, also known as   is the U+00A0: NO-BREAK SPACE.
∼ is U+223C: TILDE OPERATOR.

HTML::Entities properly handles both just fine:

>perl -e"use HTML::Entities qw( decode_entities ); printf('U+%04X', or
+d(decode_entities($ARGV[0])))" " "
U+00A0

>perl -e"use HTML::Entities qw( decode_entities ); printf('U+%04X', or
+d(decode_entities($ARGV[0])))" "∼"
U+223C
[download]

I suspect you have a bug in your output code. You're probably forgot to encode the text string returned by decode_entities into a binary string appropriate for your terminal or the file into which you outputting the string.

This can be done by adding the :encoding(...) layer on open, by adding the :encoding(...) layer using binmode, or by explicitly encoding using Encode's encode function.