Dark Energy Survey (2024): 1,635 Supernovae Test Dark Energy

The Dark Energy Survey supernova results, published in 2024 by the DES Collaboration, used 1,635 type Ia supernovae from five years of observations to measure how the universe has expanded. It is the largest supernova sample from a single survey, and it was the key ingredient in DESI’s strongest hint that dark energy changes over time.

The question before the paper

Type Ia supernovae, exploding white dwarfs of nearly uniform brightness, revealed in 1998 that cosmic expansion is accelerating. The Supernova Cosmology Project’s 1999 paper used 42 distant ones. Testing whether dark energy is truly constant needs far more, especially at large distances, and with consistent measurements. Traditionally every supernova had to be confirmed with a spectrum, an expensive bottleneck. Could a survey classify thousands of them from their light curves alone?

What the paper found

• 1,635 usable supernovae at redshifts 0.10 to 1.13 passed quality cuts, roughly five times as many high-quality supernovae beyond redshift 0.5 as the previous Pantheon+ compilation.
• The tightest supernova constraints to date. Combined with 194 nearby supernovae, the sample gave the strongest limits from any supernova data set.
• Dark energy close to, but not exactly, a constant. Combined with Planck’s cosmic microwave background and SDSS baryon acoustic oscillations: matter density Ω_M = 0.321 ± 0.007 and dark energy equation of state w = -0.941 ± 0.026. A pure cosmological constant has w = -1.

How they did it

The Dark Energy Survey imaged the sky repeatedly for five years and caught supernovae as they brightened and faded in four color bands. Instead of taking a spectrum of each explosion, the team classified them with a machine-learning algorithm trained on light curves. Redshifts came from dedicated spectroscopic follow-up of the supernovae’s host galaxies, which remain visible long after the explosion fades. Distances from brightness and redshifts from the hosts then trace the expansion history.

Why it mattered

The paper showed that photometric, machine-learning classification works for precision dark energy measurements. That approach is the template for the much larger supernova samples expected from the Vera C. Rubin Observatory, whose first data are described in Rubin’s Data Preview 1.

It also mattered for a specific result. In the DESI first-year analysis, combining DESI’s galaxy maps with this DES sample produced the strongest preference, 3.9 sigma, for dark energy that evolves over time.

What we know now

With DESI’s second data release in 2025, the preference for evolving dark energy ranges from 2.8 to 4.2 sigma depending on which supernova sample is used, still short of a 5-sigma discovery. Whether the supernova samples agree with each other has become part of the debate. See is dark energy weakening?

Who wrote it

The paper is signed by the DES Collaboration, with individual authors led by T. M. C. Abbott and M. Acevedo. It reports the final supernova cosmology from the survey’s full five-year data set.

Read the original

DES Collaboration, Abbott, T. M. C., Acevedo, M., et al. (2024). “The Dark Energy Survey: Cosmology Results with ~1500 New High-redshift Type Ia Supernovae Using the Full 5 yr Data Set.” The Astrophysical Journal Letters. doi:10.3847/2041-8213/ad6f9f. Free on arXiv: 2401.02929.