Final Report
November 4, 2015
The 2015 ACM ICPC East Central North America Regional Programming Contest (ACM ICPC ECNA RPC) was held on Saturday, October 31, 2015. We had 130 teams from 55 colleges and universities throughout western Pennsylvania, Ohio, Michigan, eastern Ontario, and Indiana (excluding the Greater Chicago Metropolitan Area).
As it was difficult for some teams in the ECNA region to travel to a single site for the contest, the ECNA RPC was held as a distributed contest. There were four contest sites:
- University of Cincinnati in Cincinnati, Ohio (31 teams from 12 schools)
- Grand Valley State University in Allendale, Michigan (26 teams from 11 schools)
- University of Windsor in Windsor, Ontario (23 teams from 10 schools)
- Youngstown State University in Youngstown, Ohio (50 teams from 22 schools)
The ECNA RPC was successful, although there were two issues of note; those are discussed more fully below. First, some observations:
- The software environment was largely unchanged from last year, with the exception of updating versions of compilers. A more thorough update is planned for next year.
- We used Kattis this year for the contest control system. ECNA staff received a great deal of positive feedback on the switch, especially from the Grand Valley site.
- The contest started promptly at 10am and ended at 3pm.
- The contest consisted of 9 problems with 97 teams solving at least one problem.
- The maximum number of problems solved by any team was 9.
Issues
Problem B Judges' Solutions Incorrect
An appeal was made regarding the interpretation of the procedure described in problem B. In the problem, a rectangle was to be drawn on a screen whose bounds were specified in the input. The appeal claimed that the judges’ solutions allowed the rectangle to be drawn beyond the borders of the screen. While the intent of the judges was to keep the rectangle on-screen, there were four test cases where the rectangle would have to be extended beyond the borders in order to get answers that matched the judges’ output.
It was discovered that both of the judges’ solutions were flawed and neither solution realized that the rectangle extended beyond the screen border.
The four test cases were corrected and all 57 submissions for the problem were reexamined. The results of the examination are summarized below.
- 22 gave wrong answers for the initial and corrected test cases
- 14 gave incorrect answers for test cases where test cases did not need correction
- 9 submitted after the team submitted an accepted solution
- 5 were initially accepted
- 1 had a compile error
- 1 exceeded the time limit on a test case on which it was not previously tried due to failing an earlier test case
- 1 crashed with the corrected data
- 4 were initially denied but accepted with the corrected test cases
A corrected problem was sent to Kattis to provide corrected results.
Of the four submissions rejudged to be accepted, two were by teams that had a later submission that was accepted on the original test data and two were by teams that were not credited with solving the problem.
The changes in standing as a result of rejudging were as follows:
- Three teams (Waterloo Black, Waterloo Red and CMU3) who were credited with solving problem B retain their credit with no time adjustment. The Waterloo teams each submitted one solution that was incorrect on both data sets before making a correct submission and CMU3 had only one submission for the problem.
- CMU2’s second submission was accepted on the corrected test data. Since they already had a submission accepted, their time was reduced by 84 minutes – 44 minutes for the time difference in correct submission and 40 penalty minutes.
- Waterloo Gold’s first submission was accepted on the corrected test data. Since they also had a submission accepted, their time was reduced by 42 minutes – two minutes for the time difference in correct submissions and 40 penalty minutes.
- Waterloo White’s fifth submission was accepted on the corrected test data. Since they were not credited with solving the problem, their problem solved count increased to six and 361 minutes were added – 281 minutes to a correct solution and 80 penalty minutes.
- CMU1’s eighth submission was accepted on the corrected test data. Since they were not credited with solving the problem, their problem solved count increased to six and 433 minutes were added – 293 minutes to a correct solution and 140 penalty minutes.
This caused some movement among the top nine teams; Waterloo White and OSU_1 exchanged places, CMU1 moved from ninth place to fifth place, Toronto Blue moved from fifth place to sixth place, CMU3 moved from sixth place to eighth place and Waterloo Red moved from eighth place to ninth place.
Password issue
For increased security, Kattis used two sets of passwords – one for the practice contest and a second set for the actual contest. The second passwords were delivered to the RCD about two hours after the practice contest ended; these were then distributed to the sites.
At the start of the contest, teams at two sites continued to use the first set of passwords and were unable to log in at the start of the contest.
After discussion with the sites and confirmation from Kattis staff, the problem was recognized and the second passwords were distributed. This caused a delay of 12 minutes at one site and somewhat longer at the second site. It should be noted that six of the top ten teams were at delayed sites.
You can view the final standings here.
You can view the problem sets, test data and judges' solutions here.
Special Awards
Congratulations to the following teams on their awards:
- Jolly Jumpers (highest jump in rank from the previous year)
Black Hornets, jumping from 42nd place in 2014 to 17th place - Extreme Programmers (earliest correct solution)
Champions of the West, solving problem D in 0:23 - Steadfast Gurus (last correct solution)
SVSU Cardinal, solving problem E with 5 minutes remaining - Solid Programmers, solving the most problems with fewest
penalties
Orange Hornets, Euclid and Oberlin Prime, each solving three problems with only two incorrect submissions for those three. - Relentless Programmers (most attempts before correct solution)
Steins; Gate, solving problem E on their 9th attempt
Note: Teams are only eligible to win one prize.