Dark energy explorer
Three models of dark energy: a cosmological constant (ΛCDM), a constant equation of state w (wCDM), and one that changes with time, w(z) = w0 + wa z/(1 + z) (CPL). The charts come in three groups: dark energy itself; the shapes of distances, compared with a cosmological constant with the same matter density and Hubble constant, so that H0 cancels; and absolute scales, compared with the Planck 2018 ΛCDM fit held fixed, where H0 matters. Each chart shows what one kind of measurement would see. The page opens on the DESI DR2 w0wa fit; it is a published fit to explore, not a claim about the true cosmology.
1. Dark energy itself
2. Shapes, where H0 cancels
Both charts compare the model with ΛCDM with the same Ωm and H0, and H0 cancels in both, so moving it changes nothing here.
3. Absolute scales, where H0 matters
Each chart compares the model (solid) with the Planck 2018 ΛCDM fit, held fixed, and with the DESI DR2 fit for the selected model; both fits are in the table below. Lensing charts use the lens at zl = 0.348.
Its minimum (dot) is where the expansion starts to accelerate. BAO and cosmic chronometers measure H(z).
What a distance indicator with an independently known zero-point sees. Without that calibration the constant offset cannot be told apart from the zero-point, which is why supernovae alone do not measure H0.
What the time delays between the images of a variable source (a lensed quasar, AGN or supernova) measure; the delays also depend on the lens mass model.
- DΔt, source at z1
- –
- Relative to Planck ΛCDM
- –
- A 100-day delay in Planck ΛCDM is here
- –
A lens model fits the convergence Σ/Σcr, so the masses it gives scale with the assumed cosmology as Σcr Dl², which is ∝ 1/H0.
- Σcr, source at z1
- –
- Mass inside θE = 30″
- –
- Change from Planck ΛCDM
- –
What this shows
Dark energy's equation of state w is the ratio of its pressure to its energy density. A cosmological constant has w = −1 at all times, so its density never changes. With w > −1 the density falls as the Universe expands, so it was higher in the past; with w < −1 (phantom) it grows. Crossing w = −1, the phantom divide, is allowed in this description, though a single canonical scalar field cannot do it.
The expansion rate H(z) sets every distance. Standard candles such as type Ia supernovae measure the luminosity distance through the distance modulus μ; the Δμ chart in the second group shows how much brighter (negative Δμ) or fainter (positive) a source would look in this model than with a cosmological constant with the same Ωm and H0. A difference of a few hundredths of a magnitude is what current supernova samples can detect when averaged over many objects.
A strong lens deflects light by an angle proportional to Ξ = Dls/Ds. For one cluster with lensed galaxies at two redshifts, the ratio of lensing efficiencies Ξ(z1)/Ξ(z2) depends only on the cosmology, but measuring it requires fitting the cluster's mass model at the same time, and degeneracies between the mass model and the cosmology limit the precision. The effect is small: at most about one per cent for the published fits marked on the sliders (up to 1.0 per cent for the DESI DR2 CPL fit when one source sits just behind the lens, at z = 0.4; the same fit shifts the ratio by −0.11 per cent for the default source redshifts), though it grows at extreme slider settings, to about 19 per cent at the corners of the ranges. It is a ratio of distances, so it does not depend on H0.
In the first two groups H0 cancels: w(z) and the dark energy density do not involve it, and Δμ and Ξ compare distances with those of a cosmological constant with the same H0, so every distance scales by the same factor c/H0. The third group compares absolute quantities with one fixed reference, the Planck 2018 ΛCDM fit, so H0 matters there. The expansion rate H(z)/(1 + z) is the rate of change of the scale factor, ȧ; it falls while the expansion decelerates and rises once it accelerates, so its minimum marks the onset of acceleration. BAO and cosmic chronometers measure H(z). The absolute Hubble diagram is what a distance indicator with an independently known zero-point sees: H0 shifts it by a constant 5 log10(H0,Planck/H0) on top of the difference in shape from Planck ΛCDM, which is the second group's Δμ only when Ωm equals Planck's 0.3153. Without such a calibration the constant cannot be told apart from the zero-point, which is why supernovae alone do not measure H0.
The two lensing charts of the third group use the same lens. When a variable source, such as a lensed quasar, AGN or supernova, is imaged several times, the delays between the images are DΔt/c times a difference of Fermat potentials set by the lens mass model; for a fixed lens model they scale with DΔt, which is ∝ 1/H0 (Refsdal 1964). The delays also depend on the mass model, which is the main limitation of the method. The critical surface density Σcr converts the convergence a lens model fits into a surface mass density, so the mass a lens model gives inside a fixed Einstein radius, M = π(θEDl)²Σcr, depends on the assumed cosmology as DlDs/Dls, which is ∝ 1/H0.
Model and assumptions
- A flat universe with matter and dark energy only. Radiation is left out: with Ωr ≈ 9 × 10−5 it would change distances by less than 0.05 per cent out to z = 10, and the expansion rate itself by 0.17 per cent at z = 10.
- The dark energy equation of state is w(a) = w0 + wa(1 − a), with a = 1/(1 + z) the scale factor (Chevallier & Polarski 2001; Linder 2003). w0 is its value today and w0 + wa its value in the distant past. Then H(z)² = H0²[Ωm(1 + z)³ + (1 − Ωm)(1 + z)3(1 + w0 + wa) e−3waz/(1 + z)].
- In the first two groups the dashed reference is a cosmological constant with the same Ωm and H0, in every model. In the ΛCDM model it is the model itself, so the Δμ and Ξ charts are flat at zero, and Ωm and H0 show only in the age, q0, the onset of acceleration and the third group. Real fits do not work this way: when w0 and wa change, Ωm and H0 shift too, to keep matching the best-measured distances, so the differences shown here are larger than those between models that fit the same data equally well. Compare the published fits below: each model's fit has its own Ωm and H0.
- In the third group the references are fixed: the Planck 2018 ΛCDM fit (dashed) and the DESI DR2 fit for the selected model (dotted): DESI + CMB for ΛCDM, DESI + CMB + DESY5 for wCDM and CPL. The charts show H(z)/(1 + z) = H0E(z)/(1 + z) in km/s/Mpc; μ − μPlanck ΛCDM, which tends to 5 log10(H0,Planck/H0) as z → 0; the time-delay distance DΔt = (1 + zl)DlDs/Dls; and the critical surface density Σcr = c²Ds/(4πG DlDls), the last two as ratios to Planck ΛCDM for a lens at zl = 0.348 against source redshift. All D are angular-diameter distances. The readouts give DΔt and Σcr for the source at z1, the delay that a 100-day delay in Planck ΛCDM becomes (100 days times the DΔt ratio, for the same lens model), and the mass inside θE = 30″, M = π(θEDl)²Σcr = c²θE²DlDs/(4G Dls), with G = 6.6743 × 10−11 m³ kg−1 s−2, 1 Mpc = 3.0857 × 1022 m and M☉ = 1.989 × 1030 kg. When only H0 changes, DΔt scales exactly as 1/H0, Σcr as H0 and the mass as 1/H0.
- The three models differ only in which of w0 and wa may vary. ΛCDM fixes w0 = −1 and wa = 0, wCDM frees w0 and keeps wa = 0, and CPL frees both. A slider the model does not let vary is greyed out and tagged "fixed"; each model remembers its own w0 and wa when you switch away and back, while Ωm and H0 are shared.
- The deceleration parameter is q = ½[1 + 3w(z)ΩDE(z)]; acceleration began where q first turns negative going forward in time. Some choices of w0 and wa give no such redshift (always decelerating, or accelerating since before z = 20) or more than one.
- Distances are integrated numerically (Gauss–Legendre quadrature, relative error below 10−9), in
src/lib/cpl.ts.
The published fits
The marks under the sliders and the Planck 2018 and DESI DR2 buttons use these values, for the model selected. All are marginalised posterior means with 68 per cent intervals, as the papers state them; values in brackets are fixed by the model.
| Model | Data | Ωm | H0 (km/s/Mpc) | w0 | wa | Source |
|---|---|---|---|---|---|---|
| ΛCDM | Planck TT,TE,EE + lowE + lensing | 0.3153 ± 0.0073 | 67.36 ± 0.54 | (−1) | (0) | Planck VI, Table 2 |
| ΛCDM | DESI DR2 BAO + CMB | 0.3027 ± 0.0036 | 68.17 ± 0.28 | (−1) | (0) | DESI DR2, Table 5 |
| wCDM | Planck TT,TE,EE + lowE + lensing + SNe + BAO | not given | not given | −1.028 ± 0.031 | (0) | Planck VI, Eq. 50 |
| wCDM | DESI DR2 BAO + CMB + DESY5 | 0.3098 ± 0.0050 | 67.34 ± 0.54 | −0.971 ± 0.021 | (0) | DESI DR2, Table 5 |
| CPL | Planck TT,TE,EE + lowE + lensing + SNe + BAO | not given | 68.31 ± 0.82 | −0.957 ± 0.080 | −0.29 (+0.32, −0.26) | Planck VI, Table 6 |
| CPL | DESI DR2 BAO + CMB + DESY5 | 0.3191 ± 0.0056 | 66.74 ± 0.56 | −0.752 ± 0.057 | −0.86 (+0.23, −0.20) | DESI DR2, Table 5 |
- Planck's SNe are the Pantheon sample and its BAO are 6dFGS, SDSS MGS and BOSS DR12. Planck alone constrains w poorly, so its extended-model results combine it with these. The paper does not tabulate Ωm for those fits, nor H0 for the constant-w one, so those sliders have no Planck mark in those models and the Planck button leaves them as they are.
- DESI's CMB is Planck temperature and polarisation (PR4 CamSpec at ℓ ≥ 30, with Commander and SimAll at low ℓ) plus the combined Planck and ACT DR6 lensing, and DESY5 is the DES 5-year supernova sample. Its ΛCDM results do not include supernovae, so the ΛCDM entry is DESI + CMB; DESI BAO alone gives Ωm = 0.2975 ± 0.0086 and hrd = 101.54 ± 0.73 Mpc, which does not fix H0 without the sound horizon rd. Combining DESI + CMB with Pantheon+ or Union3 supernovae instead of DESY5 gives (w0, wa) = (−0.838, −0.62) or (−0.667, −1.09) in CPL, and w = −0.995 or −0.997 in wCDM.
- These are the means of each parameter's marginal posterior, not the best-fitting point of the fit; with correlated parameters the two can differ.
References
- Chevallier, M. & Polarski, D. 2001, Int. J. Mod. Phys. D, 10, 213
- Linder, E. V. 2003, Phys. Rev. Lett., 90, 091301
- Refsdal, S. 1964, MNRAS, 128, 307
- Planck Collaboration VI 2020, Planck 2018 results. VI. Cosmological parameters, A&A, 641, A6, arXiv:1807.06209
- DESI Collaboration 2025, DESI DR2 Results II: Measurements of Baryon Acoustic Oscillations and Cosmological Constraints, arXiv:2503.14738