[LON-CAPA-users] Puzzling situation...

[Stuart Raeburn]

Hi,

> Thank you. Would that then explain the following bomb message:
>
> Sun Feb 24 06:15:32 pm 2013 (CST): Computer's answer is incorrect
> ("0.00"). It is likely that the tolerance range [5.488e-17, 5.712e-17] or
> significant figures [1, 2] need to be adjusted.

Yes.

This is the same thing as the student saw.  A submitted value of 0.00  
was compared with the acceptable range: 5.488e-17 - 5.712e-17 and  
found to lie outside the range, and so INCORRECT was the award.   
Machine precision meant that in this case the computation of $dY11  
resulted in a value of 5.6e-17, and with a 2% relative tolerance this  
gave the range of acceptable values for the answer of 5.488e-17 to  
5.712e-17.

One way to solve this issue would be to compute $dY11, and then test  
in the script block if it would be zero, allowing for machine  
precision. If that is the case, $dY11 would be explicitly set to 0

$dY11=$EAY11-$AY11;
if (abs($dY11) < 1.0E-16) {
     $dY11 = 0;
}
$dY11=&format($dY11,"2s");

The particular bomb message you saw would have originated when the  
"Computer's answer" of 0.00 was compared with the computed value of  
$dY11 which in this case was 5.6e-17. That comparison occurs:

(a) post-answer date, when the computer's answer is shown

or

(b) when a student's submissions are being viewed.


Before "testing" the answer, LON-CAPA converts the "Computer's answer",
as generated in the script block, to the format specified for the answer
(i.e., whatever is defined in the format attribute in the
numericalresponse tag) -- in this case 2f.

For a numericalresponse items, if the "grade" returned is not "correct",
and significant figures have been defined then the error message:

"It is likely that the tolerance range or significant figures need to be
adjusted." is set.


Stuart Raeburn
LON-CAPA Academic Consortium


Quoting "Mills, Douglas G" <dmills at illinois.edu>:

> Hi Stuart,
>
> Thank you. Would that then explain the following bomb message:
>
> Sun Feb 24 06:15:32 pm 2013 (CST): Computer's answer is incorrect
> ("0.00"). It is likely that the tolerance range [5.488e-17, 5.712e-17] or
> significant figures [1, 2] need to be adjusted.
> This error occurred while processing response 25 in part 320
> The error occurred on host uiucaccess2
> This error occurred on machine uiucaccess2
>
> And if so, how should we be accomplishing what we are trying to do?
> Thanks!
>
>
> Doug
>
> Douglas Mills
> Director of Instructional Technologies
> Department of Chemistry
> University of Illinois
> dmills at illinois.edu
> (217) 244-5739
>
>
>
>
>
>
>
> On 2/22/13 1:23 PM, "Stuart Raeburn" <raeburn at msu.edu> wrote:
>
>> Hi,
>>
>>> Any ideas anyone on why in the second instance ($db12, id=25) 0 is
>>> accepted as a correct answer but in the other instance ($dy11,
>>> id=22), 0 is returned as incorrect???
>>
>> Perhaps you are seeing an artifact of floating point calculations in
>> the situation where the answer is zero, and you are using a relative
>> tolerance?
>>
>> Stuart Raeburn
>> LON-CAPA Academic Consortium
>>
>>
>> Quoting "Mills, Douglas G" <dmills at illinois.edu>:
>>
>>> HI All,
>>>
>>> We're greatly perplexed by this situation.  We have a student whose
>>> results for two different responses in a part should be 0, but the
>>> answer is accepted for one of the responses and not accepted for the
>>>  other. To the best of my abilities to discern, everything about the
>>>  two items is identical (and having written these, it's almost
>>> certain that I cut and pasted my way from one to the other) and the
>>> data for the two are analogous.
>>>
>>> The formulas for computing the two answers are:
>>>
>>> $dY11=&format($EAY11-$AY11,"2s");
>>> $dB12=&format($EAB12-$AB12,"2s");
>>>
>>> The values for the variables in her instance are:
>>>
>>> $EAY11=4.2e-1
>>>
>>> $AY11=.42
>>>
>>> $EAB12=2.6e-1
>>>
>>> $AB12=.26
>>>
>>> The coding in the problems looks (is!?) identical as well:
>>>
>>> <numericalresponse format="2f" answer="$dY11" id="22">
>>> <responseparam name="tol" type="tolerance" default="2%"
>>> description="Numerical Tolerance" />
>>> <responseparam name="sig" type="int_range" default="1,2"
>>> description="Significant Digits" />
>>> <textline /> $PART20msg[1]
>>> </numericalresponse>
>>>
>>> <numericalresponse format="2f" answer="$dB12" id="25">
>>> <responseparam name="tol" type="tolerance" default="2%"
>>> description="Numerical Tolerance" />
>>> <responseparam name="sig" type="int_range" default="1,2"
>>> description="Significant Digits" />
>>> <textline /> $PART20msg[4]
>>> </numericalresponse>
>>>
>>> Any ideas anyone on why in the second instance ($db12, id=25) 0 is
>>> accepted as a correct answer but in the other instance ($dy11,
>>> id=22), 0 is returned as incorrect???
>>>
>>> Doug
>>>
>>> Douglas Mills
>>> Director of Instructional Technologies
>>> Department of Chemistry
>>> University of Illinois
>>> dmills at illinois.edu<mailto:dmills at illinois.edu>
>>> (217) 244-5739