Re: [WWdevel] Re: extraAnswerEvaluators.pl

[Davide P.C. <dp...@un...>]

> Now that I am thinking about it, I don't understand why 0 is special 
> at all when using absolute error.  I see why it is special when 
> dealing with relative error, but it should be just like any other 
> number when we have absolute error, or am I missing something?

As I mentioned, I had originally done it with the scientific notation 
in mind, and hadn't been thinking about the absolute tolerance case, so 
the result was s little surprising to me, too.  I also said that I 
didn't have a problem changing it in the absolute case.  Another 
possibility is to use the zeroLevel tolerances for this all the time so 
as to get scientific notation turned into zero in both absolute and 
relative cases, but not use the standard tolerances for that at all.

But I did want to consider the possibility of make the output results 
be more realistic for ALL answers (not just zero) by using the 
tolerances to show the displayed results with the proper number of 
digits of precision as set by the problem's tolerances.  This would 
make the student answer more closely correspond to how WW actually 
treats it, and would make it much clearer how many digits of precision 
must be used in the student's answer.  This would also help problem 
authors be aware of the effect of the tolerances on their answers, 
particularly when those answer involve either very large or very small 
numbers.  The standard tolerances won't distinguish between 12345 and 
12335, for example, and I doubt most problem writers would take that 
into account.

Davide