as filtering out uniq elements at the top of the function call instead. Here are the results:

╭───────────────┬────────────┬──────────╮
│ Test          │ Iterations │ Time(ms) │
├───────────────┼────────────┼──────────┤
│ Original code │    320,101 │      162 │
│ No dupe check │    478,298 │      234 │
│ uniq @L       │    213,531 │      124 │
╰───────────────┴────────────┴──────────╯

Sometimes my curiosity takes me to interesting places. Other times it has me drawing tables in Vim.

Yet Ebreo

Yet Ebreo’s solution also implements the O(n²) wiki solution:

sub get_triplets {
    my ($list, $target) = @_;
    my @sorted = sort { $a - $b } @{$list};
    for my $i (0..@sorted-2) {
        my $a = $sorted[$i];
        my $start = $i + 1;
        my $end = @sorted - 1;
        while ($start < $end) {
            my $b = $sorted[$start];
            my $c = $sorted[$end];
            if ($a + $b + $c == $target) {
                say "$a $b $c";
                $start++;
                $end--;
            } elsif ($a + $b + $c > $target) {
                $end--;
            } else {
                $start++;
            }
        }
    }
}


Task #2 - Colourful Number

Original task description:

Write a script to display all Colorful Number with 3 digits.

A number can be declare Colorful Number where all the products of consecutive subsets of digit are different.

For example, 263 is a Colorful Number since 2, 6, 3, 2x6, 6x3, 2x6x3 are unique.


Solution types

There was really only one way to do it, with some variations:

Check every number O(n)

Since our range only spans 900 numbers, it’s reasonable to simply check them all, and see which ones are Colourful. To check whether a number is Colourful, one simply has to try all of the multiplications, and then make sure they are all unique:

sub colourful {
    my ($x, $y, $z) = split '', shift;
    6 == uniq $x*$y, $x*$y*$z, $y*$z, $x, $y, $z;
}

I use uniq here, but using a hash works equally well.

The task specifically calls for 3-digit numbers, but hopefully you can see expanding the above to handle longer numbers would be possible. Some of you did that, but most stuck to the task description.

No matter how you do it, the efficiency of this method is O(n) on the number of integers in the target range. I strongly suspect you can do no better than O(n), but I don’t have time to consider a proof. I’d be very happy to be proven wrong!


Adam Russell

Adam Russell’s solution is another Prolog relation using AI::Prolog. First, the supporting Perl code:

use AI::Prolog;
MAIN:{
    my $prolog = do{
        local $/;
        <DATA>;
    };
    $prolog = new AI::Prolog($prolog);
    for my $n (100..999){
        my($x, $y, $z) = split(//, $n);
        $prolog->query("colorful($x, $y, $z).");
        my $result = $prolog->results;
        if($result){
            print "$n: colorful number\n";
        }
        else{
            print "$n: not a colorful number\n";
        }
    }
}

And here is the Prolog, from the __DATA__ segment:

colorful(X, Y, Z) :-
    A is X * Y,
    B is Y * Z,
    C is X * Y * Z,
    X \= Y,
    Y \= Z,
    X \= Z,
    X \= A,
    X \= B,
    X \= C,
    Y \= A,
    Y \= B,
    Y \= C,
    Z \= A,
    Z \= B,
    Z \= C.

Even if you don’t know any Prolog, the above should be comprehensible: the colorful relation first sets A, B, and C to the results of the possible multiplications, and then the rest of the relation asserts that all of these (and the initial X, Y and Z inputs) must be different from each other.

BlogMore Perl & Prolog

Alicia Bielsa

Alicia Bielsa’s solution handles numbers of arbitrary length:

sub isColorful {
    my $number = shift;
    my @aDigits = split ('', $number);
    my %hProducts = ();
    my $lengthNumber = scalar(@aDigits);
    foreach my $subset  (1..$lengthNumber){
        foreach my $i (0..$#aDigits){
            my $setFound = 1;
            my $product  = 1;
            foreach my $s (0..$subset-1){
                if (defined $aDigits[$i+$s]){
                    $product *= $aDigits[$i+$s];
                } else {
                   $setFound = 0;
                }
            }
            if ($setFound){
                if (exists $hProducts{$product}){
                    return 0;
                } else {
                    $hProducts{$product} = 1;
                }
            }
        }
    }
    return 1;
}

I like this solution, in that it challenges the limits of the original task.

Andrezgz

Andrezgz’s solution uses a %unique hash to ensure he gets exactly six unique results from the multiplications:

for my $n (100 .. 999) {
    my %unique;
    my ($f,$s,$t) = split //, $n;
    @unique{$f, $s, $t, $f*$s, $s*$t, $f*$s*$t} = (1) x 6;
    print $n.$/ if keys %unique == 6;
}

The hash slice makes this one quite compact. I like it.

Cheok-Yin Fung

Cheok-Yin Fung’s solution uses Algorithm::Permute to generate all possible lists of $digits digits from 2..9. (Cheok Yin made the observation that the digits 0 and 1 will never appear in a number.)

use Algorithm::Permute;
use List::Util qw{product any};
my $digits = 3;   #can be changed to 2, 4 or 5,
                  #both produce colourful numbers within reasonable waiting time
my $permat    = Algorithm::Permute->new( [ 2 .. 9 ], $digits );
my @colourful = ();
while ( my @m = $permat->next ) {
    my @box = (0) x ( product( 10 - $digits .. 9 ) - 2 );

    # $box[0] is for result of multiplication as 2,
    # $box[1] is for result of a result of multiplication as 3,
    # $box[70] is for result of a result of multiplication as 72, etc..
    for my $j ( 0 .. $#m ) {
        my @a = ();
        for my $i ( 0 .. $#m - $j ) {
            push @a, $m[ $i + $j ];
            $box[ product(@a) - 2 ]++;
        }
    }
    push @colourful, scalar join( "", @m ) unless any { $_ > 1 } @box;
}
print join "\n", sort @colourful;

Colin Crain

Colin Crain’s solution uses a hash to determine if all products are unique:

sub colorful3 {
    my $number = shift;
    my ($hundreds, $tens, $ones) = split //, $number;
    my %products = map { $_ => 1 } ($hundreds, $tens, $ones, $hundreds * $tens,
                                    $tens * $ones, $hundreds * $tens * $ones);
    keys %products == 6 ? 1 : 0;
}

Cristina Heredia

Cristina Heredia’s solution includes a checkNumber sub that prints a message if the given number (passed in as a list of digits) is colourful:

sub checkNumber {
    $first = $aNumber[0];
    $second = $aNumber[1];
    $third = $aNumber[2];
    $firstMult = $first * $second;
    $secondMult = $first * $third;
    $thirdMult = $second * $third;
    $fourthMult = $first * $second * $third;
    if ($first != $second and $first != $third and $second != $third
        and $firstMult != $first and $firstMult != $second and $firstMult != $third
        and $secondMult != $first and $secondMult != $second and $secondMult != $third
        and $thirdMult != $first and $thirdMult != $second and $thirdMult != $third
        and $fourthMult != $first and $fourthMult != $second and $fourthMult != $third
        and $firstMult != $secondMult and $firstMult != $second and $secondMult != $third) {
        print "The number $number is a colorful Number";
    }
    else {
        print "The number $number is not a colorful Number";
    }
}

Note that instead of using uniq or a hash, Cristina checks every pair of products, which works equally well.

Dave Cross

Dave Cross’s solution uses a hash to check for uniqueness:

NUMBER:
for my $n ( 100 ... 999 ) {
    my %product;
    my ( $x, $y, $z ) = split //, $n;
    $product{$_}++ for ( $x, $y, $z, $x * $y, $y * $z, $x * $y * $z );

    say "$n is colourful" if keys %product == 6;
}

Dave Jacoby

Dave Jacoby’s solution fights back against feature creep, stating in a comment that handling arbitrarily large numbers “is a topic for another time”:

sub is_colorful( $d ) {
    my %test;
    my ( $i, $j, $k ) = split //, $d;    # 263 becomes 2,6,3
    map { return 0 if ++$test{$_} > 1; } $i, $j, $k, ( $i * $j ),
        ( $j * $k ), ( $i * $j * $k );
    return 1;
}

Dave also states that he “has been told that using map as a cheat array is dirty pool, but it makes this solution a bit cleaner.” Yet it works, is efficient, and is easy to understand and maintain. If that’s what “dirty pool” looks like then sign me up.

Duncan C. White

Duncan C. White’s solution implements a colourful function that handles arbitrarily long integers, fairly compactly:

use Function::Parameters;

fun colourful($x) {
    my @dig = split( //, $x );    # find all digits.
    my $n   = @dig;
    my %seen;                     # combinations already seen.
    foreach my $startpos ( 0 .. $n - 1 ) {
        foreach my $endpos ( $startpos .. $n - 1 ) {

            my $prod = 1;
            foreach my $p ( $startpos .. $endpos ) {
                $prod *= $dig[$p];
            }

            return 0 if $seen{$prod}++;
        }
    }
    return 1;
}

E. Choroba

E. Choroba’s solution also handles integers of arbitrary length:

use List::Util qw{ product };
sub is_colorful_number {
    my ($n) = @_;
    my $max_length = length $n;
    my %uniq;
    my $count = 0;
    for my $length (1 .. $max_length) {
        for my $pos (0 .. $max_length - $length) {
            my @pattern = ((0) x $pos, (1) x $length);
            undef $uniq{
                product((split //, $n)[ grep $pattern[$_], 0 .. $#pattern ])
            };
        }
    }
    return ($max_length ** 2 + $max_length) / 2 == keys %uniq;
}
say for grep is_colorful_number($_), 100 .. 999;

Although I use List::Util in most of my Perl modules, my brain for some reason still has a hard time remembering product exists, and the muscle memory of reduce { $a * $b } @_ is a stubborn habit to break. My lame excuse is that a List::Util with product was “just” added to core in Perl 5.20, six years ago.

Blog3 Sum and Colourful Numbers

Jaldhar H. Vyas

Jaldhar H. Vyas’s solution looks at 3-digit numbers, first gathering all the @products, then using a counting hash to see if any product is repeated more than once:

sub isColorful {
    my ($n) = @_;
    my %subsets;
    my @digits = split //, $n;
    my @products = @digits;
    push @products, $digits[0] * $digits[1];
    push @products, $digits[1] * $digits[2];
    push @products, $digits[0] * $digits[1] * $digits[2];
    map { $subsets{$_}++ } @products;
    return !grep { $_ > 1 } values %subsets;
}

Javier Luque

Javier Luque’s solution uses a 3-nested loop for $hundreds, $tens and $ones places, instead of looping from 100..999 and then using split:

my @solutions;
for my $h (2 ... 9) {
    for my $t (2 .. 9) {
        for my $o (2 .. 9) {
            if ( $h * $t != $t * $o &&
                 $h * $t != $h * $t * $o &&
                 $h * $o != $h * $t * $o &&
                 $t * $o != $h * $t * $o) {
                push @solutions, "$h$t$o";
            }
        }
    }
}
say join ' ', @solutions;

Note that this also returns numbers with repeated digits. Javier calculates the products correctly, but does not check whether the individual digits themselves are unique, as per the example in the task spec. This leads to numbers with repeated digits (e.g., 788, 998) being accepted instead of rejected.

Blog051 – Perl Weekly Challenge

Mohammad S Anwar

Mohammad S Anwar’s solution first asserts that all digits must be unique, and then goes about computing the various products, and then checks whether the number of products matches the number of unique products:

use List::Util 1.45 qw(uniq);
foreach my $i (234 .. 987) {
    my @N = split //, $i;
    next if ( scalar(uniq(@N)) != scalar(@N) );
    my @S = ($N[0], $N[1], $N[2], ($N[0] * $N[1]), ($N[1] * $N[2]), ($N[0] * $N[1] * $N[2]));
    print "[$i] => [", join(", ", @S), "]\n"
        if ( scalar(uniq(@S)) == scalar(@S) );
}

I’m glad to see Mohammad continuing to blog about his participation in the challenge.

BlogBLOG: The Weekly Challenge #051

Phillip Harris

Phillip Harris’s solution handles arbitrarily large integers, checking uniqueness with a hash:

sub checkcolorful() {
    my %products;
    my @digits = split //, $_[0];
    for ( my $x = 0 ; $x <= $#digits ; $x++ ) {
        for ( my $y = 0 ; $y + $x <= $#digits ; $y++ ) {
            my $total = 1;
            for ( my $z = 0 ; $z <= $x ; $z++ ) {
                $total = $total * $digits[ $y + $z ];
            }
            if ( $products{$total} == 1 ) {
                return 0;
            }
            else { $products{$total} = 1 }
        }
    }
    return 1;
}

Roger Bell West

Roger Bell West’s solution loops over the digits, $a, $b, and $c to get all candidate 3-digit numbers using the digits 2..9:

use integer;
use List::Util qw(max);
foreach my $a ( 2 .. 9 ) {    # 1?? will never be colourful
    foreach my $b ( 2 .. 9 ) {    # ?0? and ?1? will never be colourful
        if ( $a == $b ) {
            next;
        }
        foreach my $c ( 2 .. 9 ) {    # ??0 and ??1 will never be colourful
            if ( $a == $c || $b == $c ) {
                next;
            }
            my %p;
            $p{$a}++;
            $p{$b}++;
            $p{$c}++;
            $p{ $a * $b }++;
            $p{ $b * $c }++;
            $p{ $a * $b * $c }++;

            if ( max( values %p ) < 2 ) {
                print "$a$b$c\n";
            }
        }
    }
}

The products (and digits) are calculated and tallied in %p. Any hash whose max value is less than two does not have any repeated values.

Ruben Westerberg

Ruben Westerberg’s solution uses a nice map construct together with split and substr to generate the list of all possible products for an arbitrarily long number:

sub contiguousSubs {
    my ($string) = @_;
    my $len = length $string;
    map {
        my $subLen = $_;
        map {
            my @d = split "", substr( $string, $_, $subLen );
            List::Util::reduce { $a * $b } @d
          } 0 .. $len - $subLen
    } 1 .. $len;
}

Ryan Thompson

My solution uses a %seen hash to return 0 if any of the products have already been seen:

sub is_colourful3 {
    my ($x, $y, $z) = split //, $_[0];
    my %seen;
    $seen{$_}++ and return 0 for $x, $y, $z, $x*$y, $y*$z, $x*$y*$z;
    return 1;
}

Blog3Sum and Colourful Numbers

Saif Ahmed

Saif Ahmed’s solution handles arbitrary integers, using a %products hash to check for duplicates:

sub testColourful {
    my %products = ();
    my @digits   = split //, shift;

    foreach my $startDigit ( 0 .. $#digits ) {
        my $p = 1;
        foreach ( 0 .. $#digits - $startDigit ) {
            $p *= $digits[ $startDigit + $_ ];
            $products{$p}++;
        }
    }

    return keys %products == @digits * ( @digits + 1 ) / 2;
}

It’s interesting that the number of products is the @digits-th triangular number.

Ulrich Rieke

Ulrich Rieke’s solution computes all products and uses %numHash to tally the number of times each product appears:

sub isColourful {
    my $number = shift;
    my %numHash;
    if ( $number =~ /(\d)(\d)(\d)/ ) {
        $numHash{ $1 * $2 }++;
        $numHash{ $2 * $3 }++;
        $numHash{ $1 * $2 * $3 }++;
        $numHash{$1}++;
        $numHash{$2}++;
        $numHash{$3}++;
        $numHash{ $1 * $3 }++;
        return scalar keys %numHash == 7;
    }
    return 0;
}

There is one small issue, however: $numHash{ $1 * $3 }++ computes the product of the first and last digits, but only consecutive digits should be considered, per the task description. It’s an easy fix, though: remove that line, and change the next line to check %numHash == 6.

User Person

User Person’s solution handles 3-digit numbers with a %seen hash to check for duplicate products:

LOOP:
for (my $i = $MIN; $i <= $MAX; ++$i) {
    my %seen = ();
    my @d = (split //, $i );
    foreach ( $d[0], $d[1], $d[2], $d[0]*$d[1], $d[1]*$d[2], $d[0]*$d[1]*$d[2]) {
        exists $seen{$_} ? next LOOP : $seen{$_}++ ;
    }
    print "$i\n";
}

Walt Mankowski

Walt Mankowski’s solution handles arbitrary integers as well, using substr and split to build up the products to be calculated:

use List::Util qw(product);
sub is_colorful($n) {
    my %prods;
    for my $len (1..length($n)) {
        for my $i (0..length($n) - $len) {
            my $p = product split //, substr($n, $i, $len);
            if (defined $prods{$p}) {
                return 0;
            } else {
                $prods{$p} = 1;
            }
        }
    }
    return 1;
}

Wanderdoc

Wanderdoc’s solution handles arbitrarily large integers, with a few helper routines. The main is_colorful just calls prod_con_subsets and checks whether all of the products are unique:

sub is_colorful {
    my $num = $_[0];
    my @arr = prod_con_subsets($num);
    return scalar @arr == scalar uniq @arr;
}

prod_con_subsets does most of the heavy lifting, partitioning the number into all possible configurations for calculating products:

sub prod_con_subsets {
    my $num = $_[0];
    my $len = length($num);
    my @set;
    for my $width ( 1 .. $len ) {
        for my $pos ( 0 .. $len ) {
            my $chunk = substr( $num, $pos, $width );
            if ( length($chunk) == $width ) {
                push @set, $chunk;
            }
        }
    }
    my @products = _products(@set);
    $DEBUG_STORE{$num} = [@products];
    return @products;
}

The _products sub then takes a set of substrings, splits each one and calculates the product of its digits with reduce { $a * $b }:

sub _products {
    my @arr      = @_;
    my @products = map {
        my $el = $_;
        my $res =
          length($el) == 1
          ? $el
          : reduce { $a * $b }
        split( //, $el )
    } @arr;
    return @products;
}

is_colorful returns a boolean, but %DEBUG_STORE hash used above seems to be used to contain the list of products, so they can be printed out to prove a number is colourful:

my $counter;
for my $num ( 100 .. 999 ) {
    next if $num =~ /[01]/;    # cannot be colorful if contains 0 or 1.
    print "The Nr.", ++$counter, ' is ', $num, '; Proof: ',
      join( " ", @{ $DEBUG_STORE{$num} } ), $/
      if is_colorful($num);
}

Yet Ebreo

Yet Ebreo’s solution splits the number with a regex, and then places all possible products into @seq, sorted numerically:

sub is_colorful {
    my @digits = pop =~ /./g;
    my @seq = sort {$a - $b } (@digits, $digits[0] * $digits[1], $digits[1]* $digits[2], $digits[0] * $digits[1] * $digits[2]);
    return "@seq "!~ /(\d+ )\1/
}

Instead of using uniq or a hash, Yet Ebreo concatenates @seq into a string, and uses a regex to check for two consecutive instances of the same number.



See Also

Perl Blogs this week:

Adam RussellMore Perl & Prolog

E. Choroba3 Sum and Colourful Numbers

Javier Luque051 – Perl Weekly Challenge

Mohammad S AnwarThe Weekly Challenge #051

Ryan Thompson3Sum and Colourful Numbers

SO WHAT DO YOU THINK ?

If you have any suggestions or ideas then please do share with us.

Contact with me