Dark matter distorts the view of more distant galaxies, creating a blurred-halo effect as seen from Earth. Astronomers study dark energy by tracking this distortion through time. (NASA)

If dark matter is the glue holding galaxies together, dark energy is its doppelganger, pushing the universe apart at increasing speeds. Dark energy is thought to make up three-quarters of the universe yet its basic nature remains poorly understood.

In a new approach to cracking the dark energy puzzle, Columbia astronomers are working with computer scientists to wring more information from high-resolution images of about a billion galaxies in our universe. Their project, drawing on statistics, and computer self-learning and face-recognition algorithms, is funded by a two-year, $200,000 grant awarded by the Office of the Provost and administered by the Data Science Institute.

Though invisible, dark energy can be inferred from its effects on the shining stars and galaxies seen from telescopes on Earth and in space. In 1998 astronomers noticed that the distance between supernovas, or exploding stars, was getting bigger, faster, and coined the term dark energy to describe its cause.

After the Big Bang 13.8 billion years ago, the universe grew rapidly before gravity slowed down its expansion. Then, about six billion years ago, dark energy is thought to have mysteriously caused it to pick up speed again. By pinning down the nature of dark energy, astronomers hope to understand the ultimate fate of our universe.

For now, dark energy is studied by monitoring the night sky. The most comprehensive technique involves tracking the subtle distortions of light around distant galaxies caused by gravitational lensing. When light travels to Earth, it bends around clumps of invisible dark matter en route. By measuring the changing shape of this distortion, or shear, at different distances from Earth, astronomers can trace dark matter’s gravitational pull through time.  

More distortion, or shear, is associated with more dark matter and either less, or weaker, dark energy. In this cosmological simulation, more yellow and red shear suggests less, or weaker, dark energy. (Haiman and colleagues)

“Looking at galaxies at different distances from Earth is like traveling through time,” said Columbia astronomer Zoltán Haiman, who is leading the dark energy project. “It allows us to reconstruct the time-evolution of the dark matter clumps. If they grow bigger, rapidly over time, we can infer there is either less dark energy filling the universe, or that dark energy is weaker.”

Using a supercomputer, Haiman, with his graduate students Andrea Petri and Jia Liu, has generated nearly a hundred models for how dark energy produced the universe we see today. Computer scientist Daniel Hsu is now applying statistical analysis and image-matching techniques to those models, each containing thousands of variations, to pin down ratios for the three variables thought to be most important for defining dark energy.

The idea for the project, said Haiman, came from a discussion he had years ago with computer scientist Shree Nayar, who helped develop the technology that lets computers quickly tell male and female faces apart. Haiman wondered if similar techniques could be used to compare pictures of the evolving universe to pick out shear features most predictive of dark energy, much as the mouth and nose are features predictive of male and female human faces.

One challenge for astronomers investigating dark energy is a shortage of observational data. To identify where clumps of dark matter have formed across the universe, shear maps for a large number of galaxies at a wide range of distances from Earth are needed.

In this simulation, reduced shear indicates either more or stronger dark energy. (Haiman and colleagues)

To get around this, Haiman and his colleagues recently came up with a technique to get better dark matter estimates from limited data. Testing their method on shear shapes produced by six million galaxies—less than 1 percent of the sky–they showed that such estimates could be markedly improved.

As powerful new telescopes come online, including the Large Synoptic Survey Telescope (LSST) in Chile, an explosion of shear data is expected. Larger surveys will allow astronomers to measure the shapes of up to a billion galaxies, including those in our deep cosmic past. Data science tools like those being developed by Haiman and his colleagues will allow astronomers to extract more information from surveys large and small.

Until the mystery is solved, dark energy could be many things. “It could be a fundamental property of vacuum–the fundamental quantum property of nothing–or a new and exotic elementary particle,” said Haiman.

It might also turn out that dark energy does not exist, and Einstein’s general theory of relativity does not apply on cosmic scales.

“We may just be interpreting the data with the wrong equations,” said Haiman. “We can only solve this riddle with further analysis and measurements.” 

— Kim Martineau