In the version of Library/TCNJ/TCNJ_Eigenvalues/problem10.pg that we have at U. C. Davis there appears to be a problem if the student's seed is 3077. In this case the eigenvalues are -8 and -6 and the eigenvectors are (1,1) and (1,0) respectively. When I, the instructor, enter these values and click on "Show correct answers" WeBWorK displays the answers thus: 1; 1; 1; 0; but says that only one of the answers is correct. A simple check (or Wolfram Alpha) will show that (1,1) and (1,0) are indeed the correct answers, but WeBWorK will not accept (1,0) as a correct answer. However, when I change the seed to 77 the problem numbers change and WeBWorK will accept the correct answers as correct.

By the way in MATLAB notation the matrix is A = [-6, -2; 0, -8].

I am not sufficiently familiar with the ins and outs of "answer rules" to have been able to debug the problem in a half hour or so. I'm attaching the problem file to this post.

Re: There is a subtle bug in Library/TCNJ/TCNJ_Eigenvalues/problem10.pg

by Gavin LaRose - Wednesday, 21 March 2012, 9:22 AM

Hi Gerry,

It looks as if John Jones patched this in the SVN in January. Could you check to see if your NPL revision is earlier than 1479 (or simply update the problems to determine if that resolves the problem)?