11 views

Skip to first unread message

Mar 4, 2013, 4:18:03 AM3/4/13

to sage-algebra, sage-...@gmail.com

In a ring of characteristic 0, it seems that 0^0 (= 1) is well-

defined.

In my view this is correct. It makes it much simpler to define the

matrix (e.g. FF a finite field):

G = matrix([ [ a^i for a in FF ] for i in range(k) ])

However, in finite fields or any of the the following rings, 0^0

gives an error:

FF.<w> = FiniteField(32)

FF = FiniteField(31)

PF.<x> = PolynomialRing(FF)

The correct behaviour can be faked in an isomorphic ring which

is a quotient of something in characteristic zero:

PZ.<x> = PolynomialRing(ZZ)

R = PZ.quotient_ring([x-1,31])

Here R(0)^0 (= 1) is fine. Is there any objection to reporting this

as a bug and fixing it?

--David

defined.

In my view this is correct. It makes it much simpler to define the

matrix (e.g. FF a finite field):

G = matrix([ [ a^i for a in FF ] for i in range(k) ])

However, in finite fields or any of the the following rings, 0^0

gives an error:

FF.<w> = FiniteField(32)

FF = FiniteField(31)

PF.<x> = PolynomialRing(FF)

The correct behaviour can be faked in an isomorphic ring which

is a quotient of something in characteristic zero:

PZ.<x> = PolynomialRing(ZZ)

R = PZ.quotient_ring([x-1,31])

Here R(0)^0 (= 1) is fine. Is there any objection to reporting this

as a bug and fixing it?

--David

Reply all

Reply to author

Forward

0 new messages

Search

Clear search

Close search

Google apps

Main menu