NemoViz: a visual interactive system for atomistic simulations design

Abstract

Introduction: Given that the nanoscale regime has been reached, atomistic simulations are being used as predictive tools on a nanoscopic scale in nanoelectronics, materials science, and computational fluid dynamics. These simulations are generally supported by complex and specific simulation engines that require a deep knowledge of the engine as well as the field. The inputs required by these engines are usually described in long-text files, and their composition is error prone. These files contain details about the corresponding materials, geometries, algorithms, initial values, etc. The introduction of visualizations to the simulation creation process could alleviate some common user issues with simulation creation, in particular regarding the input-text files. In other fields, visual analytics have been used to facilitate user interaction with complex data. This work features a case study on the creation of a nanoelectronic device simulation and provides evidence of how the use of visual analytics reduces the cognitive complexity of defining an atomistic simulation. Case description: NEMO5 is a tool designed to simulate the electronic properties of nanoelectronic devices on an atomistic level. In this work we introduce NemoViz, an interactive visualization tool that enables users to define the simulation inputs required by NEMO5. Regular NEMO5 users were exposed to NemoViz, and users’ effectiveness and efficiency was measured while debugging the input for a simulation. Discussion and Evaluation: The results of this work show that NemoViz reduced the time that users spent defining the inputs of a NEMO5 simulation. When using NemoViz, Expert NEMO5 users detected errors twice as fast as when NemoViz was not used, and non-expert NEMO5 users were able to detect errors as effectively and efficiently as expert users. Conclusions: These results suggest that the use of visual analytics as a simulation design process tool reduces the cognitive load of complex simulators such as NEMO5. Users’ interaction with visualization facilitates their understanding of output results and input descriptors, which may lead to new research codes from their widespread user base.