06 Feb 2020

Three UH professors use high performance computing resources to enhance their courses

Three University of Hawai‘i (UH) at Mānoa professors made use of Mana, the UH high performance computing (HPC) cluster this fall semester to teach three graduate-level courses, taking advantage of the speed, efficiency, and volume of data that can be processed with the cluster. 

Professor of Entomology Dr. Daniel Rubinoff, Assistant Professor of Chemistry Dr. Rui Sun and Assistant Professor of Civil and Environmental Engineering Dr. Jonghyn Harry Lee taught Systemics and Phylogenetics (PEPS 662), Computational Study of Condensed Matters: from Theories to Applications (CHEM 761) and the newest course, Deep Learning in Civil and Environmental Engineering and Earth Science (CEE 696) respectively. All three courses allowed students the opportunity to use MANA, a 298 node (6308 core) compute cluster housed with Information Technology Services (ITS) throughout the fall semester. 


Rubinoff, Sun and Lee engaged students in computational science not only to create awareness of this multidisciplinary field but also to equip students with the skills necessary to use advanced computing to solve and understand complex problems. 


Rubinoff taught PEPS 662 students basic computational skills for biologists known as bioinformatics. According to Rubinoff, bioinformatics has become an integral part of almost all fields of biology and prepares students to become competitive in whatever field they choose to work in the future. 


“In recent years due to technological advances and reduced costs of DNA sequencing, it is not uncommon for students to have large genome-scale datasets to analyze,” Rubinoff said. “The computational power required to estimate phylogenies – evolutionary relationships between species – from such data grows exponentially making it impossible to run on your basic desktop computer.”

Running computationally heavy class assignments was made possible for these three courses through the support of ITS Cyberinfrastructure, a group within ITS that provides HPC services and training for UH System. ITS has provided HPC resources and training at no cost since the inception of the cluster in 2014, all UH faculty, staff, students and affiliates.

Sun created an opportunity for CHEM 761 to tour the Information Technology Centerʻs Data Center to see first-hand where the UH HPC resides and learn about the infrastructure that supports this resource. 


CHEM 761 students tour the Information Technology Centerʻs Data Center to see first-hand where the UH HPC resides.

Sun, who described the first module of his course as challenging, teaches students much-needed skills to write their own programs and apply them to real-life problems such as simulating natural selection. Sun, who completed the second iteration of his course this past semester, emphasized the importance of having courses that teach computational skills for non-computer science majors.  


“I realized that there is an urgent need to incorporate basic training of computation into the natural science catalog,” Sun wrote in a grant proposal. “I have decided to dedicate my expertise to introduce computational modeling and simulations to enhance students’ learning experience and to better prepare them for their careers as data processing with computers becomes increasingly important.”

Lee who also emphasized the importance of high performance computing led efforts to incorporate programming skills in CEE 696 where Python programming knowledge was a prerequisite. High performance computing allows users to solve large problems in areas of science, business, and engineering by using aggregate computer power that allows for higher performance than that of an individual computer. This becomes important for innovation in areas such as artificial intelligence (AI). 

CEE 696 student Jinwen Xu presents his deep learning project on rainbow detection from images

According to Lee, deep learning, a subset of AI and machine learning that aims to train computers to learn by taking in new information, deciphering it and producing an output, has become popular because of its use of large data sets. 

“Thanks to recent advances in data acquisition and computational power, deep learning algorithms can construct layered high-level representations of nature and engineered systems in a way that maximizes performance on a given task, which has shown a great potential to complement traditionally established domain models.” Lee wrote in the course syllabus.

A number of students attested to the rigorous nature of the three courses where programming skill sets varied for each course. However, despite the level at which students started, they walked away from the courses knowing more than when they began.

“There is no limit to the applications”, CEE 696 student Madeline Mckenna said. McKenna, a UH Mānoa Ph.D. student in atmospheric science admitted not initially realizing the application of deep learning to climate science. McKenna, encouraged by her advisor Assistant Professor of Atmospheric Science Dr. Christina Karamperidou took the course and learned skills to help the many projects in which she is involved. “Taking the course was definitely eye-opening and challenging”.

With enhanced focus on Data Science education and skills at the state and nation-wide levels, efforts from faculty such as Rubinoff, Sun and Lee to incorporate these skills put UH students at an advantage when pursuing educational and career goals. 

In addition, UHʻs Hawai‘i Data Science Institute offers a Data Science Fellows program training initiative that funds graduate assistants and undergraduate fellows to further their knowledge in the theory, techniques and applications of data science. 

The Data Science Institute was recently awarded a $1 million National Science Foundation grant and will add computational resources to the cluster throughout the coming year with equipment arriving as early as March 2020.