From blue to red spirals: Slow galaxy transformation via ram pressure stripping in TNG-50

The paper investigates whether ram-pressure stripping (RPS) alone can transform blue spiral galaxies into red spirals in low-mass group environments, using the TNG-50 cosmological simulation. The authors selected 41 subhalos at z=0 with disky morphologies, stellar masses > 5×10 9 M⊙, gas masses < 5×10 9 M⊙, and excluded systems with significant mergers or flybys (stellar mass growth < 5% since z=1). They verified that stellar rotational support (κrot) remains constant over time, confirming that gravitational perturbations are not dominant.

The sample was classified by g-r color into blue (30%), green (35%), and red (35%) spirals. Red subhalos have infall times > 2 Gyr (93%), with 75% exceeding 3.5 Gyr, 50% exceeding 6 Gyr, and 25% exceeding 7 Gyr. The transformation from blue to red spirals is gradual, with quenching timescales ≳ 6 Gyr after infall, longer than the 4 Gyr typically associated with merger-driven evolution and the 1-2 Gyr rapid quenching seen in massive clusters.

The 11 red subhalos show two distinct behaviors: those with infall times ≳ 6 Gyr show a gradual color increase to 0.7-0.8 mag over > 4 Gyr, while those with shorter infall times (< 6 Gyr) show similar but faster evolution. Some subhalos experience temporary SFR enhancements (rejuvenation) near pericentric passages due to ram-pressure-induced gas compression, but subsequently quench as gas is stripped.

The authors developed a RAMness parameter (RAMness = FS+/FSTot) to quantify the competition between ram pressure and gravitational restoring force. High RAMness (≳ 0.85) coincides with elongated gas tentacles and active stripping, confirming RPS as the dominant mechanism. Red subhalos experience multiple jellyfish phases, with gas not fully removed during the first pericenter passage, allowing repeated stripping episodes persisting up to 7 Gyr.

No significant correlation was found between transformation timescale and host halo mass (Spearman ρ = 0.07, p = 0.83), indicating that orbital history and cumulative RPS exposure are more important than halo mass. The resulting galaxies are predominantly red, anemic spirals rather than fully transformed S0 systems, indicating that gas removal alone is insufficient for complete morphological transformation. The paper concludes that RPS in group environments can generate long-lived quenched spirals, representing an intermediate evolutionary pathway preceding lenticular galaxy formation.

Improvements to AI Systems:

1. Temporal Quenching Prediction Model

• Improvement: Train a time-series transformer on TNG-50 subhalo trajectories (infall time, gas mass, color evolution) to predict the gradual color transition (Δg-r over 4–7 Gyr) rather than binary blue/red classification.

• Capability: The AI can forecast the exact quenching timescale for a given galaxy’s orbital history, distinguishing between slow RPS-driven quenching (>6 Gyr) and rapid merger-driven quenching (4 Gyr) in group environments.

2. RAMness-Aware Physical Feature Extractor

• Improvement: Integrate the RAMness parameter (FS+/FSTot) as a physics-informed input feature into a convolutional neural network (CNN) analyzing gas morphology (e.g., tentacles, jellyfish tails).

• Capability: The AI can autonomously identify active stripping events in real-time from simulated or observational gas maps, flagging subhalos where ram pressure exceeds gravitational restoring force (RAMness ≥ 0.85) and predicting imminent gas loss.

3. Rejuvenation Event Detector

• Improvement: Add a recurrent neural network (RNN) module that monitors pericentric passages and detects temporary SFR spikes caused by ram-pressure-induced gas compression.

• Capability: The AI can separate genuine star-forming galaxies from those undergoing temporary rejuvenation, preventing misclassification in large surveys and improving star-formation history reconstruction.

4. Multi-Episode Stripping Simulator

• Improvement: Use a graph neural network (GNN) to model repeated gas-stripping episodes across multiple pericenter passages, learning that gas is not fully removed in the first pass.

• Capability: The AI can simulate cumulative RPS effects over 7+ Gyr, predicting the final gas fraction and color of a galaxy given its full orbital path—useful for generating synthetic galaxy catalogs or forecasting future observations.

5. Morphological Transition Classifier

• Improvement: Train a classifier on the paper’s finding that RPS produces red anemic spirals (not S0s) to distinguish between gas-stripped spirals and merger-transformed lenticulars using kinematic (κrot) and color data.

• Capability: The AI can accurately label galaxies as “quenched spiral” vs. “true S0” in surveys, improving morphological evolution studies and reducing contamination in lenticular galaxy samples.

6. Orbital-History Encoder for Halo-Mass Independence

• Improvement: Replace host-halo-mass features with an orbital-history encoder (e.g., infall time, pericenter distance, cumulative RAMness) in a regression model, leveraging the null correlation (ρ=0.07) with halo mass.

• Capability: The AI can predict quenching timescales independent of halo mass, enabling robust applications across different group/cluster environments without mass-based bias.

Late-type galaxies lose gas through ram-pressure stripping (RPS) after falling into a massive halo. Because this mechanism primarily removes the gaseous component while leaving the stellar disk largely undisturbed, it provides a pathway for quenching star formation without immediate morphological transformation. While RPS is well established in galaxy clusters, galaxy evolution in low-mass groups is often attributed to mergers, leaving open the question of whether RPS alone can drive the transition from blue, star-forming spirals to quenched systems in these environments. We use the high-resolution cosmological simulation TNG-50 to investigate the evolution of blue spiral galaxies, after their infall into group-scale halos. We excluded systems undergoing significant mergers, thus isolating the effect of RPS. We find that RPS in low-mass groups (M group<10 14.5 M) can efficiently quench star formation while preserving the stellar disk structure. The transformation is gradual, with quenching timescales 6 Gyr after infall, longer than the about4 Gyr typically associated with merger-driven evolution. The resulting galaxies are predominantly red, anemic spirals rather than fully transformed S0 systems, indicating that gas removal alone is insufficient to produce complete morphological transformation. Our results show that RPS in group environments can generate long-lived quenched spirals which might represent an intermediate evolutionary pathway preceding the formation of lenticular galaxies.