FAQs on the Project assignment
(Last update: Wed Oct 21 08:53:37 CEST 2009)

######################################################################

Q: For each of my heuristics they come up with the same results for the
   E-instances E-17 to E-24. When looking at the instance files they
   look a lot alike, is that intended?
   - Anders Nielsen Helmar (Wed Oct 21 08:53:37 CEST 2009)

A: No it was not intended :( I made a mistake while transforming these instances.
   I updated today the package. Please download it again.
   They should now be different.

######################################################################

A: Extended sets of instances for race testing are now available. They
   contain 24 instances of type E and 21 of type U.
   - Marco Chiarandini (Tue Oct 13 14:14:43 CEST 2009)


######################################################################

Q: When I try to run the solution checker on solutions with events that
   has not been assigned (i.e lines with '-1 -1' in the solution file)
   the checker aborts with the following message:

   fctt-checker: fctt-checker.cpp:176: int main(int, char**): Assertion
`a>= 0 && a<n_of_events' failed.  Abort

   The checker runs fine on solutions without these '-1 -1' lines. Is
   that an error in en solution checker?
   - Anders Nielsen Helmar (Wed Oct  7 18:10:32 CEST 2009)


A: I updated the checker so that it accepts also '-1 -1' assignments.

######################################################################
Q: What I understand from the project text is that the fourth part of
   the cost function ("number of students attending more than one event
   in a period") only increases with 1 for each student that has an
   event clash (even if the student has more than 2 events that
   clash). But it seems that fctt-checker counts the number of pairs of
   clashes for each student instead (which might be a better cost
   function). Is this intended?	    
   - Jakob Lykke Andersen (Thu Oct  1 08:57:03 CEST 2009)


A: You are absolutely right: the checker and the text were not
   congruent. I agree with your suggestion of counting the pairs and
   I updated the text accordingly:

   These are the changes in the new version of the text:
   Point 4 on page 3:

   OLD: number of students attending more than one event in a period,
   NEW: number of students attending two events that are scheduled in the
        same period,


   The formula (4) has also changed accordingly.
   The checker remained the same.

######################################################################
Q: Is it correctly understood that RandomTimetable first randomly picks
   a period checks whether or not the current event is available in that
   period, if not randomly pick another (assuming that there is at least
   one available period for each event).  then it randomly picks a room,
   but what if that room is occupied in that period?  should it pick
   another at random until it has checked all rooms or just output -1
   -1?	   
   - Kasper Tordrup (Wed Sep  30 18:57:03 CEST 2009)


A: No, the RandomTimetable does not do any feasibility check, it simply
   picks a random integer from the set of integer numbers that represent
   the periods and the rooms. It does not check whether the event is
   available or whether the room is not used already. It does not
   include the possibility to choose -1.

######################################################################
Q: What was the name of the random generator for java mentioned at the
   lecture? 	
   - Kasper Tofte (Wed Sep  30 09:57:03 CEST 2009)

A: See slides from lecture 9 that I put on the right-most column of the
   lecture plan. Direct link:

   http://www.imada.sdu.dk/~marco/Teaching/Fall2009/DM811/Slides/dm811-lec9.pdf