EDE Playground

Dark energy through cosmic history

Drag any of the 32 points up or down to add or remove dark energy at that redshift. Everything else on the page follows.

Amplitude of each component, as a share of the ΛCDM density at its redshift

Each component is constant until its redshift and then decays quickly. You can also click in a column, or focus a point and use the arrow keys.

The result: fraction of the total density in dark energy

your model posterior of the paper, 68% and 95%

Against the data, relative to the ΛCDM best fit

your model Planck ACT DR6 DESI, Pantheon+ ΛCDM best fit

Temperature, TT

Change in the power spectrum. Bandpowers are combined for display.

Cross-spectrum, TE

Difference divided by √(TT × EE).

Polarisation, EE

Change in the power spectrum.

CMB lensing

Change in the lensing potential power.

BAO distances

Solid and filled: D_M/r_d. Dashed and open: D_H/r_d. Diamond: D_V/r_d.

Supernova distances

Distance modulus in magnitudes, binned, calibrated by the SH0ES Cepheid hosts.

How to read this

The model is the one of Moss, Copeland, Bamford and Clarke (arXiv:2109.14848): 32 dark energy components at fixed redshifts in one fluid with sound speed c_s². Every curve and every χ² here comes from a neural-network emulator of the Boltzmann code CAMB, running in your browser, with the same likelihoods as the paper. The two calibration parameters are set to their best values for each model.

The emulator was trained where the paper's chains go. Near the presets it reproduces the exact χ² to about 1. Far from them it can be badly wrong even inside its range, so treat a surprisingly good fit with suspicion.

With the acoustic scale held fixed, H₀ and ω_c follow your edits so that the angular size of the sound horizon and r_dh stay where they are. Without it, raising one component mostly just shifts the acoustic peaks. The polarisation efficiency of ACT enters χ² but is not drawn in the plots.