## Forum archive 2000-2006

### Lars Jensen - fun_cmp broken?

by Arnold Pizer -
Number of replies: 0
 fun_cmp broken? topic started 9/20/2005; 3:04:25 PMlast post 9/20/2005; 7:07:39 PM
 Lars Jensen - fun_cmp broken?  9/20/2005; 3:04:25 PM (reads: 354, responses: 1) Is the fun_cmp evaluator broken in the webwork 2.1 ? When I try for example the problem rochester_problib/setAlgebra15Functions/jay5.pg the answer evaluator doesn't seem to care about the variable part of the answer. The correct answer is listed as: ((7)*x+20000) but it takes 20000 as a correct answer. It also takes ((2)*x+20000) as correct... ((100)*x+20000) is marked incorrect! -Or is this an artifact of the large numerical difference between the two numbers? Lars. <| Post or View Comments |>

 John Jones - Re: fun_cmp broken?  9/20/2005; 7:07:39 PM (reads: 467, responses: 0) Hi Lars, Yes, this is an artifact. Using ANS(fun_cmp($CO, limits=>[(-$b-10)/$a, (-$b+10)/$a]));ANS(fun_cmp($R));ANS(fun_cmp($P, limits=>[($b-10)/($c-$a), ($b+10)/($c-$a)]));ANS(num_cmp($BE)); assures checking so that the range of the 1st and 3rd parts is in [-10,10]. Then it will be more sensitive. John <| Post or View Comments |>