Immutable HammingCode?

[Gerli Viikmaa]

Hi,

Consider the following code (the field is arbitrary in this case)

sage: code = codes.HammingCode(2, GF(4,'a'))

The codewords (vectors) cannot be changed (note the absence of an error):

sage: code[0]
(0, 0, 0, 0, 0)
sage: code[0][0] = 1
sage: code[0]
(0, 0, 0, 0, 0)

yet they are not immutable, even when trying to set them so:

sage: code[0].is_immutable()
False
sage: code[0].set_immutable()
sage: code[0].is_immutable()
False

Is there any particular reason why the codewords are not immutable if they cannot be changed anyway?

If they were, using them as dictionary keys would be less of a hassle:

sage: {code[0]: "element"}
TypeError: mutable vectors are unhashable

I guess the current options are

sage: v = code[0]
sage: v.set_immutable()
sage: {v: "element"}

or

sage: {tuple(code[0]): "element"}

neither of which is particularly elegant.

Is there a way to set the all of the codewords as immutable?

I'm using Sage version 6.1.1 on a Linux machine.

Gerli

[P Purkayastha]

The reason why it behaves this way is because the code[0] command regenerates the codeword every time you call it. The reason for doing so is in http://trac.sagemath.org/ticket/13694

So, code[0][0]=1 is indeed changing the vector, but it is not changing the vector that is displayed by the next code[0] command. If you want a set of immutable vectors, then you can do something like this:

sage: C = codes.HammingCode(3, GF(2))
sage: codevectors = C.list()
sage: for c in codevectors: c.set_immutable()
sage:
sage: D = {c: c.hamming_weight() for c in codevectors}

[P Purkayastha]

It may not be such a bad idea to return codevectors which are immutable. I have opened http://trac.sagemath.org/ticket/16338 to discuss this.

On Monday, May 12, 2014 11:12:19 PM UTC+8, Gerli Viikmaa wrote: