From Earlham Cluster Department
The XSEDE12 Student Programming Contest is a competitive programming event which will take place as part of the XSEDE12 Conference in Chicago, IL. Teams consisting of up to five students (high school, undergraduate, and/or graduate students) will be given eight to twelve problems from various scientific problem domain areas. To accommodate travel impediments, at least one team member is required to be physically present at XSEDE12, since all program development, testing, and deployment are restricted to the on-site computational resources. The contest will be based on the LittleFe cluster platform (http://LittleFe.net) and the Bootable Cluster CD software stack (http://BCCD.net). Contestants are encouraged to familiarize themselves with them before the contest.
Problem descriptions will be released one week prior to the on-site competition. Full problem descriptions and data sets will only be accessible to the teams on the day of the actual event. Students will have the bulk of the day, from 8:00 am until 5:00 pm, to submit code and solutions to as many problems as they choose. Part of the judging criteria will involve documentation of the team's activities. Each team will keep an engineering journal using electronic lab notebooks provided by the contest organizers.
To register a team, ask questions, etc. send email to firstname.lastname@example.org
A team consists of up to five students from recognized high school, undergraduate, and/or graduate degree programs. Faculty, and other coaches not part of the student team, are encouraged to mentor students up to, and including, release of the problem descriptions; but should have no involvement with the students on any aspect of the actual competition once the full problem details are released on the day of the event.
Every team will be provided access to identical computational resources, a LittleFe v4 unit running the BCCD. Teams may not share resources. Team members may use their own laptops and the contest organizers remote computational resources if they request access to them. Team members will be able to install software on the LittleFe/BCCD unit during the contest.
How to use your USB drive
High Level Description of Problems
Folding, but not your laundry
This problem is based on the processes and chemistry used by chains of amino acids to "fold" into proteins. Students will design a parallel algorithm considering chains of hydrophobic and polar protein elements.
Encrypting and decrypting messages has been a long-standing use of a supercomputer of the day. While not a state secret cipher, this problem provides a message in a message in a message to uncover. The moral of the story is while there is a well-established place for high-level, highly-abstracted languages, algorithmic and performance issues may sometimes dictate other solutions.
This problem challenges you to discover algorithm(s) to simplify and solve a well defined logic puzzle, given that for a contest lasting only a few hours, a brute force algorithm could take days of run-time to complete.
Measure twice, cut once
In our constantly-evolving, many-core world, the task of load balancing is taking on more and more importance, as is the algorithmic load balancing of computational resources done within the parallel codes that computational scientists have written to implement their mathematical models. This problem offers students the opportunity to develop their algorithm for making load balancing decisions in the universe of hybrid computing, by deciding which tasks will be run together in shared memory mode, communicating between groups using message passing.
Problems involving a large number of objects where everything interacts with everything else is found throughout science, from the very small to the very large and a lot in between. This class of problems is called an n-body problem.
There is a neurotoxin outbreak in an underground laboratory. The scientists inside are completely unprepared for this type of epidemic so their best bet is to get out as fast as possible. The only way out of the lab is through the escape elevators (they’re too high-tech for stairs). This will involve random walking (or running in this case), probability, and spread of poisonous gas.
Quasicrystals can be modeled as tilings of two dimensional space. The cardinality of the set of possible tilings, ie the size of state-space, is directly related to the entropy of the crystal. Consequently, the problem of counting the number of tilings of 2-D space with a fixed number of shapes of tiles is as important as it is difficult. Solving the problem requires the application of numerical techniques of discrete math in a physics/materials science context.
Has it cooled down yet?
Simulated Annealing is used in function minimization as a method of finding the global minimum for problems with many free parameters and many local minima.
How fast can you run?
Often the best way to compare two or more computational resources is to analyze how quickly and efficiently they can each solve a given problem. Some of these applications give better light into the resource's performance than others. Be prepared to use some common benchmarking tools.
If you did the air traffic analysis problem, you would see that the data leave a lot to be desired. One way to “fix” the data is to create your own by simulation. This problem is the basis for the real air traffic management systems used in Europe. The USA is still mired in old manual technology. The object here will be to simulate the air traffic in a 2-D square sector.
General Purpose Graphical Processing Units (GPGPUs) are developing into a powerful and efficient computing platform. They are especially suited for data-processing applications, many of which can be modeled as matrix operations. In some cases, it can be advantageous to perform operations using groups, rings, or fields other than the integers or reals. Solving this problem requires implementing matrix operations over a Galois field on CUDA-enabled GPGPUs.
The following factors will be taken into consideration in ranking and awarding team honors:
- Process toward a solution
- A complete solution
- Extenuating circumstances of a complete solution, i.e. insightful elegant solutions
- Robustness of the solution.
- Is the code general, so that it will work with other data sets?
- Will perturbations in the data destabilize the code?
- Accuracy of the solution
- Quality of the write-up describing the teams activities to solving the problem
- Engineering journal
- How sound is the documented engineering process?
- Is the documented process consistent with the runs made through the development and scoring queues?
- Resources for Preparation and Team Support
- Computational Resources
- Looking into these resources ahead of time will help prepare students for the challenges they will face during the competition.
For more information, please direct questions to email@example.com
Maximum of 15
- Clark Atlanta University (Peter Molnar) - Travon Haynes, Stephen Jones, Jr., David Williams, Zerotti Woods, Jimmy Zhangj, Fatima Ojeda
- XSEDE Scholars 1 (Alice Fisher) - Manuel Zubieta, David Manosalvas, Grace Silva, Nancy Carloa
- XSEDE Scholars 2 (Alice Fisher) - Lindley Graham, Mechel'le Saenz, Agnes Ramos-Beauchamp, Alejandro Weibel
- XSEDE Scholars 3 (Alice Fisher) - Elizabeth Alvarez, Ivy Krystal Jones, Brian Flowers, Kolawole Oni
- XSEDE Scholars 4 (Alice Fisher) - Jasmine Bunch, Mariela Barrera, Malika Harris, Paul Delgado, Jaime Guajardo
- XSEDE Scholars 5 (Alice Fisher) - Amanuel Weldemariam, Alemsthay Abeje, Erika Jones, Ivan Zecena, Julie Pedraza
- XSEDE Scholars 6 (Alice Fisher) - Carla Smith, Abigail Groff, Peter Njenga, Daniel Calderon, Malcom Chitsa
- XSEDE Scholars 7 (Alice Fisher) - Kenisha Ford, Lourdes Espinoza, Joseph Young, Mylo Carter, Eric Santos
- XSEDE Scholars 8 (Alice Fisher) - Melissa Estrada-Maravilla, Alex Grabacki, Carlos Sanchez, Ifueko Igbinedion, J.D. DeVaughn-Brown
- XSEDE Scholars 9 (Alice Fisher) - Nitocris Perez, Adolfo Escobedo-Pinto, Rolando Olivares, Stefano Rensi, Anja Guillory
- University of Michigan (Benson Muite) - Brian Leu (firstname.lastname@example.org), Albert Liu (email@example.com), Parth Sheth (firstname.lastname@example.org), Zheyin Zhang (email@example.com)
- University of Notre Dame Summer Scholars - Team One (Timothy Stitt, Timothy.Stitt.firstname.lastname@example.org) - Rebecca Liu, Jacob Kassman, Evelyn Ding, Anna McMahon, Alex Cao
- University of Notre Dame Summer Scholars - Team Two (Timothy Stitt, Timothy.Stitt.email@example.com) - Austin Gwiazdowski, Alex Brizius, Nicholas DeMarshall, Jacob Miller, John Heintschel
- Tables, chairs, power strips (2 per) from the conference A/V group
- Monitors (we bring or conference A/V group?)
- Disable DHCP
- 15 LittleFes, gigabit switch, jumpers, spares, gear bags