The Nearly Universal Disk Galaxy Rotation Curve

Raj Patel, Nikhil Arora, Stéphane Courteau, Connor Stone, Matthew Frosst, Lawrence M. Widrow

Abstract

The Universal Rotation Curve (URC) of disk galaxies was originally proposed to predict the shape and amplitude of any rotation curve (RC) based solely on photometric data. Here, the URC is investigated with an extensive set of spatially-resolved rotation curves drawn from the PROBES-I, PROBES-II, and MaNGA data bases with matching multi-band surface brightness profiles from the DESI-LIS and WISE surveys for 3,846 disk galaxies. Common URC formulations fail to achieve an adequate level of accuracy to qualify as truly universal over fully sampled RCs. We develop neural network (NN) equivalents for the proposed URCs which predict RCs with higher accuracy, showing that URC inaccuracies are not due to insufficient data but rather non-optimal formulations or sampling effects. This conclusion remains even if the total RC sample is pruned for symmetry. The latest URC prescriptions and their NN equivalents trained on our sub-sample of 579 disk galaxies with symmetric RCs perform similarly to the URC/NN trained on the complete data sample. We conclude that a URC with an acceptable level of accuracy ($\Delta V_{\rm circ} \lesssim 15$ per cent) at all radii would require a detailed modelling of a galaxy's central regions and outskirts (e.g., for baryonic effects leading to contraction or expansion of any dark-matter-only halo).

Introduction

The rotation curve (RC) of a galaxy—the orbital velocity of its stars and gas as a function of distance from the center—is a fundamental tool in astrophysics. Historically, RCs provided the first robust evidence for the existence of dark matter, as the observed velocities in the outer regions of galaxies far exceed what would be expected from visible matter alone. The "Universal Rotation Curve" (URC) hypothesis posits that a galaxy's entire rotation profile can be predicted using only its photometric properties, such as luminosity and size. If a truly universal relationship exists, it would allow for the determination of mass distributions in distant galaxies where high-resolution spectroscopy is unavailable. However, previous URC models have often relied on limited or biased samples, leaving their "universality" in question.

Data and Sample Selection

This study utilizes an extensive compilation of spatially-resolved data from several major surveys. The primary sources for rotation curves are the PROBES-I and PROBES-II databases, supplemented by 2D velocity maps from the MaNGA (Mapping Nearby Galaxies at Apache Point Observatory) survey. To provide the necessary photometric context, multi-band surface brightness profiles were obtained from the DESI Legacy Imaging Surveys (LIS) and the WISE satellite.

The final sample consists of 3,846 disk galaxies, representing a diverse range of luminosities and morphologies. To ensure the highest quality results, a "gold standard" sub-sample of 579 galaxies was identified, pruned for high degrees of rotational symmetry. The reliability of the stellar mass estimates derived from this multi-band photometry is crucial for the URC analysis.

Comparison of stellar mass estimates from Pipe3D versus the values used in this study, showing strong consistency across the sample.

Methodology

The analytical pipeline involved several critical steps to ensure a fair comparison between traditional models and machine learning benchmarks. First, raw rotation curves and surface brightness profiles were smoothed and truncated to minimize noise from local perturbations like spiral arms. A key component of the analysis was the bulge-disk decomposition, which separates the galaxy into its central bulge and outer disk components. This is essential because the mass distribution of the bulge significantly influences the rotation curve in the inner regions.

Bulge-to-total (B/T) ratios across different Hubble types for the various galaxy samples used in the study, highlighting the morphological diversity of the dataset.

The researchers tested two prominent URC formulations from the literature: the classic PSS96 model and the more recent LSB-D (Low Surface Brightness-Dwarf) model. To evaluate whether these models are limited by the data or their own mathematical structure, neural network (NN) equivalents were developed. According to the Universal Approximation Theorem, an NN can find the optimal mapping between inputs and outputs. If an NN outperforms a traditional formula using the same input data, it indicates that the formula itself is mathematically non-optimal.

Key Results and Insights

The comparison revealed a significant "math gap." Neural networks consistently outperformed traditional URC formulas, achieving higher accuracy and lower systematic bias. This suggests that the information required to predict rotation is indeed present in the photometric data, but current physics-based equations fail to capture the full complexity of the relationship.

Interestingly, the models performed just as poorly on the symmetric "gold standard" sub-sample as they did on the full dataset. This implies that the failure of URC models is not merely due to "messy" or asymmetric galaxies, but is a fundamental issue with the model prescriptions. The study also highlighted the critical role of the bulge; models that ignored the bulge component significantly overestimated velocities in the central regions of galaxies.

Conclusion

While the dream of a truly universal rotation curve remains elusive with current formulas, this study shows that a "nearly universal" curve is achievable with more sophisticated modeling. The authors suggest that future URC prescriptions must incorporate detailed bulge components and better account for the interaction between baryons and dark matter, such as adiabatic contraction. Furthermore, extending the data to include neutral hydrogen (HI) observations could help resolve discrepancies in the outer regions of galaxies. Ultimately, the transition from simple analytical formulas to more flexible machine learning models may be necessary to achieve the desired accuracy of $\Delta V_{\rm circ} \lesssim 15$ per cent across all radii.

Summary updated

Abstract

The Universal Rotation Curve (URC) of disk galaxies was originally proposed to predict the shape and amplitude of any rotation curve (RC) based solely on photometric data. Here, the URC is investigated with an extensive set of spatially-resolved rotation curves drawn from the PROBES-I, PROBES-II, and MaNGA data bases with matching multi-band surface brightness profiles from the DESI-LIS and WISE surveys for 3,846 disk galaxies. Common URC formulations fail to achieve an adequate level of accuracy to qualify as truly universal over fully sampled RCs. We develop neural network (NN) equivalents for the proposed URCs which predict RCs with higher accuracy, showing that URC inaccuracies are not due to insufficient data but rather non-optimal formulations or sampling effects. This conclusion remains even if the total RC sample is pruned for symmetry. The latest URC prescriptions and their NN equivalents trained on our sub-sample of 579 disk galaxies with symmetric RCs perform similarly to the URC/NN trained on the complete data sample. We conclude that a URC with an acceptable level of accuracy ($ΔV_{\rm circ} \lesssim15$ per cent) at all radii would require a detailed modelling of a galaxy's central regions and outskirts (e.g., for baryonic effects leading to contraction or expansion of any dark-matter-only halo).
BibTeX
@article{patel2024nearly,
  author = {Raj Patel and Nikhil Arora and Stéphane Courteau and Connor Stone and Matthew Frosst and Lawrence M. Widrow},
  title = {{The Nearly Universal Disk Galaxy Rotation Curve}},
  journal = {The Astrophysical Journal},
  year = {2024},
  doi = {10.3847/1538-4357/ad58bc},
}
Figure