When physicists propose theories beyond the Standard Model—such as Supersymmetry, extra dimensions, or sterile neutrinos—they introduce new parameters like hypothetical masses, coupling constants, and mixing angles. Testing these theories is notoriously difficult because a single parameter space can influence dozens of entirely different physical phenomena, from the debris of high-energy collider impacts to the relic density of dark matter frozen out since the Big Bang.
If we test these parameters against experimental results one by one, we suffer from a form of tunnel vision. A specific parameter value might perfectly explain an anomaly in a B-meson decay, yet be completely incompatible with cosmological observations. To truly understand if a model describes reality, we must look at the entire picture at once.
Global statistical fitting solves this by pooling multi-modal experimental constraints into a single, unified mathematical framework. It allows physicists to map out high-dimensional parameter spaces, expose hidden statistical tensions, and determine whether a theory is genuinely viable or merely overfitted to an isolated anomaly.
The Intuition: Breaking Degeneracies by Pooling Information
Why does analyzing experiments simultaneously work better than analyzing them in isolation? The answer lies in parameter degeneracies.
Imagine a theoretical model with two free parameters. A single experiment might constrain these parameters along a "flat direction"—a continuous ridge where changing both parameters in tandem yields the exact same experimental prediction. From this experiment alone, an infinite combination of parameters looks equally valid.
Experiment A Constraint: ====== (Flat direction / Degeneracy)
Experiment B Constraint: |||||| (Different flat direction)
Combined Global Fit: X (Unique intersection point)
Global fitting breaks these degeneracies by multiplying independent likelihood functions together (or summing their $\chi^2$ values). If Experiment A and Experiment B probe different physical regimes, their flat directions will likely intersect rather than run parallel.
As shown by Roberto Ruiz de Austri et al. in Prospects for direct dark matter detection in the constrained MSSM, combining cosmological dark matter relic density with collider observables into a unified likelihood function completely reshapes the allowed parameter space. What looked like a vast, unconstrained plain in the Minimal Supersymmetric Standard Model (MSSM) contracts into narrow, highly predictive islands of validity.
Mapping the Landscape: Bayesian and Frequentist Scans
Once a joint likelihood function is constructed, the challenge shifts to exploration. How do we find the regions of a multi-dimensional parameter space that match all constraints? Two primary statistical paradigms are used to scan these spaces:
The Bayesian Approach
Bayesian global fits use algorithms like Markov Chain Monte Carlo (MCMC) or nested sampling to map out the posterior probability distribution of the parameters. This approach excels at identifying the most probable regions of parameter space given the data, naturally accounting for volume effects in high dimensions.
For instance, Gianfranco Bertone et al. in Global fits of the minimal universal extra dimensions scenario used a global Bayesian framework to evaluate the Minimal Universal Extra Dimensions (MUED) model. By simultaneously processing electroweak precision data, dark matter relic density, and collider limits, the authors could map out the exact regions where the theory remains viable. Similarly, M. Ciuchini et al. applied a Bayesian global fit in On flavourful Easter eggs for New Physics hunger and lepton flavour universality violation to integrate multiple rare decay channels, revealing a coherent $3\sigma$ signal for new physics that isolated measurements could not definitively confirm.
The Frequentist Approach
Frequentist global fits rely on $\chi^2$ minimization to find the absolute best-fit point and draw confidence intervals. This method focuses purely on the goodness-of-fit, asking how well the model can possibly perform, regardless of how small or fine-tuned the valid parameter region is.
C. Giunti et al. heavily utilized this approach in Update of short-baseline electron neutrino and antineutrino disappearance and Pragmatic view of short-baseline neutrino oscillations. By using global least-squares fitting to synthesize data from radioactive sources, nuclear reactors, and accelerator experiments, the authors could test whether a "3+1" sterile neutrino framework could consistently explain anomalies across different baselines and energy scales.
Diagnosing Model Health and Statistical Tension
Global fits do more than just find best-fit parameters; they act as a diagnostic "health check" for theoretical models. When a model is fundamentally flawed, a global fit will expose statistical tension—a situation where different datasets pull the parameters in opposite, irreconcilable directions.
Consider what happens when a new, high-precision measurement is made. In Quantifying the tension between the Higgs mass and(g−2)μin the constrained MSSM, María Eugenia Cabrera et al. demonstrated that if the Higgs boson mass is around 125 GeV, the parameter space of the constrained MSSM is pushed into a corner where it can no longer resolve the anomalous magnetic moment of the muon $(g-2)_\mu$. The global fit quantifies this tension, showing that the model cannot simultaneously satisfy both experimental constraints without breaking.
Higgs Mass Constraint ----> [ Parameter Space ] <---- Muon (g-2) Constraint
|
(Severe Statistical Tension)
This diagnostic power is equally useful for evaluating localized experimental anomalies. When the "Gallium Anomaly" suggested the existence of a sterile neutrino, C. Giunti et al. in Gallium Anomaly: critical view from the global picture of νe and $$ {\overline{\nu}}_e $$ disappearance performed a global fit including reactor, tritium, and solar neutrino data. Using a goodness-of-fit metric, the authors proved that the sterile neutrino parameters required to explain the Gallium data were in deep, systemic tension with the rest of the global neutrino dataset, suggesting the anomaly likely stems from an unidentified systematic error rather than new physics.
Universality Across Physical Scales
While global fitting is a staple of particle physics and dark matter searches—such as the comprehensive sub-GeV dark matter scans performed by Sowmiya Balan et al. in Resonant or asymmetric: The status of sub-GeV dark matter—the underlying mechanism is entirely universal. It applies to any field where complex, multi-parameter models must be reconciled with diverse, multi-modal data.
For example, in nuclear physics, M. C. Atkinson et al. applied this exact methodology in Dispersive optical model analysis of Pb-208 generating a neutron-skin prediction beyond the mean field. Instead of analyzing dynamical scattering data and static ground-state nuclear properties separately, the authors used a global fit to optimize a dispersive optical model against elastic scattering cross sections, charge density distributions, and single-particle energy levels simultaneously. This multi-modal integration ensured that the resulting prediction for the "neutron skin" of Lead-208 was consistent across both low-energy static and high-energy dynamical regimes.
Historically, even the internal structure of the proton was mapped this way. In Measurement of the proton and the deuteron structure functions, F2 and F2, M. Arneodo et al. used a global phenomenological parametrization to unify data from multiple independent deep inelastic scattering experiments. By reconciling measurements across vastly different kinematic regions, the global fit provided a continuous, mathematically coherent description of nucleon structure functions.
Summary
Global statistical fitting works because it prevents us from evaluating theories in a vacuum. By mathematically pooling independent constraints, it:
- Eliminates flat directions by forcing parameters to satisfy multiple physical criteria simultaneously.
- Quantifies model health by exposing statistical tensions when independent datasets disagree within the framework of the theory.
- Provides a rigorous roadmap for future experimental searches, pointing exactly to the remaining "surviving" islands of parameter space.
Key Foundational Work
On the detectability of the CMSSM light Higgs boson at the Tevatron (2007) — contributed 2 concept elements to this direction.
Prospects for direct dark matter detection in the constrained MSSM (2007) — contributed 1 concept elements to this direction.
Measurement of the proton and the deuteron structure functions, F2 and F2 (1995) — contributed 1 concept elements to this direction.