Forum archive 2000-2006

Brian Camp - error messages and an error

Brian Camp - error messages and an error

by Arnold Pizer -
Number of replies: 0
inactiveTopicerror messages and an error topic started 9/24/2006; 7:35:30 PM
last post 9/25/2006; 12:39:35 AM
userBrian Camp - error messages and an error  blueArrow
9/24/2006; 7:35:30 PM (reads: 149, responses: 5)
I am not sure if this is the appropriate place to post this (if it is not, please let me know). Anyway ...

I think I have found an error in the following problem: rochesterLibrary/setIntegrals2Indefinite/ur_in_2_1.pg

In particular, a student with a seed number of 2451 has already entered the answer correctly, such as ...

(-2/3)*[x^(-6)]+(8/3)*ln(x)

Yet WW is reporting this as incorrect. I have not seen this problem come up with other seeds.

To make matters worse, the student is now putting in an incorrect answer and getting an Message that says "Can't solve for adaptive parameters". They are using ** notation for exponents although I am not sure that this makes a difference.

Unfortunately, I am not good enough yet at decoding .pg files to see what the error might be. I suspect there may be three possibilities. 1. there really is an error in the .pg code that affects some random seeds but not others 2. this function for some reason is having difficulty being checked using the fun_cmp macro (??) 3. the student already inputted the correct answer, and now is inputting an incorrect answer. Does this have an adverse reaction in WW?

Well, I am perplexed. Any help would be appreciated.

Brian

<| Post or View Comments |>


userJohn Jones - Re: error messages and an error  blueArrow
9/24/2006; 9:26:51 PM (reads: 181, responses: 0)
There is a problem in the way the answer is checked - the standard domain is not suitable. In this case, apparently it was only an issue for some seeds (different seeds lead to different formulas for the answer, so some may be harder to check than others). There is a version of this in the National Problem Library, which is now fixed.

Note, the student's answer is not correct in that it is missing absolute value bars for the log. This wasn't the problem for your student, but the "NPL" version mentioned above requires that they be there to mark the student's answer correct.

John

<| Post or View Comments |>


userMichael Gage - Re: error messages and an error  blueArrow
9/24/2006; 9:57:53 PM (reads: 179, responses: 0)
This is a good place to post questions like this.

When you suspect there is just a mistake in the problem (as in your case) you can also report the bug directly to our bug database bugzilla.

To do this. While looking at the student's version of the problem, click "edit this problem" and then on the editor page click "report problem bugs" (a link to the upper right of the problem code textarea box).

The first time you do this you will have to "signup", which involves sending in your email so that bugzilla can email you the results of the bug fix. Bugzilla will email you a password which you should then change to something you can remember. After that signin is automatic if you stay on the same machine (bugzilla leaves a cookie) or just a matter of entering your email and password.

In any case clicking on the "report problem bugs" link automatically posts the address of the problem in the library and the seed being used. You can add any further comments. Someone will take care of fixing the bug fairly quickly.

As to short term fixes when a student has found a bug in a problem. You can.

(1) Change his seed -- this is done from the homework set editor by clicking in the "edit assigned users" column and then clicking on the link to "edit data for ..." the individual student. You will see an entry for the seed for each problem. If the error in the problem very rarely occurs then this can work pretty well.

(2) Give the student credit for the problem. This is done by proceeding as above, but when you get to the student's data instead of changing the seed for the problem, change the "status" from 0 to 1.

(3) If the problem is causing troubles for everyone you can set everyone's status to 1 by going to the homework set editor; clicking in the "edit problem" column and changing the status in for the problem to 1 -- this page edits information for all student versions of the problem, so setting the status equal to 1 will set it to 1 for everyone.

-- Mike

<| Post or View Comments |>


userBrian Camp - Re: error messages and an error  blueArrow
9/24/2006; 10:31:38 PM (reads: 177, responses: 0)
Mike, thanks for the note about changing seeds and status. I changed the seed for this student.

Is it also possible to delete a problem from a homework set? For example, if this turns out to be a problem for a significant portion of the class, then what happens if I delete the problem from the set before the due date?

John, I will look for that problem in the National Library. Thanks.

To anyone, about bugs in general. If one is found while students are in the middle of a problem set, and someone else goes in and fixes the bug, what effect does this have on the problem set? Are the problems local in some way so that fixes to the database at the various schools do not actually effect the homework version that I have put together for my students?

Generally, I am somewhat curious to know what sort of activities on my part are safe to do while my students are actively working on a homework set. That is, if I try to edit a problem that they are working on, will they see this? Will it effect the recorded scores that they have obtained to that point in any way? [added later: I notice that the previous message thread addresses this to some extent although I am still interested in whatever others have experienced].

I would love to be able to have the confidence to make changes to a homework set that is in progress. However, I am afraid that I will do something that will lead to the loss of the student's work and they will have to start over at square one. Not usually the way to become endeared to the students :)

Brian

<| Post or View Comments |>


userMichael Gage - Re: error messages and an error  blueArrow
9/25/2006; 12:25:19 AM (reads: 173, responses: 0)
Is it also possible to delete a problem from a homework set? For example, if this turns out to be a problem for a significant portion of the class, then what happens if I delete the problem from the set before the due date?

this is usually a bad idea, since it is hard to know what to do with the scoring -- also -- should the problems be renumbered? and if so when I refer to problem 5 is it before the deletion or afterward? It's been hard to figure out a "right" behavior for this situation which fits the various circumstances.

I usually just mark the problem correct for all students (set status to 1) one could also change the problem so it has 0 "weight" meaning that the problem doesn't count at all. These accomplish approximately the same ends but psychologically students prefer the former.

Fixes in the library will not be propogated to your problems until or unless you do a cvs update on your library.

Mild editing of a problem while it being used will not cause much trouble. It will not change the scores of students who already have credit for the problem (perhaps by entering an incorrect answer that webwork accepted). Any edits -- changes in wording, or modification of how the answer is checked will be used in the future. Obviously, updates won't appear on hard copies that have already been donwloaded.

I edit problems to change wording or to add a hint from time to time when a student's question has made me realize that the problem's presentation might be confusing some students.

Other than deleting either a problem or worse a student's entire problem set, any modifications you make will give the benefit of the doubt to the student. Unless you explicitly set it to 0 their score is the best score they have received in all of their attempts.

<| Post or View Comments |>


userJohn Jones - Re: error messages and an error  blueArrow
9/25/2006; 12:39:35 AM (reads: 171, responses: 0)
Hi,

On updating problems while they are in progress, I have done it many times. If the person making changes knows well enough what they are doing, they can fix the problem without changing the randomization. That is possible on this problem. If you want to modify it, the new ANS line is

 

ANS(fun_cmp($answer, mode=>"antider", limits=>[[-100,100]]) );

The problem library version would be at

Library/Rochester/setIntegrals2Indefinite/ur_in_2_1.pg

but one would have to update the library from cvs to get the new corrected version.

John

<| Post or View Comments |>