Trac #18269: A new structure for experimentation on decoding: communi…

…cation channels

For now, there is no structure to easily add errors in codewords. If one
wants to experiment with decoding algorithms on codes, the only possible
way is to add errors "by hand", which is rather tedious.

We propose here a new structure, based on communication channels, on
purpose to facilitate the experimentation process with decoding
algorithms.

For now, our structure consists of:

- an abstract class for channels,
- a channel which adds a specific number of errors at random positions
to each provided vector
- a channel which adds a specific number of errors at random positions
to each provided vector, and erases a specific number of random
positions

With this new structure, creating `n` errors in a vector can be done in
one line of code into Sage.
Adding a new Channel class should also be easy: all one needs to do is
to inherit from the abstract class, and override and implement a method.

For better consistency, channels can only be accessed using
channels.<name> (see #15445) from the global namespace.

URL: http://trac.sagemath.org/18269
Reported by: dlucas
Ticket author(s): David Lucas
Reviewer(s): Vincent Delecroix

src/doc/en/reference/coding/index.rst

@@ -6,6 +6,8 @@ Coding Theory
 .. toctree::
    :maxdepth: 1
 
+   sage/coding/channels_catalog
+   sage/coding/channel_constructions
    sage/coding/codes_catalog
    sage/coding/linear_code
    sage/coding/code_constructions

src/sage/coding/all.py

@@ -74,3 +74,6 @@
     ["Krawtchouk", "delsarte_bound_hamming_space", "delsarte_bound_additive_hamming_space"])
 
 lazy_import('sage.coding', 'codes_catalog', 'codes')
+lazy_import('sage.coding', 'channels_catalog', 'channels')
+
+import sage.coding.channel_constructions