From 8050bc367aefab125ffc207b9895e080ef25ecf6 Mon Sep 17 00:00:00 2001
From: Jörg Sommrey <28217714+jo-37@users.noreply.github.com>
Date: Mon, 20 Jul 2020 15:29:18 +0200
Subject: solution for task 2

---
 challenge-070/jo-37/perl/ch-2.pl | 62 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 62 insertions(+)
 create mode 100755 challenge-070/jo-37/perl/ch-2.pl

diff --git a/challenge-070/jo-37/perl/ch-2.pl b/challenge-070/jo-37/perl/ch-2.pl
new file mode 100755
index 0000000000..bbb999f225
--- /dev/null
+++ b/challenge-070/jo-37/perl/ch-2.pl
@@ -0,0 +1,62 @@
+#!/usr/bin/perl
+
+# Deeply impressed by E. Choroba's wonderful solution to task 1 from
+# challenge 068, I realized the ease of performing linear algebra
+# within PDL.  So I switched from Math::Matrix to PDL.
+use PDL;
+# PDL and Test2::V0 both export 'float' by default.
+# Resolve the conflict:
+use Test2::V0 '!float';
+
+# As described in the (German) Wikipedia, the Gray code is a linear
+# transformation (modulo 2) acting on the bit representation of its input
+# value.  Let
+# G = [[1, 1, 0, ..., 0, 0],
+#      [0, 1, 1, ..., 0, 0],
+#      ...
+#      [0, 0, 0, ..., 1, 1],
+#      [0, 0, 0, ..., 0, 1]]
+# be a N x N matrix with all elements of the main diagonal as ones,
+# all elements of the super diagonal as ones and all other elements zero.
+# Then the Gray encoding of a number with bit representation b as a row
+# vector of length N, is the (matrix) product b . G.
+
+# Generate G for the given length and return a sub ref that Gray encodes
+# a number. 
+sub gray_encoder {
+	my $len = shift;
+
+	# Construct G.
+	my $g = PDL->zeroes(byte, $len, $len);
+	# Set main and super diagonal to ones.
+	$g->diagonal(0, 1) .= 1;
+	$g->slice('1:,:-2')->diagonal(0,1) .= 1;
+
+	sub {
+		# Create a row vector from the bits of the given number.
+		my $in = pdl split //, sprintf "%0${len}b", shift;
+		die "arg not valid for encoder" unless $in->dim(0) == $len;
+
+		# Calculate the bits of the Gray encoded number.
+		# Note: with PDL, the 'x' operator represents the usual matrix
+		# product, not the vector cross product.
+		my $out = ($in x $g) % 2;
+
+		# Return the Gray encoded number.
+		# Note: $out is not a row vector, but a 1 x N matrix
+		# where the two dimensions need to be flattened into one.
+		local $" = '';
+		oct "0b@{unpdl $out->squeeze}";
+	}
+}
+
+# One single encoder with the maximum required bit length fits for all
+# smaller numbers.
+my $gray = gray_encoder 5;
+
+my $n = 4;
+my @out = map $gray->($_), (0 .. 2**$n - 1);
+is \@out, [0, 1, 3, 2, 6, 7, 5, 4, 12, 13, 15, 14, 10, 11, 9, 8],
+	'example from challenge';
+
+done_testing;
-- 
cgit