Source-linked AI summary
New features in the simulation of neutrino oscillation experiments with GLoBES 3.0
Patrick Huber, Joachim Kopp, Manfred Lindner, Mark Rolinec, Walter Winter
TL;DR
GLoBES 3.0 addresses the need for flexible simulation and analysis of short- and long-baseline neutrino oscillation experiments. It introduces user-defined systematics, priors, and probability engines, with the key result that arbitrary discrete source-detector combinations and non-standard physics can be handled without recompilation. The release also updates experiment prototypes and supports beta beam fluxes.
Problem
Existing GLoBES extensions were needed to support arbitrary systematics, multi-source or multi-detector configurations, external information, and non-standard physics.
Method
GLoBES 3.0 adds user-defined systematics and priors, a replaceable probability engine, and expanded AEDL and experiment-prototype support.
Results
GLoBES 3.0 simulates arbitrary combinations of discrete sources and detectors, incorporates external constraints, and supports non-standard physics without recompilation.
Takeaways & Limitations
The three user-defined concepts—oscillation probabilities, systematics, and priors—provide flexibility for experimentalists and theorists across software levels.
Takeaways & Limitations
GLoBES assumes one stationary point source per experiment and cannot study geometrically distributed or physically time-dependent sources such as supernovae.
Abstract
from arXiv · showhide
We present Version 3.0 of the GLoBES (``General Long Baseline Experiment Simulator'') software, which is a simulation tool for short- and long-baseline neutrino oscillation experiments. As a new feature, GLoBES 3.0 allows for user-defined systematical errors, which can also be used to simulate experiments with multiple discrete sources and detectors. In addition, the combination with external information, such as from different experiment classes, is simplified. As far as the probability calculation is concerned, GLoBES now provides an interface for the inclusion of non-standard physics without re-compilation of the software. The set of experiment prototypes coming with GLoBES has been updated. For example, built-in fluxes are now provided for the simulation of beta beams.
1 Introduction
GLoBES 3.0 responds to the need for accurate, efficient simulation and analysis of neutrino oscillation experiments with improved flexibility. Its new features support arbitrary systematics, multi-detector simulations, external priors, and non-standard physics without recompilation.
- Motivation: GLoBES simulates the performance of the average neutrino oscillation experiment, providing a major performance gain over Monte Carlo simulation.It is designed for highly accurate beam and detector simulation and analysis of correlations and degeneracies.
- New features: GLoBES 3.0 introduces arbitrary systematics for multi-detector experiments and extends AEDL with beta beam fluxes and list variables.These additions target experimental implementation flexibility.
- New features: User-defined priors simplify incorporating external information from different experiments.The feature is intended for phenomenological analyses.
- New features: The probability engine can simulate non-standard physics without recompiling GLoBES.This separates probability-model extensions from the core software build.
2 Concept of GLoBES
GLoBES provides an abstract, flexible framework for simulating and analyzing short- and long-baseline neutrino experiments. Version 3.0 extends this framework to arbitrary systematics and combinations of discrete sources and detectors while supporting detailed probability and χ2 analyses.
- Software framework: GLoBES uses AEDL, a C-library, and a binary testing program to define experiments, compute probabilities and rates, obtain ∆χ2-values, and validate definitions.The package supports complete three-flavor descriptions of short- and long-baseline experiments.
- Systematics: GLoBES 3.0 can cross-define systematic errors across experiments, enabling combinations of multiple discrete sources and detectors.This includes several sources with one detector, several detectors with one source, or several sources and detectors.
- Experiment definition: Experiment definitions can specify built-in or user-defined fluxes, cross sections, efficiencies, resolution functions, channels, backgrounds, and user-defined systematics.Energy ranges, windows, and bin widths can also be chosen almost arbitrarily.
- Analysis: The C-library extracts ∆χ2, probabilities, and event rates for individual or combined experiments with arbitrary matter-density profiles.It also supports matter-density uncertainty and precision extraction.
- Analysis: Local minimization projects multi-parameter χ2 functions onto parameter subspaces and can localize degenerate solutions.External constraints, including solar-parameter constraints, can be imposed during projection.
3 Oscillation probabilities and the simulation of non-standard physics
GLoBES calculates oscillation probabilities by diagonalizing the Hamiltonian across constant-density matter layers, then uses those probabilities for event rates. GLoBES 3.0 adds a reconfigurable probability engine that supports non-standard physics and extra parameters without recompilation.
- Standard probability calculation: The standard probability calculation diagonalizes the Hamiltonian in layers of constant matter density within the three-flavor oscillation framework.The resulting probabilities feed the event-rate computation.
- Layered propagation: For k matter-density layers, the oscillation probability is constructed from evolution operators associated with layer densities and thicknesses.The Hamiltonian in each layer is diagonalized using the mixing matrix in matter and its eigenvalues.
- Non-standard physics: GLoBES 3.0 allows users to replace the Hamiltonian and complete probability engine by registering modified functions without recompiling the software.A user can copy, modify, and register the standard probability engine.
- Non-standard physics: The interface supports more than the standard six oscillation parameters plus matter density by adding n non-standard parameters η_j.These parameters are accessed through the usual parameter-handling mechanisms.
- Applications: Non-standard physics implementations include damping effects, Hamiltonian-level effects, and mass-varying neutrinos.The example distribution includes an analytical implementation of wave-packet coherence loss through σE.
4 Systematics implementation and user-defined systematics
GLoBES 3.0 implements systematic errors through nuisance parameters and a pull-based χ2 framework, while user-defined systematics extend correlations across rules and experiments.
- Default systematics: GLoBES supports signal and background normalization errors plus tilt or energy-calibration errors by default.Tilt linearly distorts the spectrum around its center, whereas calibration stretches the reconstructed energy scale.
- Pull method: The pull method introduces nuisance parameters that scale signal or background rates, with Gaussian penalties corresponding to systematic uncertainties.For signal and background normalization errors, rates are multiplied by factors involving the nuisance parameters, and χ2 is minimized over them.
- Detector correlations: Standard near–far detector systematics leave normalization between the two detectors uncorrelated.GLoBES 3.0 replaces this fixed construction with a modifiable χ2 function for detector-specific uncertainties.
- Example calculation: Figure 3 evaluates sin2 2θ13 sensitivity versus luminosity at 90% confidence for several systematic-error assumptions.Its thick curve uses σR = σcorr = 2.8%, σN = σF = σuncorr = 0.6%, plus an energy-calibration error; the calculation uses example5.c.
- User-defined systematics: User-defined systematics can correlate errors across different rules or experiments and introduce additional error types.Examples include calibration, shape, and uncorrelated bin-to-bin errors, with each user-defined systematic identified by a name matched by the application.
5 Adding external information with user-defined priors
GLoBES 3.0 extends marginalization and external-information handling through user-defined priors, supporting non-Gaussian errors and broader combinations of experimental data.
- Marginalization: GLoBES projects the fit manifold onto a chosen oscillation-parameter subspace by scanning one parameter and marginalizing over the others.For sin^2 2θ13 precision, the sin^2 2θ13 direction is scanned while the remaining oscillation parameters are marginalized.
- External constraints: External Gaussian constraints can be added to oscillation parameters, including matter density, before marginalization.An external constraint on θ12 is added before the marginalization step.
- User-defined priors: User-defined priors f(λ) accommodate non-Gaussian external errors and other external information beyond predefined Gaussian constraints.This replaces the fixed Gaussian form when external errors are non-Gaussian or additional information is needed.
- Applications: The priors support combinations of terrestrial neutrino data with neutrino-telescope information and long-baseline data with atmospheric-neutrino data.These combinations are cited as experimental uses of the feature.
- Caveat: The marginalization-order description is retained for pedagogy even though it is not strictly valid for GLoBES 3.0’s new hybrid minimizer.The qualification applies specifically to the systematics–correlations marginalization order.
6 AEDL changes and experiment prototypes
GLoBES 3.0 broadens AEDL experiment descriptions by decoupling systematics from rules, adding user-defined correlations and errors, and supplying modifiable experiment prototypes.
- AEDL structure: AEDL is a text-file syntax for describing GLoBES experiments through constructions such as channels, rules, experiments, and experiment combinations.These constructions organize fluxes, oscillation channels, backgrounds, detectors, baselines, and experiment-level χ2 values.
- Channel: A channel maps source flux, flavors, particle type, interaction, and cross section to an event-rate vector.Channels provide the inputs used to build signal and background components.
- Rules and combinations: A rule combines signal and background channels and applies their associated systematics to produce a systematics χ2.An experiment sums the rule contributions, while an experiment combination sums across experiments.
- AEDL changes: User-defined systematics break the strict coupling between systematics and rules, allowing correlations across rules or experiments and application-defined systematic errors.AEDL also adds version requirements, list variables, and interpolation.
- Experiment prototypes: Predefined experiment prototypes can be modified for users’ experiments or used to test new physics concepts.Integrated luminosities, baselines, fluxes, efficiencies, and other factors can be adjusted.
- Compatibility: Old AEDL prototypes should continue to run but are no longer maintained by the GLoBES team.
7 Internal changes and improved performance
GLoBES 3.0 improves performance by restructuring minimization and replacing general-purpose diagonalization with a specialized hermitian 3×3 algorithm, while preserving compatibility by default.
- Motivation: Sophisticated multi-experiment simulations can take several days, motivating optimization of GLoBES’s multidimensional χ2 minimizations.
- Minimization: GLoBES separates minimization over oscillation parameters from nuisance systematics because only oscillation-parameter changes require recomputing oscillation probabilities.
- Hybrid minimizer: An alternative interleaved minimizer performs the full minimization in one pass and is several times faster in most cases.Each Powell iteration over oscillation parameters is followed by one iteration over systematics parameters.
- Compatibility: The faster minimizer is disabled by default because pathological cases may behave differently; users must select it through an API call.The default preserves compatibility with older application programs.
- Probability calculation: GLoBES replaces LAPACK-based diagonalization with a hermitian 3×3-specific analytical algorithm that falls back to the QL method in rare inaccurate cases.The change targets the small, well-conditioned matrices arising in GLoBES.
- Computing environments: Performance improvements also support 64-bit systems, Condor clusters, parallel clusters, and static binaries for parallelization.
8 Summary and conclusions
GLoBES 3.0 broadens simulation flexibility across experiment configurations, external information, and oscillation-probability models. Three user-defined concepts support adjustments by experimentalists and theorists at multiple levels.
- GLoBES 3.0 simulates arbitrary combinations of discrete sources and detectors rather than remaining restricted to one-source-one-detector configurations.User-defined systematical errors enable cross-definition of errors across different experiments.
- User-defined priors incorporate external constraints from atmospheric-neutrino experiments or neutrino telescopes and help localize complicated degeneracies.
- GLoBES 3.0 supports non-standard physics beyond three-flavor oscillations without recompiling the software.The implementation also permits more than six oscillation parameters.
- New experiment prototypes use these features and are illustrated by examples in the manual.The Double Chooz example requires user-defined systematics for correlated errors between near and far detectors.
- The release introduces user-defined oscillation probabilities, user-defined systematics, and user-defined priors as three flexible concepts.These concepts allow easier adjustment of the software for experimentalists and theorists at different levels.