---
title: 'GeMS-E: Multifaceted Scientific Frameworks'
url: https://www.emergentmind.com/topics/gems-e
type: topic
---

# GeMS-E: Multifaceted Scientific Frameworks

GeMS-E denotes several distinct, technically advanced systems and datasets across astrophysics, particle detection, and computer vision. The most prominent scientific usages are (1) the "Globular cluster Extra-tidal Mock Stars" catalogue in galactic dynamics, (2) an event-assisted 3D Gaussian Splatting framework for computer vision under extreme motion blur, and (3) the GEM–Emulsion hybrid tracking detector in neutrino physics. Each instance of GeMS-E is unified by the acronym—but refers to a domain-specific system, summarized rigorously below.

## 1. GeMS-E: Globular Cluster Extra-tidal Mock Stars Catalogue

GeMS-E in galactic dynamics is the acronym for the "Globular cluster Extra-tidal Mock Stars (E)" catalogue, an extensive library of star and binary orbits ejected from the cores of 159 Galactic globular clusters due to three-body dynamical interactions [2310.09331]. The dataset provides mock sky coordinates, phase-space distributions, and stellar parameters for 50,000 escapers per cluster, serving as a foundation for associating field stars to globular cluster progenitors with quantified statistical rigor.

The scientific objective is to facilitate "chemo-dynamical tagging"—the identification of real stellar escapers in the Galactic halo (e.g., in Gaia/SDSS/APOGEE) using simulated action–angle distributions, probe contamination of thin tidal streams by core-ejecta, and quantify cluster core dynamics and evolutionary history.

## 2. Dynamical Encounter Modelling and Corespray Algorithm

The catalogue is generated using the Corespray particle-spray code, which implements a rapid three-body scattering formalism for ejections:

- For each close 1+2 stellar encounter in the cluster core, the cross-section is 
  $$
  \sigma(E, L) = \int_{b < b_{\max}} 2\pi b\, db \approx \pi b_{\max}^2,
  $$
  with $b$ the impact parameter and $b_{\max}$ determined by sampled separation radii.

- The speed distribution for ejected singles is 
  $$
  f(v_{\rm recoil}) \propto v_{\rm recoil}^2 \exp\left(-\frac{v_{\rm recoil}^2}{2\sigma_c^2}\right),
  $$
  truncated at $v_{\rm recoil} > v_{\rm esc}$, where $\sigma_c$ is the cluster core's 1D velocity dispersion.

- Each cluster's parameters (mass, core/tidal radius, velocity dispersion, escape velocity, etc.) are taken from established catalogues (e.g. Baumgardt+2018), and triple stellar masses are sampled from a Kroupa IMF, evolved over 12 Gyr with McLuster to include realistic mass segregation.

- Binaries are assigned separations from a distribution $p(a) \propto 1/a$ (Öpik’s law), producing a range of soft and hard targets.

- The code integrates the Hamiltonian orbits of escapers in seven Milky Way potentials using galpy, outputting present-day 6D phase-space for each escaper.

## 3. Action–Angle Coordinates and Galactic Potential Models

The orbits of escaped stars and binaries are characterized in action–angle space, which isolates orbital families within the Galaxy. For axisymmetric potentials, actions are computed as
$$
J_R = \frac{1}{2\pi} \oint v_R\, dR, \quad
J_z = \frac{1}{2\pi} \oint v_z\, dz, \quad
J_\phi = L_z = R v_\phi.
$$
Corespray uses galpy’s “Stäckel fudge” to approximate the local galactic potential as a Stäckel potential, solving action integrals with percent-level accuracy for each escaper over seven galactic models—including MWPotential2014, McMillan17, bar- and spiral-perturbed disks, triaxial halos of varying flattening, and Large Magellanic Cloud (LMC) perturbations.

A notable result is the model-independence of the action distributions for most clusters: mean shifts in normalized action space remain $\lesssim 0.1$, suggesting robust statistical predictions across a range of plausible Milky Way potentials.

## 4. Stream Contamination and Probabilistic Association Tool

GeMS-E quantifies the contamination of classical tidal stellar streams by stars ejected through binary-single scattering ("core ejecta"), especially for clusters with pericentre radii $R_{\rm peri} > 5$ kpc. In clusters such as Pal 13, mock escapers from Corespray overlap the observed stream, demonstrating that a nontrivial fraction of stars in thin streams may arise from dynamical ejection rather than simple outer-layer stripping.

The contamination fraction is approximately given by
$$
f_{\rm mass,\,2+1} \equiv \frac{\dot M_{2+1}}{\dot M_{\rm tidal}} \approx \frac{\tau_{2+1}}{M/T_{\rm diss}},
$$
where $\tau_{\rm tidal} = M/T_{\rm diss}$ (Baumgardt & Makino 2003) and $\tau_{2+1}$ is the ejection timescale for binaries (Leigh & Sills 2011). For Pal 13, $f_{\rm mass, 2+1}\sim4\%$.

The catalogue includes a probabilistic tool enabling users to assign a field star (with measured actions) a likelihood of origin from each GC. Given PDF $\mathbb{P}_c(\boldsymbol J)$ for cluster $c$ estimated from the mock sample via kernel density estimation, and a candidate star with actions $\boldsymbol J$, the method computes
$$
P(c | \boldsymbol J) \propto \mathbb{P}_c(\boldsymbol J) \sim 
\exp\left[ -\frac{1}{2} (\Delta\boldsymbol J)^T \Sigma_c^{-1}\Delta\boldsymbol J \right]
$$
and returns the maximum-likelihood cluster and log-odds over the runner-up [2310.09331].

## 5. Access, Format, and Research Use

The GeMS-E catalogue is publicly hosted at Zenodo (https://zenodo.org/record/8436703). It provides two CSV files (singles and binaries) per cluster, each row encoding RA, Dec, heliocentric distance, proper motions ($\mu_\alpha$, $\mu_\delta$), radial velocity ($v_r$), mass, escape time, escape velocity, and computed actions ($J_R$, $J_\phi$, $J_z$), among other parameters.

Researchers can:
- Compute actions for observed stars in the same galactic potential (e.g., MWPotential2014 via galpy)
- Employ the accompanying Python script to compare to catalogue KDEs for cluster-matching in action space
- Optionally combine dynamical with chemical abundance data for enhanced "chemo-dynamical tagging" in field star populations

The GeMS-E framework thus furnishes a key reference for the identification of cluster-origin field stars, quantitative assessment of cluster mass loss histories, and modelling of stream contamination throughout the Galactic halo [2310.09331].

---

## Table: Overview of GeMS-E in Major Research Domains

| Domain               | System/Usage                                                        | Key Purpose/Contribution                      |
|----------------------|---------------------------------------------------------------------|-----------------------------------------------|
| Galactic dynamics    | Extra-tidal Mock Stars catalogue (GeMS-E)                          | Field star–cluster association, stream analysis|
| Computer vision      | Event-assisted 3D Gaussian Splatting (GeMS-E pipeline)             | 3D reconstruction from motion-blurred images  |
| Particle physics     | GEM–Emulsion hybrid tracking demonstrator (GeMS-E test chamber)    | High-resolution charged-particle tracking     |

---

Each GeMS-E system is distinct in methodology, but all represent high-fidelity, quantitative approaches within their respective research fields, as detailed above.

Source: https://www.emergentmind.com/topics/gems-e