-
Event-by-event Comparison between Machine-Learning- and Transfer-Matrix-based Unfolding Methods
Authors:
Mathias Backes,
Anja Butter,
Monica Dunford,
Bogdan Malaescu
Abstract:
The unfolding of detector effects is a key aspect of comparing experimental data with theoretical predictions. In recent years, different Machine-Learning methods have been developed to provide novel features, e.g. high dimensionality or a probabilistic single-event unfolding based on generative neural networks. Traditionally, many analyses unfold detector effects using transfer-matrix--based algo…
▽ More
The unfolding of detector effects is a key aspect of comparing experimental data with theoretical predictions. In recent years, different Machine-Learning methods have been developed to provide novel features, e.g. high dimensionality or a probabilistic single-event unfolding based on generative neural networks. Traditionally, many analyses unfold detector effects using transfer-matrix--based algorithms, which are well established in low-dimensional unfolding. They yield an unfolded distribution of the total spectrum, together with its covariance matrix. This paper proposes a method to obtain probabilistic single-event unfolded distributions, together with their uncertainties and correlations, for the transfer-matrix--based unfolding. The algorithm is first validated on a toy model and then applied to pseudo-data for the $pp\rightarrow Zγγ$ process. In both examples the performance is compared to the single-event unfolding of the Machine-Learning--based Iterative cINN unfolding (IcINN).
△ Less
Submitted 25 October, 2023;
originally announced October 2023.
-
Uncertainty components in profile likelihood fits
Authors:
Andrés Pinto,
Zhibo Wu,
Fabrice Balli,
Nicolas Berger,
Maarten Boonekamp,
Émilien Chapon,
Tatsuo Kawamoto,
Bogdan Malaescu
Abstract:
When a measurement of a physical quantity is reported, the total uncertainty is usually decomposed into statistical and systematic uncertainties. This decomposition is not only useful to understand the contributions to the total uncertainty, but also required to propagate these contributions in subsequent analyses, such as combinations or interpretation fits including results from other measuremen…
▽ More
When a measurement of a physical quantity is reported, the total uncertainty is usually decomposed into statistical and systematic uncertainties. This decomposition is not only useful to understand the contributions to the total uncertainty, but also required to propagate these contributions in subsequent analyses, such as combinations or interpretation fits including results from other measurements or experiments. In profile-likelihood fits, widely applied in high-energy physics analyses, contributions of systematic uncertainties are routinely quantified using "impacts", which are not adequate for such applications. We discuss the difference between impacts and actual uncertainty components, and establish methods to determine the latter in a wide range of statistical models.
△ Less
Submitted 14 March, 2024; v1 submitted 8 July, 2023;
originally announced July 2023.
-
An unfolding method based on conditional Invertible Neural Networks (cINN) using iterative training
Authors:
Mathias Backes,
Anja Butter,
Monica Dunford,
Bogdan Malaescu
Abstract:
The unfolding of detector effects is crucial for the comparison of data to theory predictions. While traditional methods are limited to representing the data in a low number of dimensions, machine learning has enabled new unfolding techniques while retaining the full dimensionality. Generative networks like invertible neural networks~(INN) enable a probabilistic unfolding, which map individual eve…
▽ More
The unfolding of detector effects is crucial for the comparison of data to theory predictions. While traditional methods are limited to representing the data in a low number of dimensions, machine learning has enabled new unfolding techniques while retaining the full dimensionality. Generative networks like invertible neural networks~(INN) enable a probabilistic unfolding, which map individual events to their corresponding unfolded probability distribution. The accuracy of such methods is however limited by how well simulated training samples model the actual data that is unfolded. We introduce the iterative conditional INN~(IcINN) for unfolding that adjusts for deviations between simulated training samples and data. The IcINN unfolding is first validated on toy data and then applied to pseudo-data for the $pp \to Z γγ$ process.
△ Less
Submitted 10 January, 2024; v1 submitted 16 December, 2022;
originally announced December 2022.
-
Presenting Unbinned Differential Cross Section Results
Authors:
Miguel Arratia,
Anja Butter,
Mario Campanelli,
Vincent Croft,
Aishik Ghosh,
Dag Gillberg,
Kristin Lohwasser,
Bogdan Malaescu,
Vinicius Mikuni,
Benjamin Nachman,
Juan Rojo,
Jesse Thaler,
Ramon Winterhalder
Abstract:
Machine learning tools have empowered a qualitatively new way to perform differential cross section measurements whereby the data are unbinned, possibly in many dimensions. Unbinned measurements can enable, improve, or at least simplify comparisons between experiments and with theoretical predictions. Furthermore, many-dimensional measurements can be used to define observables after the measuremen…
▽ More
Machine learning tools have empowered a qualitatively new way to perform differential cross section measurements whereby the data are unbinned, possibly in many dimensions. Unbinned measurements can enable, improve, or at least simplify comparisons between experiments and with theoretical predictions. Furthermore, many-dimensional measurements can be used to define observables after the measurement instead of before. There is currently no community standard for publishing unbinned data. While there are also essentially no measurements of this type public, unbinned measurements are expected in the near future given recent methodological advances. The purpose of this paper is to propose a scheme for presenting and using unbinned results, which can hopefully form the basis for a community standard to allow for integration into analysis workflows. This is foreseen to be the start of an evolving community dialogue, in order to accommodate future developments in this field that is rapidly evolving.
△ Less
Submitted 17 November, 2021; v1 submitted 27 September, 2021;
originally announced September 2021.
-
Evaluation of the Strong Coupling Constant alpha_s Using the ATLAS Inclusive Jet Cross-Section Data
Authors:
Bogdan Malaescu,
Pavel Starovoitov
Abstract:
We perform a determination of the strong coupling constant using the latest ATLAS inclusive jet cross section data, from proton-proton collisions at sqrt{s}=7 TeV, and their full information on the bin-to-bin correlations. Several procedures for combining the statistical information from the different data inputs are studied and compared. The theoretical prediction is obtained using NLO QCD, and i…
▽ More
We perform a determination of the strong coupling constant using the latest ATLAS inclusive jet cross section data, from proton-proton collisions at sqrt{s}=7 TeV, and their full information on the bin-to-bin correlations. Several procedures for combining the statistical information from the different data inputs are studied and compared. The theoretical prediction is obtained using NLO QCD, and it also includes non-perturbative corrections. Our determination uses inputs with transverse momenta between 45 and 600 GeV, the running of the strong coupling being also tested in this range. Good agreement is observed when comparing our result with the world average at the Z-boson scale, as well as with the most recent results from the Tevatron.
△ Less
Submitted 3 July, 2012; v1 submitted 24 March, 2012;
originally announced March 2012.
-
An Iterative, Dynamically Stabilized(IDS) Method of Data Unfolding
Authors:
Bogdan Malaescu
Abstract:
We describe an iterative unfolding method for experimental data, making use of a regularization function. The use of this function allows one to build an improved normalization procedure for Monte Carlo spectra, unbiased by the presence of possible new structures in data. We unfold, in a dynamically stable way, data spectra which can be strongly affected by fluctuations in the background subtracti…
▽ More
We describe an iterative unfolding method for experimental data, making use of a regularization function. The use of this function allows one to build an improved normalization procedure for Monte Carlo spectra, unbiased by the presence of possible new structures in data. We unfold, in a dynamically stable way, data spectra which can be strongly affected by fluctuations in the background subtraction and simultaneously reconstruct structures which were not initially simulated.
△ Less
Submitted 15 June, 2011;
originally announced June 2011.
-
An iterative, dynamically stabilized method of data unfolding
Authors:
Bogdan Malaescu
Abstract:
We propose a new iterative unfolding method for experimental data, making use of a regularization function. The use of this function allows one to build an improved normalization procedure for Monte Carlo spectra, unbiased by the presence of possible new structures in data. We are able to unfold, in a dynamically stable way, data spectra which can be strongly affected by fluctuations in the back…
▽ More
We propose a new iterative unfolding method for experimental data, making use of a regularization function. The use of this function allows one to build an improved normalization procedure for Monte Carlo spectra, unbiased by the presence of possible new structures in data. We are able to unfold, in a dynamically stable way, data spectra which can be strongly affected by fluctuations in the background subtraction and simultaneously reconstruct structures which were not initially simulated. This method also allows one to control the amount of correlations introduced between the bins of the unfolded spectrum, when the transfers of events correcting the systematic detector effects are performed.
△ Less
Submitted 22 July, 2009;
originally announced July 2009.
