What specific gaps exist between cosmology simulations and real observational astronomy data?

# Gaps Between Cosmology Simulations and Observations

The cosmological simulation field faces several significant mismatches with real data—some well-characterized, others emerging only recently.

Observable Tensions in the Standard Model

The most pressing gap is a *direct conflict between different cosmological measurements* rather than between simulations and any single dataset [AG-2025.04-033]. The standard ΛCDM model, tested against diverse probes (cosmic microwave background, galaxy surveys, supernovae, baryon acoustic oscillations), shows "discordances primarily resting on divergence in the measurement of core cosmological parameters." These tensions may stem partly from *systematic errors in the observations themselves*, but they also suggest that simulations calibrated to one dataset may fail on others—a hint that the underlying physics or measurement methods need revision [AG-2025.04-033].

Simulation-to-Simulation Inconsistency

A deeper structural problem: even when different simulation codes (Abacus, Gadget, Gizmo, Ramses, etc.) are given identical starting conditions, they produce *measurably different matter fields* [AG-2025.05-214]. Using field-level comparison—comparing the actual spatial structure, not just summary statistics like the power spectrum—researchers found that "resolution effects, such as those arising from adaptive mesh refinement, have a significant impact" on whether one simulation can reliably infer cosmological parameters from another [AG-2025.05-214]. This means simulations disagree on small scales in ways that *matter for precision cosmology*, yet these discrepancies are often hidden when only power spectra are compared.

Baryonic Physics in Galaxy Clusters

Simulations like IllustrisTNG reproduce the *scaling relations* of galaxy clusters reasonably well (within 1–2σ of observations), but they misplace roughly 14% of the expected baryonic mass, with this "missing" material appearing as intracluster light and warm gas in unexpected locations [AG-2025.04-410]. The gap is subtle—the simulations match observations statistically—but the *detailed spatial distribution of baryons diverges*, suggesting that the sub-grid models for stellar feedback and hot gas cooling are incomplete [AG-2025.04-410].

Underexplored Cosmological Regimes

Cosmic voids—the vast low-density regions between galaxy clusters—encode information about neutrino masses and gravity itself [AG-2025.03-414], yet simulations and observations of void properties remain sparsely compared. The challenge is that voids are diffuse and hard to observe cleanly, so simulations trained on high-density regions may perform poorly here. Similarly, tensions between different dark energy probes (supernovae, BAO, quasars, CMB) suggest that simulations built around one cosmology may not correctly predict structure in alternative models [AG-2025.06-678].

The Bottom Line

The real gap isn't a single discrepancy but a *layered set of incompatibilities*: observations disagree with each other; simulations disagree with observations *and with each other*; and the sub-resolution physics needed to connect small-scale processes (star formation, black hole growth) to large-scale structure remains poorly constrained. Modern precision cosmology now has the data to expose all three problems simultaneously [AG-2025.12-585].