Results from the DESI (Dark Energy Spectroscopic Instrument) YR1 Data Release: a summary

Title: DESI 2024 VI: Cosmological Constraints from the Measurements of Baryon Acoustic Oscillations

Authors: The DESI Collaboration, A. G. Adame et al 

First author institution: Lawrence Berkeley National Laboratory, Berkeley, California 

Status: submitted to JCAP (preprint on ArXiv) 

Figure 1: Lower image: one of the eight petals that makes up part of the DESI focal plane (a plane perpendicular to the focus of light rays where images appear) showing the robots that control each optical fibre. Top image: the DESI instrument installed on the Mayall Telescope.

The Dark Energy Spectroscopic Instrument (DESI) is a robotic instrument and spectrograph mounted on the Mayall Telescope in Kitt Peak, Arizona. The DESI collaboration aims primarily to understand the elusive Dark Energy. This is an energy of unknown source causing the Universe to accelerate in its expansion; this accelerating expansion is not predicted to occur for a universe that is filled with just ordinary matter and radiation (some more detail can be seen in this Astrobite). Since we still know so little about Dark Energy, a large galaxy survey can allow us to explore the history of the expansion of the Universe in more detail. The DESI instrument has 5000 individual optical fibres controlled by robots that allow it to measure individual spectra of up to 5000 galaxies in just a mere 20 minutes! Due to this design, and an observing program that optimises targets in the sky based on observing conditions, the survey will measure spectra of up to 35 million galaxies over 5 years. This will allow DESI to perform precise cosmological measurements, as a great volume of space and number of galaxies can be probed, and noise in the data products is reduced. This bite looks at the cosmology results from the collaboration’s analysis of the recently released Year 1 Data (YR1), in particular, via a signal that can be seen in the data known as Baryon Acoustic Oscillations.

DESI tracers 

For the cosmological results in this work DESI uses information from various different ‘tracers’ – galaxies that trace the Large Scale Structure of the Universe. These consist of low redshift bright galaxies (BGS) that are measured when the moon lights the sky (and thus dimmer galaxies are less visible), and higher redshift galaxies measured during the dark time. The dimmer objects include luminous red galaxies (LRGs) which are elliptical galaxies that are extremely bright, emission line galaxies (ELGs) which are younger galaxies with emission line features in their spectra, and quasars (QSOs) which are very distant and bright galaxies that contain active galactic nuclei. The sample used also includes QSOs detected using Lyman-alpha forest measurements, or a method of tracing matter that utilises a series of absorption lines detected due to light from distant QSOs passing through neutral hydrogen in the space between us and the distant galaxies. 

Baryon Acoustic Oscillations analysis 

Baryon Acoustic Oscillations (or BAOs) refer to sound waves that propagated through the early Universe. These sound waves left an imprint in the distribution of matter in the Universe that seeded the growth of galaxies which led to the Large Scale Structure that we see today, in the form of an excess probability of seeing galaxies separated by a distance called the sound horizon scale – the furthest distance the sound waves could travel in the early Universe. If we look at the probability of finding galaxies separated at different distance scales, we can see there is increased probability at a distance corresponding to this sound horizon, which can essentially be used as a giant ruler (see this excellent visualisation). 

To detect the BAO signal, the authors measure two quantities from the observed galaxy distribution of each tracer; 1) the preferred separation of galaxies in the sky from their angles (by looking at the distance between objects at separations transverse to our line-of-sight in the sky) and 2) the preferred separation of distances between objects in the sky along the line-of-sight, from there redshifts. 

A preferred angular separation of galaxies in the sky is a BAO signal, and the angle corresponds to a ratio which is a physical distance to those galaxies in the sky and the sound horizon scale. A preferred separation in redshift between galaxies along the line-of-sight is also a BAO signal, and the separation \Delta z corresponds to a ratio that is the expansion rate of the Universe known as the Hubble constant and the sound horizon scale. 

We can compare these ratios to the ratios we expect from a theoretical model, allowing us to test theories of cosmology, which includes the Universe’s expansion history. 

Cosmology results 

The DESI collaboration uses the BAO data to test the most generally accepted cosmological model, \LambdaCDM, in which the Universe contains cold dark matter (CDM) and a cosmological constant Dark Energy density \Lambda, which is a Dark Energy component with an equation of state (EOS) w = -1. For  this kind of Dark Energy, its energy density doesn’t change with the expansion of space or with time. They also consider extensions in which the Dark Energy might actually vary (not a cosmological constant), either by having an EOS w that is either not strictly given by w = -1 (w can vary) or has a time dependence, w = w_0 + w_a(1-a). Here, a is the scale factor of the Universe and is thus a proxy for time.

Figure 2 shows model predictions for the BAO distance measurements vs the best fits to the BAO scales from the different tracers, in the context of the \LambdaCDM model. The model is able to provide a good fit to the data, although the orange point in Figure 2 is slightly discrepant – the authors believe that this is likely a fluctuation, but this may be confirmed with future DESI data. Combining the DESI BAO data with Cosmic Microwave Background (CMB) data allows a constraint on the Hubble constant, and they find H_0 = 67.97 \pm 0.38 km/s/Mpc in \LambdaCDM. 

Figure 2: Figure 1 from the paper, showing an angle-averaged distance to sound horizon ratio from the different tracers on the left, and the ratio of the distances measured along the line-of-sight and parallel to the line of sight, on the right plot. The best fit to the standard \LambdaCDM model from the DESI data (black solid line) and from CMB data (dashed line) are shown. The bottom panel shows the ratio of the data to the fit to the DESI data.

Time-varying dark-energy: The authors also test the case where Dark Energy varies with time (models with a varying EOS from w = -1 for Dark Energy). The DESI results alone or in combination with external data sets find results consistent with a cosmological constant. This leads us to the most interesting result, which comes from testing a more complex model with a time-varying EOS; in this case, finding w_0 consistent with -1.0 and w_a consistent with zero would suggest Dark Energy is a cosmological constant. But in this case, the DESI data alone or in combination with external data sets, prefers w_0 greater than -1 and w_a less than 0.0, see Figure 3. The combinations of datasets for DESI BAOs + CMB data have a statistical preference for the Dark Energy model with a time-varying EOS compared to the standard \LambdaCDM model. There is some moderate preference for the time-varying EOS models when including DESYR5 data in an analysis; the preference is weaker for other datasets explored.

Figure 3: Figure 6 from the paper, showing the best fitting regions for the parameters w_0 and w_a from the DESI results alone (grey dashed lines in the left panel) and combined with various external datasets. The right panel shows more constraining combinations of the data, where a discrepancy with a cosmological constant dark energy is clear (the centre of the crosshairs shows the region in the space consistent with a cosmological constant). 

Neutrinos: Cosmological data also has the constraining power to measure the masses of neutrinos! Neutrinos are a somewhat misunderstood particle in the standard model. They behave like light in the early Universe, but more like dark matter later, and leave an imprint in the structure formation history of the Universe by washing out small-scale structure formation, in a way that depends on the individual masses of each kind of neutrino (which is not actually known!) However, thanks to a phenomenon known as neutrino oscillations, neutrinos are known to have a non-zero mass (at the minimum the sum of masses for the three kinds is \approx 0.06 eV.) Due to the fact neutrinos have such an impact on the growth of structure, cosmology can provide upper bounds on the sum of neutrino masses by probing the galaxy distribution in the Universe. 

While BAOs don’t strongly constrain the effect of massive neutrinos on structure growth (a different kind of data product is required), the presence of massive neutrinos do have an effect on the Universe’s expansion history, which is constrained by BAOs. CMB data is sensitive to both effects, so the combination of CMB data with DESI BAOs data can allow us to constrain neutrino mass better than before. In the context of the \Lambda CDM model, the data tells us that the sum of neutrino masses is less than 0.072 eV at 95% confidence. Interestingly, there also seems to be a peak for the likelihood for the sum of neutrino masses at zero! This is an unexpected result which might require some more research to understand and explain. However, the authors note this preference for low neutrino mass could be driven by the fact that DESI and CMB data have some tension in the preferred value of the Hubble constant (DESI prefers a slightly higher value) which has some similar effects on the data to varying neutrino masses.


The results of the DESI BAOs have left us with some interesting questions to be addressed in the future data releases. Will future data still have a preference for Dark Energy with a time-varying EOS or is this just a fluctuation that will go away with more data? It is interesting to note that the DESYR5 data and an analysis of a compilation of supernova datasets found similar hints of time-varying Dark Energy independently. Will we continue to find that the likelihood for the sum of neutrino masses peaks at a 0 eV, suggested new physics, or are the results driven by the tensions between the combined datasets? We can only wait with excitement for the next science results from future DESI releases. 

Disclaimer: Abbé Whitford is a member of the DESI Collaboration, which has over 1000 members, however is not an author of this work. I can’t help but want to write about the results in DESI because they are just incredibly exciting! 

Edited by: Jessie Thwaites and Lucas Brown

Featured image credit: DESI Collaboration/KPNO/NOIRLab/NSF/AURA/P. Horálek/R. Proctor, CCA by 4.0, via Wikimedia Commons

About Abbé Whitford

I am a third year PhD student at the University of Queensland, studying Large Scale Structure cosmology with galaxy clustering and peculiar velocities, and using Large Scale Structure to measure the properties of neutrinos. In my spare time I like to train BJJ, play with pets, and I enjoy travelling.

Leave a Reply