Skip to content

Latest commit

 

History

History
27 lines (18 loc) · 2.37 KB

README.md

File metadata and controls

27 lines (18 loc) · 2.37 KB

AlgoSphere

This is a repo for storing all the content from the different meetups.

Meetup 1: 10th October at 100W

Registrations: 48 Attendance: 36 Companies: 14

Title: 20,000 Leagues Under MATLAB, Python and Julia: A numerical Journey

Speaker: Jorge Vieyra – Julia Lead Engineer at ASML Links: https://www.linkedin.com/in/jorge-vieyra-76280542/

Title: War stories with sensor fusion

Abstract: Practical Sensor Fusion: have you ever wondered how one combines multiple sensors from different sources, different working principles and different limitations into one single and accurate precise estimation? And what kind of application will this be useful? In this speak I will tell you the story of how a sensor fusion algorithm was used to build the foundation of a 3d scanning start-up company - all the way from a research paper into a real hardware. During the jorney of the algorithm I will share the struggle of requirememt specification, developing, testing, verifying and maintaning an algorithm. Links: https://www.linkedin.com/in/marcusdaviforte/ Speaker: Marcus Forte – System Architect @ Vanderlande / Sioux

Title: Co-Design in radio astronomy, a use case in Fourier domain dedispersion.

Abstract: The Fourier-domain dedispersion (FDD) algorithm is introduced as a new method for correcting frequency-dependent dispersion delays in radio emissions from sources like pulsars and fast radio bursts. Unlike traditional time-domain algorithms that adjust delays with time shifts, FDD uses phase rotations on Fourier-transformed data. This approach addresses the limitations of existing algorithms, which often suffer from low arithmetic intensity and memory-bandwidth constraints. Implemented on GPUs, FDD demonstrates superior performance and energy efficiency compared to both standard and optimized versions of the DEDISP software, especially when processing many trial dispersion measures (DMs). The FDD algorithm improves performance by about 20% and reduces energy consumption by 5% in high-DM scenarios. Additionally, it enhances periodicity surveys for pulsars by omitting the Fourier transform back to the time domain. Overall, FDD is well-positioned to leverage advancements in GPU technology for even greater computational gains in the future. Speaker: Steven van der Vlugt - Researcher Computer Systems at ASTRON Links: https://www.linkedin.com/in/steven-van-der-vlugt-2880223b/