Back to All Events

Interactive Visualization of Large Datasets - An Infinite Canvas

DATE:                        2.27.25

TIME:                         4:00pm

LOCATION:              31-270

SPEAKER:                Professor Graham Pullan, University of Cambridge

TOPIC:                        Interactive Visualization of Large Datasets - An Infinite Canvas Approach

 

 

ABSTRACT:

It is easy for us to generate lots of data. We can simulate one geometry at many flight conditions, or we can run computations of hundreds of different geometries as we explore a design space. Similarly, we may rapidly accumulate measurement data from flight, engine and experimental tests. Yet we lack adequate tools to interactively navigate, visualize and learn from these data. In this talk, an infinite canvas approach is presented that enables engineers to visualize large datasets of simulations (or experiments) in a flexible, dynamic and interactive manner. 

 

By separating results into high-level metrics (such as geometric parameters and integrated output metrics) and detailed simulation data (such as 3D surfaces) we can interactively filter the dataset and visualize specific cases of interest. The infinite canvas approach gives us freedom to explore. It allows meta-data and detailed data plots to be created, resized and moved in response to ideas and hypotheses as they arise. Plots can be swept off-screen and kept ready to compare with new visualizations as they are created. This approach is ideally suited to touchscreens where the data exploration process then takes on an added spatial and kinesthetic dimension. 

 

The talk focusses on an open source, web-based, infinite canvas implementation, (see dbslice.org). The utility of the approach is illustrated using an example workflow for a dataset of turbine blade optimisation simulations. The infinite canvas approach supports the engineer as they interact with the data, following ideas as they occur, switching emphasis freely between high-level representations of the whole dataset and detailed views of a chosen selection of detailed data.

 

BIO: 

Graham Pullan is Professor of Computational Aerothermal Design at the University of Cambridge. He is Co-Director of the Whittle Laboratory and leader of the Centre for Doctoral Training in Future Propulsion and Power. Graham has been a long-term collaborator with researchers at the Gas Turbine Lab. He has received six ASME Turbomachinery Best Paper Awards (two with Profs Greitzer and Spakovszky) and an ASME Gas Turbine Award (also with EG and ZS). Prof Pullan is currently on sabbatical at Aero-Astro and is deeply grateful for the warm welcome he has received.

Previous
Previous
November 7

A New Look at Air-Breathing Nuclear Propulsion