my @keys = keys %hash;

for (my $i = 0; $i < @keys; $i++) {
  for (my $ii = $i+1; $ii < @keys; $ii++) {
      compare($hash{$keys[$i]}, $hash{$keys[$ii]});
  }
}
[download]

That's really cool.Thanks to all of you,Appreciate it.

use strict;
use warnings;

my %hash = (
    key1 => [qw(a b c d e f)],
    key2 => [qw(a c d e f)]
);

my @a1 = @{$hash{key1}};
my @a2 = @{$hash{key2}};

for my $i (0 .. $#a1) {
    unless (exists $a2[$i]) { 
        print "No match for $a1[$i]\n";
        next;
    }
    if ($a1[$i] eq $a2[$i]) {
        print "equal: $a1[$i], $a2[$i]\n";
    }
    else {
        print "not equal: $a1[$i], $a2[$i]\n";
    }
}

__END__

equal: a, a
not equal: b, c
not equal: c, d
not equal: d, e
not equal: e, f
No match for f
[download]

Using exists on array elements makes no sense. The information it returns is useless.

my @a;
$a[1] = undef;
$a[2] = 'foo';

print("0: ", exists($a[0]) ?1:0,"\n");  # 0
print("1: ", exists($a[1]) ?1:0,"\n");  # 1
print("3: ", exists($a[3]) ?1:0,"\n");  # 0
[download]

exists $a2[$i] should be $i < @a2 or defined $a2[$i] depending on what you were trying to accomplish.

Using exists on array elements makes no sense.

I disagree. It seems like the documentation to which you link also disagrees:

Given an expression that specifies a hash element or array element, returns true if the specified element in the hash or array has ever been initialized, even if the corresponding value is undefined.

It clearly mentions "array"; I must be missing your point.

Here is what I was trying to accomplish. The OP has two arrays of different sizes. I decided to loop over the larger array (a1). In order to avoid a warning when checking the last element of a1 versus a non-existent element of the smaller array (a2), I decided to use exists. In my example, it Does What I Want.

Thanks for pointing out $i < @a2; I definitely prefer it to my usage of exists in this case.

That's one way to do it,But is it the best way? The values are already in the hash as arrays,Why should one store them again in separate ones?

my @a1 = @{$hash{key1}};
my @a2 = @{$hash{key2}};
[download]

You can avoid the copying by using references:

my $a1 = $hash{key1};
my $a2 = $hash{key2};
...
    if ($a1->[$i] eq $a2->[$i]) {
[download]

which is functionally the same as

    if ($hash{key1}->[$i] eq $hash{key2}->[$i]) {
[download]

but less to type if you need to access the arrays ($a1/$a2) more than once...