High-Performance Computing (HPC) at
Azerbaijan State Oil and Industry University

High-Performance Computing (HPC)

The High-Performance Computing (HPC) Center has been established within the General and Applied Mathematics department at ASOIU to facilitate research in technology, computer architecture, systems, algorithms, programs, fast computing, methods, and approaches. The primary objective of the center is to expedite and reduce the cost of solving complex problems such as computer simulations, financial tasks, and computer animations. Moreover, the HPC Center has initiated the Azerbaijan Journal of High-Performance Computing to disseminate the results of its work to the global scientific community and to publish original manuscripts detailing studies in HPC.

The center is equipped with twenty computing nodes and one server. These systems are installed with the Cent OS and Rocks versions of the Linux operating system, as well as all the necessary software for parallel programming to address the targeted challenges. Leveraging the capabilities of an HPC multi-partitioned cluster system, the following tasks have been successfully addressed:

  • Parallel numerical solution of the inverse problem of two-phase filtration
  • Resolution of the problem of selecting optimal feature vectors for recognition systems using parallel genetic algorithms
  • Training of a multi-layer neural network using HPC technology to predict the volume mass of alkali based on its concentration and percentage
  • Analysis of data collected to assess soil contamination in the Absheron Peninsula using machine learning methods and prediction of contamination at depths where data was not available
  • Examination of data obtained from the World Bank Data portal, leading to the revelation of the impact of ICT exports on the country's income and the main parameters influencing this factor's dynamics
  • Utilization of several methods, including the fuzzy inference system and the AHP method, to enhance the aggregation step in federated learning
  • Employment of convolutional neural and graph neural networks for text classification using data collected from an Azerbaijani language-based serious game.