The BlackHoles@Home Project:

Black Hole Binaries on the Desktop Computer

Zach Etienne

Funding Acknowledgements:

PHY-1806596 (Grav theory, 2018-2021)

80NSSC18K0538 (ISFM, 2017-2020)

80NSSC18K1488 (TCAN, 2018-2021)

BH

BH

Collaborators:

Thomas Baumgarte

Ian Ruchlin

​

Website/Download:

blackholesathome.net

2

Black Hole Binaries on the Desktop: Motivation

Two black holes merge, gravitational waves detected

Waves encode important info about black holes!

How to get millions from thousands?!

• Extracting black hole info from gravitational waves is tough
• Must compare observation with millions of theoretical gravitational wave predictions
• Theoretical predictions must be based on full solutions to Einstein’s equations of general relativity
• Use computers to solve (numerical relativity)
• High computational expense (supercomputers)
• Only ~3,000 predictions generated in 14 years

Gravitational Wave

Observation

Theoretical

Prediction

3

Black Hole Binaries on the Desktop: Motivation

Two black holes merge, gravitational waves detected

Waves encode important info about black holes!

How to get millions from thousands?!

• Answer: “Interpolate” between the 3,000 predictions
• This approach has worked well with current generation of gravitational wave detectors
• Looking ahead: New grav wave detectors; better sensitivity
• Need to greatly increase the number of numerical relativity predictions; 3,000 is not enough

• Extracting black hole info from gravitational waves is tough
• Must compare observation with millions of theoretical gravitational wave predictions
• Theoretical predictions must be based on full solutions to Einstein’s equations of general relativity
• Use computers to solve (numerical relativity)
• High computational expense (supercomputers)
• Only ~3,000 predictions generated in 14 years

Theoretical

Prediction

Gravitational Wave

Observation

BlackHoles@Home: Efficient Numerical Grids

Exploit near-symmetries ➔ Minimize sampling points

~20x fewer points for black hole binaries than AMR

Supercomputer ➔ Consumer-grade desktop!

4

BH

BH

AMR

Adaptive Mesh Refinement

(Most Popular Method in

Numerical Relativity)

BH

BH

Super-Efficient

BiSphere Grids

Proof of Concept:

Black Holes Collision using BiSphere Grids

5

6

Finding from BH collision test:

Numerical errors small and

converge to zero at expected rate

Post-merger num error,

BH from x = -0.5 to +0.5

7

Summary,

Current Status

BH

BH

• Black hole binaries can fit on desktop now. (About 6GB RAM up to merger; then 3GB.)
• Working to make software fast enough
• Ongoing work by at least 3 groups, including my own. Multiple solutions.
• Goal: 3-6 months/waveform/desktop
• Catalog of ~20,000 gravitational waveforms in first year
• Goal: BlackHoles@Home by EoY!
• Stay tuned: Sign up for our newsletter:

blackholesathome.net