Either I don't understand how ctricpy is supposed to work, or it is broken, or both. If a complex matrix is fed to ctricpy, the result, its type and dimensions look strange. OTOH, with a "matrix" representing Re and Im, the result looks good, but it can be achieved using tricpy as well (after reordering dims).

#!/usr/bin/perl

use v5.24;
use warnings;
use PDL;
use PDL::LinearAlgebra::Complex;

my $m1 = pdl '[[0, i], [1, 1+i]]';
ctricpy($m1, 0, my $c = null);
say $c->info;
say "complex: $c";

my $m2 = sequence 2, 3, 3;
ctricpy($m2, 0, $c = null);
say $c->info;
say "Re + Im:", $c->reorder(1, 2, 0);
say "tricpy:", $m2->reorder(1, 2, 0)->tricpy(0);

__DATA__
PDL: LDouble D [2,2,1]
complex: 
[
 [
  [0 0]
  [1 1]
 ]
]

PDL: Double D [2,3,3]
Re + Im:
[
 [
  [ 0  2  4]
  [ 0  8 10]
  [ 0  0 16]
 ]
 [
  [ 1  3  5]
  [ 0  9 11]
  [ 0  0 17]
 ]
]

tricpy:
[
 [
  [ 0  2  4]
  [ 0  8 10]
  [ 0  0 16]
 ]
 [
  [ 1  3  5]
  [ 0  9 11]
  [ 0  0 17]
 ]
]
[download]

Adding complex as type to the input and output arguments in tricpy's signature makes it complex-aware and it works for real and complex matrices. Thus ctricpy might be superfluous. The result from an accordingly modified tricpy on the complex matrix $m1 is:

[
 [  0   i]
 [  0 1+i]
]
[download]