Converting a Singular list to a Sage list

14 views
Skip to first unread message

Francisco Botana

unread,
Jun 4, 2011, 1:00:25 PM6/4/11
to sage-s...@googlegroups.com
Hi,

How can I convert a Singular list to a Sage list?
For instance:
T is a Singular list with value

[1]:
_[1]=1
[2]:
_[1]=1
[3]:
[1]:
[1]:
_[1]=0
[2]:
[1]:
_[1]=(x^2+y^2-4*y+3)
[2]:
_[1]=(x^4+2*x^2*y^2-9*x^2+y^4-9*y^2+4*y+12)

I can get a Sage string (using singular.eval(T)) but it would be more
convenient getting a (structured) Sage list. Is there a command for
doing so?

Thanks.

Simon King

unread,
Jun 5, 2011, 2:26:40 AM6/5/11
to sage-support
Hi Francisco,

On 4 Jun., 19:00, Francisco Botana <fbot...@uvigo.es> wrote:
> How can I convert a Singular list to a Sage list?

It *should* be working (if L is a list in the Singular pexpect
interface) with L.sage().

However, it doesn't at all. I am really surprised that it doesn't,
because such a feature should have been implemented a long time ago.
Perhaps other people can point us to a solution? For now I can only
offer part of a solution, as sketched below, hoping that you consider
a special case for which the partial solution is enough.

First, L.sage() does not work, e.g., in the following example:

sage: singular.eval('ring R')
sage: L = singular('ringlist(R)')
sage: L
[1]:
32003
[2]:
[1]:
x
[2]:
y
[3]:
z
[3]:
[1]:
[1]:
dp
[2]:
1,1,1
[2]:
[1]:
C
[2]:
0
[4]:
_[1]=0
sage: L.sage()
<KABOOM>
sage: L[1].sage() # that's just an integer!
<KABOOM>
sage: L[2][1].sage() # that's just a string!!
<KABOOM>


> I can get a Sage string (using singular.eval(T)) but it would be more
> convenient getting a (structured) Sage list. Is there a command for
> doing so?

I think string parsing is not a good idea. It seems better to ask for
the type of each item on the list, and then uses a specialised method.
Note that there are specialised methods such as sage_poly() for
polynomials. However, this requires that you have already a ring in
Sage that matches the basering in Singular.

Anyway, back to the example:

sage: L[1].typeof()
int
sage: Integer(L[1])
32003
sage: L[2][1].typeof()
string
sage: repr(L[2][1])
'x'
sage: L[3].typeof()
list
sage: L[3][1].typeof()
list
sage: L[3][1][1].typeof()
string
sage: repr(L[3][1][1])
'dp'

Etc.

For ideals, one may try to do

sage: singular.eval('ring r = (9,a),(x,y,z),dp')
sage: I = singular.ideal(['x^2','y*z','z+x'])
# First, convert the base ring into Sage
# One needs to do special cases for different types of base rings:
# - integers
# - rational field
# - field extension
# Here, we have a finite field, and may do
sage: RS = singular('basering')
sage: br = GF(*repr(RS.charstr()).split(','))
sage: br
Finite Field in a of size 3^2
sage: vars = [repr(singular.var(i)) for i in range(1,RS.nvars()+1)]
sage: vars
['x', 'y', 'z']
sage: P = br[tuple(vars)]
sage: P
Multivariate Polynomial Ring in x, y, z over Finite Field in a of size
3^2

And then, there is again a case of a feature that should work since a
long time. As I have mentioned above, there is a method sage_poly,
taking a ring as a additional argument. So, why does it not work?

sage: I[1]
x2
sage: I[1].sage_poly(P)
...
TypeError:

There's even no error message.

I suppose what happens here is that I[1].sage_poly(P) tries to use the
string representation, which is a bad thing to do with "x2".

Anyway, IMHO it *should* work to do
sage: P*[p.sage_poly(P) for p in I] # not implemented
Ideal (x^2, x*z, x + z) of Multivariate Polynomial Ring in x, y, z
over Finite Field in a of size 3^2

In the example that you gave, the string representation of polynomials
seems fine (i.e., x^2, not x2). So, there is some hope that it works
for you.

Best regards,
Simon

Simon King

unread,
Jun 6, 2011, 2:10:31 AM6/6/11
to sage-support
Hi Francisco,

On 5 Jun., 08:26, Simon King <simon.k...@uni-jena.de> wrote:
> Anyway, IMHO it *should* work to do
> sage: P*[p.sage_poly(P) for p in I]  # not implemented

My fault: Singular has an optional parameter, determining whether a
short or a long polynomial representation is used. So, I should have
done:

sage: singular.eval('ring R')
'ring R;'
sage: singular.eval('short=0') # That's what I forgot to do
'short=0;'
sage: RS = singular('basering')
sage: br = GF(*repr(RS.charstr()).split(','))
sage: br
Finite Field of size 32003
sage: vars = [repr(singular.var(i)) for i in range(1,RS.nvars()+1)]
sage: vars
['x', 'y', 'z']
sage: P = br[tuple(vars)]
sage: I = singular.ideal(['x^2','y*z','z+x'])
sage: I # now it is a long representation: x^2, not x2
x^2,
y*z,
x+z
sage: P*[p.sage_poly(P) for p in I]
Ideal (x^2, y*z, x + z) of Multivariate Polynomial Ring in x, y, z
over Finite Field of size 32003

I think the procedure above should be automated (but the term order
should be taken care of as well, I suppose). I'll open a ticket.

Cheers,
Simon

Simon King

unread,
Jun 6, 2011, 5:46:09 AM6/6/11
to sage-support
Hi Francisco and all other potential reviewers,

On 6 Jun., 08:10, Simon King <simon.k...@uni-jena.de> wrote:
> I think the procedure above should be automated (but the term order
> should be taken care of as well, I suppose). I'll open a ticket.

It is #11431, and it seems ready for review. One can convert quite
complicated things with the patch. For example:

sage: singular.eval('ring r10 = (9,a), (x,y,z),lp')
'ring r10 = (9,a), (x,y,z),lp;'
sage: Q = singular('std(ideal(x^2,x+y^2+z^3))',
type='qring')
sage: Q.sage()
Quotient of Multivariate Polynomial Ring in x, y, z over
Finite Field in a of size 3^2 by the ideal (y^4 - y^2*z^3 + z^6, x +
y^2 + z^3)
sage: singular('x^2+y').sage()
x^2 + y
sage: singular('x^2+y').sage().parent()
Quotient of Multivariate Polynomial Ring in x, y, z over
Finite Field in a of size 3^2 by the ideal (y^4 - y^2*z^3 + z^6, x +
y^2 + z^3)

Cheers,
Simon
Reply all
Reply to author
Forward
0 new messages