About me[up]
I studied physics with specializations in computer science and computational statistics at the Vienna University of Technology where I also did my master thesis and PhD thesis. My master thesis dealt with the extension of a Bayesian nuclear data evaluation procedure to enable the inclusion of a new type of observable, i.e. angular differential cross sections. On the way I also used statistical tools such as clustering via multivariate normal mixture models and principal component analysis to learn about the model features and to what extent they are preserved in the Bayesian evaluation procedure. In my PhD thesis I researched how Bayesian evaluation methods can be made more robust against potential misspecifications, which I solved by introducing Gaussian processes to capture model imperfections. Because nuclear models are computationally expensive and produce a lot of data, I also researched how a specific Bayesian evaluation procedure can be made more efficient and found a solution by exploiting the inherent sparse structure of the problem. As a research scholar at CEA Saclay in France and then at Uppsala university in Sweden, I continued this line of research, improving computational aspects and methodology to handle and assess nuclear data and models. At present, I am working at the IAEA dealing with nuclear data development projects, which encompasses both the development of computer codes and databases as well as their verification. A more detailed account of my education and career can be found in my CV.
In my spare time, I am passionate about language learning, learning about methods and applications of machine learning outside nuclear physics, and computer science. Whenever I can, I try to bring together activities in these different domains of interest. One publicly visible result is my Ukrainian course on memrise, which has been created semi-automatically using Python scripts to scrape information from various web resources, merge them, and upload them to the learning platform. Some of my programming activity can be followed on GitHub.
I am the maintainer of www.nucleardata.com where I provide software and information on the topic of nuclear data evaluation.
Demonstrations[up]
- Separation of different contributions in a detector output [link]
- In experimental nuclear physics during data analysis, one is facing the task to split the measured signal of a particle detector in several contributing components in order to extract the relevant one. The relevant contribution is given by the particles being produced in a controlled way within the experiment whereas the other contributions are associated with particles entering the detector from the outside (e.g., from space) or particles that are produced in the experiment but whose impact on the detector output is undesired. This interactive demonstration showcases the separation of two bivariate normal peaks from uniform background noise either by the expectation-maximization algorithm or Gibbs sampling.
- Impact of non-linearities in nuclear data evaluation [link]
- One common approach to perform nuclear data evaluation is to use a special case of the Kalman filter. The Kalman filter (in its basic form) assumes a linear mapping between the state and the observation and therefore nuclear models have to be replaced by a first-order Taylor approximation for the procedure. The linear approximation of non-linear nuclear models can lead to strongly distorted results. This demonstration enables the interactive exploration of the magnitude of such distortions at the example of the nuclear model code TALYS and the neutron-induced total cross section of 181Ta.
Publications[up]
Papers & proceedings
| G. Schnabel, R. Capote, A.J. Koning, D.A. Brown | Nuclear data evaluation with Bayesian networks | arXiv | [link] |
| G. Schnabel, H. Sjöstrand, J. Hansson, D. Rochman, A. Koning, R. Capote | Conception and software implementation of a nuclear data evaluation pipeline | NDS | [link] [preprint] |
| D. Siefman, M. Hursin, H. Sjöstrand, G. Schnabel, D. Rochman, A. Pautz | Data assimilation of post-irradiation examination data for fission yields from GEF | EPJ-N | [link] |
| J. Hirtz, J-C. David, A. Boudard, J. Cugnon, S. Leray, I. Leya, J.L. Rodriguez-Sánchez, G. Schnabel | Strangeness production in the new version of the Liège Intra-Nuclear Cascade model | PRC | [link] |
| R. Capote, S. Badikov, A.D. Carlson, I. Duran, F. Gunsing, D. Neudecker, V.G. Pronyaev, P. Schillebeeckx, G. Schnabel, D.L. Smith, A. Wallner | Unrecognized Sources of Uncertainties (USU) in Experimental Nuclear Data | NDS | [link] [preprint] |
| G. Schnabel | A computational EXFOR database | ND2019 | [link] |
| G. Schnabel, H. Sjöstrand | A first sketch: Construction of model defect priors inspired by dynamic time warping | Wonder-2018 | [link] |
| H. Sjöstrand, G. Schnabel | Monte Carlo integral adjustment of nuclear data libraries – experimental covariances and inconsistent data | Wonder-2018 | [link] |
| J-C. David, J. Hirtz, R-S. J. Luis, A. Boudard, J. Cugnon, S. Leray, I. Leya, D. Mancusi, G. Schnabel | Production of Strange Particles and Hypernuclei in Nuclear Reactions at a Few GeV. New Capabilities in the INCL Intranuclear Cascade Model | HYP2018 | [link] |
| G. Schnabel | Estimating model bias over the complete nuclide chart with sparse Gaussian processes at the example of INCL/ABLA and double-differential neutron spectra | EPJ-N | [link] |
| G. Schnabel | Fitting and Analysis Technique for Inconsistent Nuclear Data | MC2017 | [link] |
| G. Schnabel, H. Leeb | A modified Generalized Least Squares method for large scale nuclear data evaluation | NIMA | [link] |
| G. Schnabel | Adaptive Monte Carlo for nuclear data evaluation | ND2016 | [link] |
| G. Schnabel, H. Leeb | Differential Cross Sections and the Impact of Model Defects in Nuclear Data Evaluation | Wonder-2015 | [link] |
| H. Leeb, G. Schnabel, Th. Srdinko, V. Wildpaner | Bayesian Evaluation Including Covariance Matrices of Neutron-induced Reaction Cross Sections of 181Ta | NDS | [link] |
| G. Schnabel, H. Leeb | A New Module for Large Scale Bayesian Evaluation in the Fast Neutron Energy Region | NDS | [link] |
| H. Leeb, T. Srdinko, G. Schnabel, D.M. Warji | Microscopic Approach to Alpha-Nucleus Optical Potentials for Nucleosynthesis | XIII Nuclei in the Cosmos | [link] |
| H. Leeb, G. Schnabel, Th. Srdinko | What is the proper evaluation method | NEMEA-7 | [link] |
Reports
| 4/2018 | CHANDA | Task 11.5 | Development of methods for the uncertainty quantification of high-energy spallation models | [pdf] |
| 4/2015 | KKKÖ | IPN-2013-7 | Development of a method to account for model deficiencies in nuclear data evaluations | [pdf] |
| 7/2014 | F4E | F4E-FPA 168 specific grant 1 | Production of an ENDF file for neutron-induced reactions of 181Ta on the basis of experiment data from the EXFOR database and predictions from the model code TALYS | [pdf] |
Posters
| G. Schnabel | Machine Learning Summer School Skoltech 2020 | Identification of papers suitable for the nuclear reaction database EXFOR | [pdf] |
| G. Schnabel | Machine Learning Summer School Moscow 2019 | Towards automatic dependency detection in nuclear databases | [pdf] |
| G. Schnabel | ISBA World Meeting 2018 | Bayesian Analysis with Stochastic Linear Models and Its Surprising Results | [pdf] |
Presentations[up]
| Pu9(n,f) cross section covariances including USU components | Mini-CSEWG | LLNL | 27 April 2023 | [pdf] |
| Towards a consistent evaluation of Ni isotopes using a Bayesian network | INDEN CM on Structural Materials | IAEA | 6 December 2022 | [pdf] |
| Update on the EXFOR parser for layer zero with new transformers | WPEC SG50 | OECD-NEA (online) | 5 December 2022 | [pdf] |
| About an ENDF parser with some applications | TM on Nuclear Data Processing | IAEA | 1 December 2022 | [pdf] |
| On a formal ENDF specification language | Annual CSEWG meeting | BNL (online) | 4 November 2022 | [pdf] |
| Nuclear data evaluation with Bayesian networks | 2022 BayesiaLab conference | BayesiaLab (online) | 27 October 2022 | [pdf] |
| An EXFOR parser prototype for layer 0 and about transformers | WPEC SG50 | Nuclear Energy Agency (online) | 4 May 2022 | [pdf] |
| Probabilistic Methods for Nuclear Data | ICTP/IAEA Workshop on Computational Science and Engineering | Trieste, Italy | 23 May 2022 | [pdf] |
| Outlier detection with Bayesian networks | WPEC SG50 | Nuclear Energy Agency (online) | 4 May 2022 | [pdf] |
| Automated construction of Bayesian networks for large scale evaluations | JEFF week | OECD-NEA (online) | 27 April 2022 | [pdf] |
| GMAP modernization and new possibilities | Technical Meeting on Neutron Data Standards | IAEA (online) | 6 December 2021 | [pdf] |
| Nuclear data evaluation with Bayesian networks | JEFF week | OECD-NEA (online) | 23 November 2021 | [pdf] |
| Nuclear data evaluation with Bayesian networks | New York Scientific Data Summit | BNL (online) | 29 October 2021 | [pdf] |
| Probabilistic methods for nuclear data | Workshop on Computational Science and Engineering | IAEA (online) | 13 July 2021 | [pdf] |
| Containerization and microservices for nuclear data | Workshop for Applied Nuclear Data Activities | (online) | 29 January 2021 | [pdf] |
| A computational EXFOR database in JSON: Migration from MongoDB to CouchDB | WPEC SG49 | Nuclear Energy Agency (online) | 13 May 2020 | [pdf] |
| A computational EXFOR database | Int. Conf. on Nuclear Data for Science and Technology | Beijing, China | 22 May 2019 | [pdf] |
| Interfacing TALYS with a Bayesian treatment of model defects and inconsistent data | Int. Conf. on Nuclear Data for Science and Technology | Beijing, China | 21 May 2019 | [pdf] |
| Bayesian fitting of multivariate mixture models via Gibbs sampling | TK talk, Uppsala University | Uppsala, Sweden | 10 Apr 2019 | [pdf] |
| Construction of model defect priors inspired by dynamic time warping | 5th Int. Workshp on Nuclear Data Evaluation for Reactor applications | Aix-en-Provence, France | 12 Oct 2018 | [pdf] |
| Towards a holistic framework for global assessments of nuclear models | seminar talk, CEA Saclay | Saclay, France | 20 Nov 2017 | [pdf] |
| Towards an automated prediction and uncertainty quantification system for nuclear models and nuclear data | Covariance Workshop | Aix-en-Provence, France | 3 Oct 2017 | [pdf] |
| Watch out! Laplace's demon lures us into the technical singularity | PHENIICS fest | Laboratoire de l'Accélérateur Linéaire, Orsay | 31 May 2017 | [pdf] |
| Fitting and Analysis Technique for Inconsistent Nuclear Data | Int. Conf. on Mathematics & Computational Methods Applied to Nuclear Science & Engineering | Jeju, South Korea | 16 Apr 2017 | [pdf] |
| Bayesian statistics, INCL, data evaluation, and lying parrots | PhD seminar, CEA Saclay | Saclay, France | 24 Mar 2017 | [pdf] |
| Adaptive Monte Carlo for Nuclear Data Evaluation | Int. Conf. on Nuclear Data for Science and Technology | Bruges, Belgium | 15 Sep 2016 | [pdf] |
| Large Scale Bayesian Nuclear Data Evaluation with Consistent Model Defects | seminar talk, CEA Saclay | Saclay, France | 2 Oct 2015 | [pdf] |
| Large Scale Bayesian Nuclear Data Evaluation with Consistent Model Defects | 3rd European Nuclear Physics Conference | Groningen, Netherlands | 3 Sep 2015 | [pdf] |
| Nuclear Data Evaluation with Consistent Model Defects | seminar talk, Technical University of Vienna | Vienna, Austria | 18 May 2015 | [pdf] |
| Evaluation of Nuclear Data with a Novel Approach | 1. Fusionstag, Graz | Graz, Austria | 21 Nov 2014 | [pdf] |
| Extending Bayesian Methods to Differential Angle Cross Sections and Spectra | Annual meeting of the Austrian Physical Society | Pöllau bei Hartberg, Austria | 26 Sep 2014 | [pdf] |
| A New Module for Large Scale Bayesian Nuclear Data Evaluation | Int. Conf. on Nuclear Data Covariances | Santa Fe, New Mexico | 30 Apr 2014 | [pdf] |