Dark-energy equations of state and model tests

Let epsilon be dark-energy density in energy per physical volume and p its pressure, so w equals p divided by epsilon with no extra c squared. Assume separate local conservation in an FLRW background: no energy transfer to matter and no modification of the conservation equation. Expansion dilutes energy through both volume increase and pressure work. A constant w gives a power law, while a time-dependent w requires integrating its history.

The common two-parameter form w of a equals w zero plus w a times one minus a remains finite toward early times, unlike a linear function of unbounded redshift. It is still a chosen parametrization and can fail to represent transitions or rapid features. Extending it into epochs outside the observations can also imply substantial early dark energy. A prior that excludes such behavior contributes information and should be stated when combining late-time distances with microwave-background data.

For a flat late-time general-relativistic model with pressureless matter and this dark-energy component, the Friedmann equation adds matter dilution and the integrated dark-energy density. Radiation is neglected only where appropriate. The deceleration parameter measures the sign and size of the scale factor’s second time derivative. A sufficiently negative weighted pressure produces acceleration. Its condition depends on both equation of state and density fractions; negative pressure alone does not guarantee acceleration when matter dominates.

Standard candles and rulers constrain integrals of inverse expansion rate as well as radial expansion information. Their sensitivity to w is consequently entangled with matter density, curvature if allowed, and absolute calibration. Growth observables add information but require a gravity law and prescriptions for dark-energy perturbations. A background w curve does not fully specify sound speed, anisotropic stress, stability, or the perturbations of a proposed field theory. Crossing w equal to minus one particularly requires care in such physical interpretations.

A lower best-fit residual in a model with extra parameters is expected even if those parameters describe no new physics. A likelihood-ratio interpretation requires its regularity and sampling assumptions; Bayesian evidence additionally depends on prior volume. Report which datasets, nuisance treatments, and model restrictions define the comparison. Replacing one supernova compilation, ruler calibration, or curvature prior can change an apparent preference because each change modifies the inference problem.

As a dated example, the DESI 2024 analysis examined evolving-equation-of-state fits whose reported preferences depended on the chosen supernova combination and other data. That is a model-conditioned observational result, not a proof that dark energy is a particular evolving substance. We do not identify that historical analysis with the latest observational status. Stronger physical claims require consistent fits across observables, systematic cross-checks, and a perturbatively and dynamically specified theory that survives constraints beyond the background distance history.