Colin Hill - Early dark energy doesn't make cosmology concordant again

Colin tells us about how even though early dark energy can alleviate the Hubble tension, it does so at the expense of increasing other tension. Early dark energy can raise the predicted expansion rate inferred from the cosmic microwave background (CMB), by changing the sound horizon at the last scattering surface. However, the early dark energy also suppresses the growth of perturbations that are within the horizon while it is active. This mean that, to fit the CMB, the matter density must increase (and the spectral index becomes more blue tilted). The consequence is that the matter power spectrum should get bigger.

In their paper, Colin and his coauthors show that this affects the weak lensing measurements by DES, KiDS and HSC, and therefore including those experiments in a full data analysis makes things discordant again. The Hubble parameter is pulled back down, restoring most of the tension between local and CMB measurements of H0, and the tension in S_8 gets magnified by the increased mismatch in the predicted and measured matter power spectrum.

It is also worth noting that, if you exclude the local measurements of H0, there is no preference for early dark energy in the data.

There is hope, perhaps. If the sound horizon could be changed without altering the growth of perturbations that might still be a valid resolution, but it is unlikely to be caused by early dark energy (alone).

[talks_curator]

*Index to Key Parts of the Talk:*
[00:02] Shaun's intro to the topic and Colin's work
[01:15] Colin's summary of the 2003.07355 paper
[04:52] The broader context to Colin's motivation for this work
[08:16] First slide: The Hubble Constant
[08:54] Early Dark Energy: earlier EDE papers; summary plot; the chain of reasoning leading to the EDE models with a reference to Knox and Millea for a nice review, 1908.03663; EDE models decrease the physical size of the sound horizon imprinted in the CMB - they have added flexibility due to a set of 3 additional new parameters that are relevant in the calculation of the physical size of the sound horizon
[12:27] How the EDE model has been implemented from a more physical perspective: a new (pseudo)-scalar field important prior to recombination, whose energy density contribution decays away rapidly just before the redshift of last scattering; the mass of the scalar field needs to be ~ 10 ^-27 eV
[17:36] The actual parameters used to describe this model: 3 physical parameters - mass of the field, a decay constant, and its initial position on the potential; these are converted into phenomenological parameters which are more closely related to what the data can actually constrain; plot showing fractional contribution of EDE to cosmic energy budget as function of z
[20:05] EDE maintains a good fit to CMB power spectrum data with higher H0: plot shows a sub-percent difference in the TT power spectrum between ΛCDM (H0=68.21) and EDE model (H0=72.19)
[21:40] What about large-scale structure?: motivation for the project - no one had made a plot of the matter power spectrum P(k); modified version of the Boltzmann code, CLASS_EDE, is public on github
[22:26] Plot of the matter power spectrum comparing ΛCDM and EDE models: the plot of their ratio highlights the significant differences
[24:19] Parameter Shifts are the driver for these differences: caused by parameter shifts in the so-called normal cosmological parameters in the EDE scenario compared to ΛCDM; also a noticeable shift in the scalar spectral index
[27:38] Interesting physical effects due to the EDE itself
[30:28] Updated EDE Analysis Including Large-Scale Structure Data Sets: includes 2 data sets not included in previous work: DES "3x2pt" full likelihood and S8 constraints from HSC and KiDS; Cobaya code used for the MCMC sampling
[33:08] Inclusion of LSS Data
[35:07] Inclusion of DES/HSC/KiDS: inclusion of LSS data leads to non-detection of EDE component
[36:28] Analysis without SH0ES: comparing all data sets without SHOES; looking at the posteriors in the EDE parameter space; broadening of error bars when analyzing EDE model, but no major shifts seen; H0 tension persists; strong upper limit on existence of EDE component; SH0ES is in 3.5σ tension with other data sets (even in EDE model); bottom line: you can't shift the parameters in the way that's needed for the CMB without messing up the large-scale structure data
[38:42] Summary: SHOES is the only data set driving preference for EDE (moderate evidence, ~ 2σ)
[40:43] Physical Priors: uniform priors on f_EDE and log(z_c) are very non-uniform on physical scalar field parameters f and m
[41:34] Final Summary:
+ No evidence for EDE component seen in CMB-only or CMD+LSS data
+ SHOES constraint is in tension, even in this model
+ Basic problem: higher H0 requires higher f_EDE, which increases σ8 and hence worsens fit to LSS data
+ In short: EDE model does not restore concordance
+ Use of physical priors (on scalar field parameters) further weakens evidence for EDE
+ Follow-up work in progress: (1) validate the BAO and RSD results here using the effective field theory of LSS; (2) demonstrate that the "all data except SHOES" analysis indeed would detect EDE if it were present in the universe
+ Theorists: back to the drawing board
[42:45] Q&A: Where to next?: upcoming release of ACT cosmological parameter measurements in the next couple months; the observables that dominate the constraining power are different between ACT (TE) and Planck (TT).
[44:19] Q&A discussion on the sound horizon issue): to be consistent with LSS, the sound horizon would need to be changed (lowered) in a way that does not then act to suppress the growth of perturbations
[45:49] Q&A: what do you think is the most probable explanation of H0 tension?
[48:13] Q&A: Outside of your own research, what do you think is the most interesting thing in cosmology at the moment?

*Other Paper References:*