Global Scalable Optimisation
Scalability of systems performance becomes a challenge for modern digital systems. Achievement of system ability to complete wide range of tasks in terms of computational performance and effective use of resources require substantial research. Significant efforts are directed towards design of large scale hardware systems. However resolving scalable tasks require also scalable software.
The aim of this Competition is to highlight the results achieved by various methods on two global optimisation test for wide range of dimensions.
Test functions
Keane's Bump Test Function is hard constrained global optimisation problem. For this completion it is transformed for maximization.
where xi are the variables (expressed in radians) and n is the number of dimensions. Optimal values for variety of dimensions of this test are unknown.
For details please see http://www.soton.ac.uk/~ajk/bump.html
Michalewicz test function is global optimisation problem. For this completion it is transformed for maximization.
The maximum is dependent on dimensions number and for variety of dimensions is unknown.
For details please see https://www.sfu.ca/~ssurjano/michal.html
Completion conditions:
1. Number of parameters- Group from 50 to 400 parameters
- Group from 401 to 2000 parameters
- Group above 2001 parameters
- Better solution – better solution has priority
- Better precision – higher number of valid digits after the decimal comma has priority.
- Authors names and affiliation
- Method name
- Optimal value
- Variables values, which generates the Optimal value
- Constraints values for Test 1
- Optionally – Objective function evaluations number, time for calculation, computer system characteristics
5. Final results will be published on the conference website.
6. Accepted submission should be presented on the Competitions Session at HPC 2019
7. Best results will receive - HPC 2019 Certificate, which indicates the place
8. Please submit to competitionhpc@parallel.bas.bg