Reconstructing orbits of galaxies in extreme regions (roger v2.0): an extension to intermediate mass systems

arXiv:2608.19429 · astro-ph.GA · Submitted 2026-08-19 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics Radio. Generated commentary on the latest astrophysics papers.

Vera: I'm Vera, and with me are Jocelyn and Subrahmanyan, guest researcher.

Jocelyn: Today's paper: "Reconstructing orbits of galaxies in extreme regions (roger v2.0)".

Vera: This research presents an updated orbital classification code, roger v2.0, which extends previous methods to analyze galaxy groups and clusters by incorporating host halo mass as an input parameter,

Jocelyn: First, who's behind it and why it matters.

Title and authors: Vera: So, we're looking at this paper titled "Reconstructing orbits of galaxies in extreme regions (roger v2.0): an extension to intermediate mass systems," and the authors are de los Rios, Martínez, Ruiz, Levis, Coenda, and Muriel <ref:2608.19429#pg0,Reconstructing orbits of galaxies in extreme regions (roger v2.0): an extension>. It seems they’re updating a classification tool for galaxies in groups and clusters by adding a new parameter that tells us about the host halo mass.

Jocelyn: That sounds really relevant for our observations because we deal with systems of various sizes out there; I wonder how incorporating the host halo mass changes what we can actually see in those deep fields.

Subrahmanyan: From a theoretical standpoint, this paper suggests that understanding these orbital dynamics within intermediate mass systems is crucial because it connects the local galaxy behavior to the larger structure formation processes occurring across different halo masses.

Vera: Exactly, Jocelyn; thinking about how we use our data to map out these structures, this extension seems like it could give us a much clearer picture of where galaxies are actually going within those massive environments.

Jocelyn: I agree; if the classification gets more robust across different scales, it means we can trust our kinematic measurements more when trying to figure out the physical processes at play.

Subrahmanyan: And that robustness is what allows us to test models of galaxy evolution because we can finally look at how host halo mass influences those orbital behaviors, which is a big step toward understanding structure growth.

The paper's summary: Vera: In this paper, the core of it is introducing roger v2 point 0, which takes the original method for classifying galaxy orbits in groups and clusters and adds the host halo mass as an extra input variable to help make those classifications more reliable <ref:2608.19429#pg0,the host halo mass as an>.

Jocelyn: So, instead of just using projected coordinates like distance and velocity relative to the cluster size, they’re feeding this new mass information into their classification process to see how it impacts the results.

Subrahmanyan: The paper highlights that even though adding the halo mass doesn't cause huge shifts in the classification itself for most things, it still improves the overall robustness of distinguishing between different galaxy orbital types.

Vera: Right, so they're using this mass input to refine how they categorize galaxies into five distinct orbital classes: Cluster Members, Recent Infallers, Backsplash galaxies, Infalling galaxies, and Interlopers.

Jocelyn: That way you can separate the populations based on their physical history around the cluster—for instance, telling the difference between something that's just passed through and something that's actually settled in.

Subrahmanyan: This helps constrain theoretical models by showing how mass affects these distributions, which is important because we need to see how galaxy evolution behaves at different scales of dark matter halos.

The paper's improvements: Vera: What’s interesting about the improvements they propose is that they are studying exactly how this host halo mass parameter influences the projected phase-space distribution, or PPSD, for each of those five orbital classes.

Jocelyn: So, they aren't just adding a variable and moving on; they are actively investigating whether the distribution of these five classes changes depending on how massive the host system is.

Subrahmanyan: The study specifically looked at how the median values of the PPSD coordinates shift with mass; for instance, they found that for Cluster Members, there's a slight growing tendency with mass.

Vera: And then they noted that for Infalling galaxies and Recent Infallers, there's actually a subtle decline in the velocity term relative to the cluster size as the halo mass gets larger.

Jocelyn: That’s fascinating because it suggests that for those specific populations, their kinematic state is more sensitive to the environment's mass than we might initially think.

Subrahmanyan: They also looked at how the fraction of galaxies in each class changes with halo mass; they found that for Cluster Members, this fraction goes down as mass increases, while for Infalling and Recent Infallers, it actually increases with mass.

Conclusion: Vera: So to wrap up on "Reconstructing orbits of galaxies in extreme regions (roger v2.0): an extension to intermediate mass systems," the paper shows that incorporating the host halo mass helps make the classification more stable by showing how its value shifts across different orbital classes <ref:2608.19429#pg0,Reconstructing orbits of galaxies in extreme regions (roger v2.0): an extension>.

Jocelyn: Basically, they confirm that this extended roger code gives us a better way to handle contamination and distinguish between those five orbital types in complex environments like galaxy groups and clusters.

Subrahmanyan: From a cosmic perspective, the implication is that these mass-dependent shifts tell us something concrete about how structure assembly dictates the orbital pathways galaxies take, which feeds directly into simulations of large-scale structure formation.

Vera: It’s really about getting a more precise handle on galaxy dynamics in those extreme regions where things are happening fastest.

Jocelyn: And with the implementation of this method, we can expect to see cleaner samples when we look at the PPSD data, which is exactly what we need for our next round of survey analysis.

Subrahmanyan: We look forward to seeing how these results integrate into larger cosmological frameworks and constrain the physics governing galaxy evolution across different halo masses.

Instituto de Astronomía Teórica y Experimental, CONICET-UNC

astro-ph.GA

Submitted: 2026-08-19

Updated: 2026-10-06

Comments: Code hosted at https://github.com/Martindelosrios/pyROGER. Accepted for its publication at MNRAS

Code: https://github.com/Martindelosrios/pyROGER

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 73/100

The gist: This research presents an updated orbital classification code, roger v2.0, which extends previous methods to analyze galaxy groups and clusters by incorporating host halo mass as an input parameter,

Key concepts

roger v2.0
An updated orbital classification code designed to categorize galaxies within galaxy groups and clusters. It uses projected position and velocity data, enhanced by host halo mass, to determine a galaxy's orbital path around the cluster.
PPSD position
A set of three input parameters used for training the classification model: the projected distance (R_p/R200), the line-of-sight velocity (|ΔV|/σ), and the parent halo's M200. These coordinates describe a galaxy's location relative to the cluster center.
Orbital Classes
Five specific categories used to classify galaxies based on their behavior around a cluster: Cluster members (CL), Recent infallers (RI), Backsplash galaxies (BS), Infalling galaxies (IN), and Interlopers (IL). These classes describe whether a galaxy is currently orbiting, recently arrived, or physically associated with the system.
Host Halo Mass
The mass of the dark matter halo that hosts the galaxy group or cluster being studied. This parameter is included in roger v2.0 to improve classification accuracy by accounting for variations in the environment's gravitational influence.

Terminology

Summary

This research presents an updated orbital classification code, roger v2.0, which extends previous methods to analyze galaxy groups and clusters by incorporating host halo mass as an input parameter, thereby improving the robustness of classifying galaxies in extreme regions.

The gist

The roger v2.0 code performs the orbital classification of galaxies in galaxy groups and clusters by using projected phase-space coordinates and incorporating the host halo mass as an additional input parameter to enhance classification robustness.

Data and Sample Selection

The main data sample utilized is the MDPL2-SAG galaxy catalog, which combines the MultiDark Planck 23 cosmological simulation (MDPL2) with the semi-analytic model of galaxy formation and evolution (SAG). The selection process involves imposing criteria on dark matter halos, specifically selecting those with a mass of M200 ≥ 1013.5 h−1M⊙, and applying isolation criteria to avoid perturbations from mergers or heavy interactions. For each cluster, galaxies are selected based on their projected distance (R p) and line-of-sight velocity (ΔV) relative to the cluster's characteristic size (R200) and velocity dispersion (σ), satisfying specific inequalities defined in equations (1) and (2). Furthermore, the sample includes interlopers, which are galaxies unrelated to clusters but appear projected near them, serving as a source of contamination.

Orbit Classification Scheme

The code classifies selected galaxies into five distinct orbital types based on their behavior around the cluster:

  1. Cluster members (CL): Galaxies that are now orbiting the cluster and have crossed R200 several times in their lifetime. Most of these are found inside R200.

  2. Recent infallers (RI): Galaxies that have crossed R200 only once, and in their way in, in the last 2 Gyr.

  3. Backsplash galaxies (BS): Galaxies located beyond R200 at z = 0 but have passed through the cluster once in the past, crossing R200 exactly twice on their way in and out.

  4. Infalling galaxies (IN): Galaxies that have been outside R200 during their entire lifetime but are now falling toward the cluster, evidenced by negative radial velocities relative to the cluster center.

  5. Interlopers (IL): Galaxies located beyond R200 with positive radial velocities relative to the cluster center, which are not physically associated with the system.

Model Training and Classification

The roger v2.0 model is trained using two machine learning techniques: K-Nearest Neighbors (KNN) and Random Forests (RF). The training procedure involves inputting three parameters for each galaxy: (i) parent halo’s M200; (ii) galaxy’s real class: CL, RI, BS, IN, or IL; and (iii) the galaxy's PPSD position (R p/R200, ΔV/σ). The training set is partitioned into an 80% training subset and a 20% testing set. Classification is achieved by computing the probability of belonging to each of the five classes, and a final classification is determined by adopting a threshold T j for each class j: "classify a galaxy as belonging to the j-class (hereafter predicted class) if (i) the maximum value of its p i corresponds to i = j, and (ii) p j > T j."

Mass Dependence and Performance Evaluation

The inclusion of host cluster mass as a third parameter was studied to determine its effect on the PPSD distribution. The analysis showed that for most classes, the median values of the PPSD coordinates do not depend on mass, or the mass dependence is mild. Specifically, for CL galaxies, there is a slight growing tendency with mass, while for RIN and IN galaxies, a subtle decline with mass in ΔV/σ is observed. The fraction of galaxies in each class as a function of halo mass was also analyzed; for instance, the fraction of CL galaxies decreases with mass, while the fraction of RI and IN galaxies increases with mass. Performance metrics like precision and sensitivity were evaluated as functions of the threshold and halo mass, revealing that the innermost classes, CL and RI, are predicted best at the lower mass end, whereas the outermost classes, IN and IL are recovered best at the high mass end. The final classification scheme uses a mass-dependent threshold T j max derived from linear fits to maximize both sensitivity and precision simultaneously.

Implementation

A Python implementation of roger v2.0, named pyroger, is provided for public availability. This framework allows users to train models (e.g., using roger2 KNN or roger2 RF) and then use the trained model to classify galaxies or estimate their probabilities. Users can predict classes by calling models.

Improvements for AI systems

Here are the specific improvements that can be made to AI systems based on this scientific paper, along with what those improved systems can do:

  1. Upgrading Galaxy Classification Robustness in Cluster Environments:

  2. Expanding Applicability to Lower Mass Systems:

  3. Enhancing Classification Accuracy via Host Halo Information:

  4. Developing Flexible and Adaptable Machine Learning Pipelines:

  1. Upgrading Galaxy Classification Robustness in Cluster Environments:

This improvement involves deploying the updated classification framework (roger v2.0) which incorporates host halo mass as an input parameter alongside projected phase-space coordinates. The improved AI system can accurately classify galaxies residing in galaxy groups and clusters, specifically distinguishing between five orbital classes (Cluster Members, Backsplash Galaxies, Recent Infallers, Infalling Galaxies, and Interlopers). It is robust against orientation variations of the host cluster due to the inclusion of three-dimensional projections (x, y, z axes) during training.

  1. Expanding Applicability to Lower Mass Systems:

The improved system can now perform orbital classifications for systems with host halo masses down to 10135 h−1M⊙ (compared to the original limit of 1015 h−1M⊙). This allows the AI system to analyze and classify galaxy populations in smaller, less massive structures, opening new avenues for studying galaxy evolution in environments previously inaccessible.

  1. Enhancing Classification Accuracy via Host Halo Information:

The AI system can utilize the host halo mass as a crucial feature for classification refinement. By training models (like K-Nearest Neighbors or Random Forests) on this additional parameter, the classification performance is improved and becomes more robust across different mass regimes. Specifically:

  • For Cluster Members (CL), the median projected distance exhibits a slight growth with mass, which can be used to refine predictions for massive systems.

  • For Recent Infallers (RI) and Infalling Galaxies (IN), the relative line-of-sight velocity shows a subtle decline with mass, allowing the AI to better predict their kinematic state in more massive halos.

  1. Developing Flexible and Adaptable Machine Learning Pipelines:

The system can be implemented using a Python library that allows for:

  • Training customized classifiers on alternative training sets.

  • Utilizing standard machine learning techniques (K-Nearest Neighbors, Random Forests, Support Vector Machines) to build the classification model.

  • Employing statistical correction methods (confusion matrix inversion) to mitigate contamination from misclassifications in the PPSD, leading to statistically corrected distributions of galaxy properties.

Sources

Related papers