# Graduate Students Seminar

### On Blackboard Collaborate

Location

Online

Date & Time

November 17, 2021, 11:00 am – 12:00 pm

Description

Session Chair: | Lara John |

Discussant: | Dr. Gobbert |

###### Speaker 1: Daniel Kelly

**Title***Performance Study on MPI Commands per Iteration of the Power Method***Abstract**- Calculating the largest eigenvalue in absolute value of a square matrix A has many applications. One iterative algorithm that approximates the largest eigenvalue is the power method. As each iteration of the power method requires a matrix-vector calculation and a norm calculation, the number of calculations increases as the number of dimensions (n) of A increases. Thus the power method was implemented utilizing parallel computing to speed up the run time. Implementing a C program in parallel requires the use of MPI commands. There exist MPI commands that can perform the same functionality as multiple other MPI commands at the same time. This means the power method can be implemented with a varying number of MPI commands per iteration. We want to run a performance study comparing the run times of the power method to the number of MPI commands per iteration.

###### Speaker 2: Lillian Chow

**Title***How to Run R Programs on the Taki Cluster in the UMBC HPCF***Abstract**- Parallel computing refers to the process of taking a large problem and breaking it down into smaller parts which can be executed simultaneously and then combined upon completion to generate one solution. The main goal of parallel computing is to increase computational power for faster problem solving. R is an open-source statistical computing software that many statisticians and data scientists use to compute statistical analyses. In this presentation, I will demonstrate how to access and use R on the taki cluster in the UMBC High Performance Computing Facility (HPCF) using the interactive mode and the batch mode, I will also demonstrate how to install R packages and how to execute serial programs and parallel programs using the Rmpi Package.

**Tags:**