Modern astronomical research is becoming more and more dependent on the power of scientific computing. Its applications range from simulations of all of the matter in the Universe evolving over time, to processing hundreds of terabytes of data taken with the Sloan Digital Sky Survey. These undertakings rely on powerful supercomputers which can operate and perform calculations millions of times faster than an average computer; but even these machines have their limitations. Supercomputers and high-performance computing clusters are immensely expensive and difficult to operate and maintain, so research collaborations have started to look towards other solutions for their computing needs. Citizen science has become increasingly popular, with volunteers using their spare time to identify circumstellar disks, map and count craters on the lunar surface, and identify transient radio sources from other galaxies.
BOINC, the Berkeley Open Infrastructure for Network Computing, is a system which harnesses the power of citizen scientists (or, their computers) to speed up scientific computing. Research projects that need computing resources can upload their jobs to BOINC, which will automatically break them up into bite-sized pieces and distribute them. Volunteers can download the BOINC client to their computer, which will automatically run these small jobs when their CPU is idle– for instance, when they step away from their computer, or when they’re only using a very small part of their computer’s processing power. On the volunteer’s side, it’s as easy as downloading the free client and picking which research projects you’d like to contribute to.
BOINC got its start in 2002, when it was developed to help manage the SETI@Home program to search for extraterrestrial signals in data from radio observatories. It very quickly ballooned in popularity, hosting dozens of projects across astronomy, mathematics, biology, and even climate science. As of November 13th, 2023, there are 39,600 active volunteers running jobs on over 132,000 computers.
Just how powerful is BOINC as a computing resource? We usually measure computing power in FLOPs: floating point operations per second, which outlines how quickly a computer can perform calculations. For reference, the average laptop can perform somewhere between 10-100 GigaFLOPs, or 10-100 billion flops. The totality of the BOINC volunteer network averages about 26 PetaFLOPs– 26 quadrillion FLOPs! Comparing it to the most powerful supercomputers as of June 2023, BOINC as a whole would rank as the 18th fastest supercomputer in the world, outdoing large expensive computing clusters like the Polaris system used by the Department of Energy.
Below, I’ll list some of my favorite BOINC projects as well as some exciting results– but this is far from a comprehensive list, and you can see all of the BOINC projects as well as all of the scientific publications which have benefited from the volunteer computing network.
- Asteroids@Home– using data from optical telescopes to try to model the shape, orbits, and rotation axis of thousands of asteroids from sparse observations
- Einstein@Home– searching for pulsar signals in gamma-ray and radio data
- Einstein@Home have discovered 94 new pulsars, including a dynamic redback pulsar system
- Milkyway@Home– creating an accurate 3D map of the Milky Way
- Universe@Home– in-depth simulations of stellar evolution, and population synthesis of many different types of stars
- These simulations are also used to study the nature of merging binary black holes
- LHC@Home– simulations to improve the quality of CERN’s Large Hadron Collider and its detectors, and analyze data from current runs
- DENIS@Home– developing electrophysiological models of the heart, understanding how it will respond to different stimuli, and helping to prevent heart disease
- GPUGrid– simulations of molecular dynamics in large systems, modeling cellular pathways involved in cancer and drug resistances that may arise, and the activation and onset of HIV/AIDS
- Rosetta@Home– simulating the complex folding process of large proteins, understanding protein binding mechanisms which may relate to disease treatment and new drug discoveries
- A Nature paper was recently published, using these data to design proteins which can bind to specific peptides, which is useful for developing targeted therapeutic drugs.
- Climateprediction.net– running large-scale climate simulations with slightly different parameters, to try to understand chaotic climate models and predict the effects of climate change
- PrimeGrid– Searching for the largest prime numbers
- The largest known prime number so far has 24 million digits!
If you’re interested in volunteering your spare CPU cycles for science, you can download the BOINC client here and sign up for projects like this. Not all projects are verified through BOINC, so be sure to follow BOINC’s advice in deciding which projects to sign up for.
Systems like BOINC can change the landscape of scientific research, as you can see from the list of results above, which were all enabled by the computing power donated by volunteers. As scientific simulations grow more complex and our datasets grow even larger in size, the power of volunteer computing and citizen science will become more and more necessary in order to answer the big questions that science tries to answer.
Edited by Lynnie Saade
Featured image credit: Clarisse Meyer
