SIRIUS Project: Dynamical Evolution of Primordial Binaries during Star Cluster Formation

Context. Binary populations are closely linked to the star formation process; however, their primordial properties can be changed by subsequent dynamical interactions within their natal clusters.

Aims. The aim of this study is to clarify how different primordial binary populations affect the evolution of multiplicity and the global structure of forming star clusters.

Methods. We investigate the dynamical evolution of primordial binaries during star cluster formation using self-consistent Nbody/smoothed particle hydrodynamics simulations that follow the collapse of a molecular cloud to a star cluster. We systematically compare three star formation models: a close binary formation model (CB), a wide binary formation model (WB), and a single star formation model (SS).

Results. In CB and WB models, the multiplicity fraction decreases with time due to dynamical interactions. In particular, the fraction in the WB model drops to a level comparable to that in the SS model. The multiplicity fraction of high-mass stars is similarly high in all models, whereas only the CB model shows a relatively high fraction for low-mass stars. Due to the assumption of equal-mass binary formation, the CB and WB models exhibit an excess at q = 1 in the mass-ratio distribution, while the SS model has no clear trend. Frequent few-body interactions generate distinct stellar populations inside and outside the cluster: the multiplicity fraction within the cluster is systematically higher, while mass functions in the outside have a shallower slope. Finally, stellar density profiles in the clusters are broadly similar among all models.

Conclusions. The primordial binary population significantly affects the final binary properties, while having only a limited impact on their host cluster structures. Our results suggest that close binaries need to form at the star formation stage to reproduce the observed multiplicity fraction of low-mass stars and the excess of equal-mass binaries.

Improvements to AI Systems:

1. Dynamic Binary Evolution Simulator – Train an AI to model the time-dependent evolution of binary fractions and mass-ratio distributions under varying primordial conditions (CB, WB, SS), enabling predictions of multiplicity fractions for low- and high-mass stars in forming clusters.

2. Cluster Structure Predictor – Use the finding that primordial binaries have limited impact on global cluster density profiles to build an AI that decouples binary evolution from structural outcomes, improving efficiency in simulations by approximating density profiles without full binary dynamics.

3. Observational Calibrator – Develop an AI that inverts the observed multiplicity fraction and mass-ratio excess (q=1) to infer the dominant primordial binary formation mode (close vs. wide) in real star-forming regions, using the paper’s result that close binaries are required to match low-mass multiplicity.

4. Subgrid Model for Galaxy Simulations – Implement a machine-learned subgrid prescription that adjusts binary properties (fraction, mass ratio) based on local stellar density and dynamical interaction rate, derived from the paper’s finding that few-body interactions reduce multiplicity, especially for wide binaries.

5. Population Synthesis Optimizer – Create an AI that optimizes initial binary parameter distributions (e.g., period, mass ratio) to match both final cluster multiplicity and field star mass functions, leveraging the paper’s observation that external populations have shallower mass slopes.

What the Improved AI System Can Do:

• Accurately forecast binary fractions and mass-ratio distributions in young clusters given initial conditions.

• Distinguish between close and wide binary formation scenarios from observable cluster properties alone.

• Accelerate star cluster simulations by predicting structural profiles without costly binary integration.

• Provide physically motivated priors for interpreting observations of low-mass binaries and equal-mass excesses.

• Generate synthetic catalogs of binary populations for testing telescope surveys and data reduction pipelines.

Binary populations are closely linked to the star formation process; however, their primordial properties can be changed by subsequent dynamical interactions within their natal clusters. The aim of this study is to clarify how different primordial binary populations affect the evolution of multiplicity and the global structure of forming star clusters. We investigate the dynamical evolution of primordial binaries during star cluster formation using self-consistent Nbody/smoothed particle hydrodynamics simulations that follow the collapse of a molecular cloud to a star cluster. We systematically compare three star formation models: close binary (CB), wide binary (WB), and single star (SS) formation model. In CB and WB models, the multiplicity fraction decreases with time due to dynamical interactions. In particular, the fraction in the WB model drops to a level comparable to that in the SS model. The multiplicity fraction of high-mass stars is similarly high in all models, whereas only the CB model shows a relatively high fraction for low-mass stars. Due to the assumption of equal-mass binary formation, the CB and WB models exhibit an excess at q= 1 in the mass-ratio distribution, while the SS model has no clear trend. Frequent few-body interactions generate distinct stellar populations inside and outside the cluster: the multiplicity fraction within the cluster is systematically higher, while mass functions in the outside have a shallower slope. Finally, stellar density profiles in the clusters are broadly similar among all models. The primordial binary population significantly affects the final binary properties, while having only a limited impact on their host cluster structures. Our results suggest that close binaries need to form at the star formation stage to reproduce the observed multiplicity fraction of low-mass stars and the excess of equal-mass binaries.