Re: [sage-devel] Re: Dealing with different signs in doctest output?
On 07/02/2012 15:43, Jason Grout wrote: On 2/7/12 9:34 AM, John H Palmieri wrote: Or as part of the doctest normalize G.round(6): multiply by -1 if the real part of the (0,0) entry is positive. If it gets too complicated, maybe it should be moved to a TESTS block instead of an EXAMPLES block. I wonder what part of the algorithm leads to a difference choice of sign on this one platform. so something like: sage: G.round(6)*sgn(G[0,0].real()) Thanks, Jason These are quite ugly solutions. How do people feel about having a solution analogous to #10952? http://trac.sagemath.org/sage_trac/ticket/10952 For example, let # sign variation ignore every +, -, and space in the output string. Marco -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
Re: [sage-devel] Re: Dealing with different signs in doctest output?
On 07/02/2012 16:06, Jason Grout wrote: On 2/7/12 9:55 AM, Marco Streng wrote: On 07/02/2012 15:43, Jason Grout wrote: On 2/7/12 9:34 AM, John H Palmieri wrote: Or as part of the doctest normalize G.round(6): multiply by -1 if the real part of the (0,0) entry is positive. If it gets too complicated, maybe it should be moved to a TESTS block instead of an EXAMPLES block. I wonder what part of the algorithm leads to a difference choice of sign on this one platform. so something like: sage: G.round(6)*sgn(G[0,0].real()) Thanks, Jason These are quite ugly solutions. How do people feel about having a solution analogous to #10952? http://trac.sagemath.org/sage_trac/ticket/10952 For example, let # sign variation ignore every +, -, and space in the output string. But then it would consider the two vectors (-1,2) and (-1,-2) to be the same, which is not what we want. Jason True, that is the biggest downside of this solution. But examples should be illustrative. It would be a shame to have a documentation of a function without any printed examples of the output of the function. Users may want to see what the output is supposed to look like, as a way of understanding what a function does. We can always add another test like M*M.transpose() below it if we really are afraid of getting (-1,-2) instead of (-1, 2). -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
Re: [sage-devel] Re: Dealing with different signs in doctest output?
On Tue, Feb 7, 2012 at 8:11 AM, javier vengor...@gmail.com wrote: How about testig for the desired mathematical properties? If this is a Gram-Schmidt test, the resulting matrix M should be orthogonal, so we can test for M*M.transpose() being the identity matrix (up to numerical accuracy). Of course, this type of indirect test should belong to the test section, not to the examples section. +1, this is the best way to test it. However, this kind of of indirect test is exactly the thing that *should* go in the examples section (whether or not it's needed for testing), as it is a good illustration of what the expected mathematical properties are. (In fact I consider tests like this to be much better than here's the answer tests; one could take the Gram-Schmidt of a random 100x100 matrix and verify it's nearly orthogonal rather than print out the meaningless coefficients themselves.) - Robert -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org