The multi-day menu-planning problem is formalized and decomposed it into several sub-problems at the daily-menu and meal-planning level. An evolutionary algorithm is developed that quickly finds a diverse set of feasible solutions, without examining all the possibilities. As the problem is constrained, infeasible solutions need to be repaired. At the meal-planning level, repairing is coupled with linear programming to balance infeasible meals. Empirical results, which showed that the evolutionary method might outperform a human, are presented.
COBISS.SI-ID: 22525223
Ant-Colony Optimisation (ACO) is a popular swarm intelligence scheme for solving combinatorial optimisation problems. However, high-dimensional continuous optimisation problems remain a challenge. We developed an ACO-based algorithm for numerical optimisation capable of solving high-dimensional real-parameter optimisation problems. The algorithm, called the Differential Ant-Stigmergy Algorithm (DASA), transforms a real-parameter optimisation problem into a graph-search problem. The parameters’ differences assigned to the graph vertices are used to navigate through the search space. We show that the DASA is a competitive continuous optimisation algorithm that solves high-dimensional problems effectively and efficiently.
COBISS.SI-ID: 23618855
A hypercube is one of the widely studied computer architecture due to its elegant properties. We study the resilience to the removal of edges or robustness of network. The problems of mutually-independent Hamiltonian paths with prescribed end-vertices and mutually-independent starting Hamiltonian cycles in the hypercube with faulty edges are studied. The study is motivated by the problem of transferring different pieces of a given message from one vertex to all recipients simultaneously such that they never meet in the same vertex. The obtained results on mutually-independent Hamiltonian paths with prescribed end-vertices is used to prove that there are n – m mutually-independent starting Hamiltonian cycles in the n-dimensional hypercube with m faulty edges.
COBISS.SI-ID: 26622247
A hardware accelerator for the compression of LIDAR data has been developed. For this purpose, hardware predictors of the point coordinates and other attributes of LIDAR data were conceived. The predictors of the point coordinates consist of two methods: linear prediction using last coordinate changes, and the search for the closest coordinate change among the most recent coordinate changes. The applied method is dynamically selected based on the resemblance of the current search result. A pipelined hardware divider, required for linear prediction, was also developed. An adjustable pipeline depth enabled us to select the most suitable divider with respect to the dividers’ latency, the usage of the hardware resources, and the clock period. The coordinate prediction and the prediction of other LIDAR data attributes are used in the prediction compression of the LIDAR data. Additionally, a variable length encoder was developed, and the arithmetic coder was improved by using the barrel shifter structure, which resulted in up to 8-times higher data throughput. Modules were developed in the VHDL language and verified in the Cadence simulation environment. Individual modules were synthesized and tested on the Xilinx XUPV5 prototype board.
COBISS.SI-ID: 26726695
Geographical information systems are ideal candidates for the application of parallel programming techniques, mainly because they usually handle large data sets. To help us deal with complex calculations over such data sets, we investigated the performance constraints of a classic master–worker parallel paradigm over a message-passing communication model. To this end, we present a new approach that employs an external database in order to improve the calculation–communication overlap, thus reducing the idle times for the worker processes. The presented approach is implemented as part of a parallel radio-coverage prediction tool for the Geographic Resources Analysis Support System (GRASS) environment. The prediction calculation employs digital elevation models and land-usage data in order to analyze the radio coverage of a geographical area. We provide an extended analysis of the experimental results, which are based on real data from a Long Term Evolution network currently deployed in Slovenia. Based on the results of the experiments, which were performed on a computer cluster, the new approach exhibits better scalability than the traditional master–worker approach. We successfully tackled real-world-sized data sets, while greatly reducing the processing time and saturating the hardware utilization.
COBISS.SI-ID: 27452711