Perl PDL中与R语言ifelse等效的实现方法是什么？

Great question! Coming from R's vectorized ifelse() function, it makes total sense to look for a more idiomatic (and efficient) PDL alternative instead of converting your PDL object to a Perl array with unpdl() and looping through elements.

Your current approach works, but it bypasses one of PDL's biggest strengths: native vectorized operations that run at C-level speed (instead of Perl-level loops). Here are a few better options:

Option 1: Vectorized arithmetic with logical masks #

PDL treats boolean operations as element-wise 0/1 values, so you can compute the result directly using arithmetic:

use PDL;

my $x = pdl(1,2,3,4);
# For odd elements: keep x; for even: x*2
my $result = $x * ($x % 2) + ($x * 2) * !($x % 2);

print $result; # Outputs [1 4 3 8]

You can even simplify this further by noticing that even elements just need to be multiplied by 2, while odds stay the same:

my $result = $x * (1 + ($x % 2 == 0));
# When $x is even, ($x%2 ==0) is 1 → 1+1=2 → x*2
# When $x is odd, it's 0 → 1+0=1 → x*1

Option 2: Use where() for explicit masking #

If you prefer something more readable (similar to how ifelse() reads in R), use PDL's where() method to select and modify elements:

use PDL;

my $x = pdl(1,2,3,4);
my $mask_odd = $x % 2; # 1s where elements are odd, 0s otherwise

# Combine the odd elements (unchanged) and even elements (doubled)
my $result = $x->where($mask_odd)->append( ($x * 2)->where(!$mask_odd) )->reshape($x->dims);

Or, if you want to modify a copy of the original array in place (super efficient for large datasets):

use PDL;
use PDL::NiceSlice; # Makes slicing syntax cleaner

my $x = pdl(1,2,3,4);
my $result = $x->copy;
# Replace even elements with their doubled values
$result->where(!($x % 2)) .= $x->where(!($x % 2)) * 2;

Why your original approach isn't ideal #

Using unpdl() converts your PDL object to a Perl array, and map() loops through each element at the Perl level. For small arrays this is fine, but with large datasets, you'll lose all of PDL's performance benefits—PDL is designed to handle bulk operations in C, avoiding Perl's slower loop overhead.

All the methods above keep the data in PDL's native format, so they're way faster for large arrays and fit better with PDL's idioms.

内容的提问来源于stack exchange，提问作者Stephan Loyd