On Thu, Dec 4, 2014 at 2:04 AM, Vincent Delecroix <20100.delecr...@gmail.com> wrote: > Hi Simon, > > 2014-12-04 9:54 UTC+01:00, Simon King <simon.k...@uni-jena.de>: > (posted on sage-support) >> Hi Vincent, >> >> On 2014-12-04, Vincent Delecroix <20100.delecr...@gmail.com> wrote: >>> sage: M = matrix(RR, [[-1]]) >>> sage: abs(M) >>> -1.00000000000000 >>> >>> So the problem is with abs(M). The reason is that abs(M) is calling >>> the method M.__abs__(). The latter one is just a shortcut for the >>> determinant. I really do not understand why and it looks like a bug to >>> me. >> >> No, see discussion on sage-devel. >> >>> For comparison, in scipy the same function just return the matrix >>> where each entry is replace with its absolute value. Which is much >>> more natural... >> >> ... which *is* a bug, IMHO. > > Nope as in scipy matrices are considered as array (e.g. exp(m) applies > exp to all of the entries and is not the exponential of the matrix). > And the reason for that is numerical analysis. > > I agree that Sage point of view on matrices is rather different. > >>> I opened a ticket for that: >>> http://trac.sagemath.org/ticket/17443 >>> It should be corrected in the next stable release. >> >> Better not. Expecting abs(M) to return the matrix formed by the absolute >> values of matrix M is a misuse. Generally, matrices in Sage are not >> considered to be arrays. Hence, element-wise operations generally don't >> seem natural. > > All right, but in that case I would prefer that there is no __abs__ at > all. What about deprecating its usage and later on return a ValueError > ? I found it very misleading, and I am not the only one. > >> The standard notation for abs(x) is |x|, and if M is a matrix then it is >> relatively common to write |M| for its determinant, which justifies that >> use of abs(M) for that purpose (although M.det() would be clearer). > > This is completely crazy!
I understand why you find this surprising. I would have no problems deprecating this use of abs on matrices. I just don't think it should be elementwise abs. David > > Vincent > > -- > You received this message because you are subscribed to the Google Groups > "sage-devel" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to sage-devel+unsubscr...@googlegroups.com. > To post to this group, send email to sage-devel@googlegroups.com. > Visit this group at http://groups.google.com/group/sage-devel. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
