On 14 November 2014 09:22, Bruno Grenet <bruno.gre...@gmail.com> wrote: > 2014-11-14 10:05 GMT+01:00 rjf <fate...@gmail.com>: >> >> My point here is that an unenlightened and obscure part of a problem >> with one computer program has (I think mistakenly) been elevated to >> a discussion of mathematics, open source, computer program reliability, >> etc. It was probably not reviewed by any computer scientist with >> expertise in computer >> algebra systems. Should AMS publish a followup? Should it try to >> find appropriate reviewers this time? > > > I may precisely be because the article was not reviewed by someone with > expertise in computer science and thus because it was a very naive article > (at least I feel so, when the authors conclude that mathematicians should > use two distinct software to check their results!), that it would be a nice > thing to have a follow-up explaining precisely that the approach proposed in > the previous article was too naive and that there are solutions to behave in > a scientifically acceptable manner. > > Bruno
I think Bruno has a good point. The fact the original article in a maths journal is quite flawed in some ways, does warrant a follow up comment. If you search the archives, you will see I contacted Wolfram Research, asking whether we could compare the output of Wolfram Alpha, and record that in the doc tests. They agreed - despite some saying they would ignore my request. So I am not totally against comparing software A to B. Using Wolfram Alpha is in some ways preferable to Mathematica, as it is free to use (or at least was). But the naive approach proposed by the trio of mathematicians can lead one into a false sense of security, because of the amount of code that is published with a license that permits its incorporation into closed-source software. I have shown, beyond any reasonable doubt, that Mathematica 7.0.1 on Solaris used both GMP and ATLAS. There is also the possibility that two or more pieces of software make use of an algorithm that is in itself flawed, or from a paper that has a typo. I personally think a follow up highlighting the danger of their proposed approach is actually more useful than documenting how a particular bug was corrected in Sage, and also has a far greater likelihood of being accepted for publication. Dave -- 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.
