Origin of the discrepancy between simulated and observed halo-mass estimates

Determine whether the discrepancy between halo-mass predictions trained on TNG100 galaxy properties and observational halo-mass estimates arises from the predictive analysis or from observational measurement errors and galaxy-classification uncertainties.

Background

The model trained on TNG100 systematically underpredicts halo masses for part of the observational comparison sample. Differences between simulated and observed stellar-mass, color, and other galaxy-property distributions may contribute to this offset, while observational halo-mass measurements and central-galaxy classifications also carry uncertainties. The paper leaves unresolved whether the discrepancy is caused primarily by the analysis and simulation-trained model or by errors in the observational data.

References

The observed masses could have uncertainties in their measurements that lead to an over-predicted halo mass as well; currently, it is unclear whether the discrepancy between our prediction and the observed masses are because of our analyses, or potential observational errors themselves.

— Inferring the Dark from the Observable: Estimating Halo Masses Using Galaxy Properties  (2608.19154 - Chen et al., 19 Aug 2026) in Section 4.3, “Predictions with Observational Data for Groups and Clusters” (label: subsection not explicitly numbered; section label: discussion)