Zenodo (CERN European Organization for Nuclear Research), Jun 9, 2023
In this work, we present a quantum walks based approach for Probabilistic Safety Assessment probl... more In this work, we present a quantum walks based approach for Probabilistic Safety Assessment problems in the dynamic Markovian framework. After presenting well known quantum walk algorithms for detecting or finding marked elements in a graph or for finding paths in specific graphs, we propose two algorithms to address the more general problem of finding paths from some point to marked elements. The first algorithm is based on move and store strategy to built failure sequences from an initial state to the marked states. It uses a full register assembling qubits representing components of the system under study and other control and ancilla qubits (between n and 2n) to identify all the sequences. The second algorithm is based on a hybrid approach to consider different cascading circuits for solving relatively large instances. We present some tests on a Qiskit simulation library that were performed to compare our algorithms against classical random walks. These tests showed that classical approach has advances in the first iteration but the hybrid approach scales better in the sense that it identifies a large number of paths and converges towards all the possible paths to these marked vertices.
HAL (Le Centre pour la Communication Scientifique Directe), Oct 11, 2022
Les Etudes Probabilistes de Sûreté semblent faire partie des problèmes à forte combinatoire pouva... more Les Etudes Probabilistes de Sûreté semblent faire partie des problèmes à forte combinatoire pouvant être un candidat intéressant pour le calcul quantique. Dans ce papier, nous donnons une perspective de ce que pourraient devenir les algorithmes de calculs dans le domaine des EPS. Différentes classes d'algorithmes quantiques pourraient être envisagées dans ce sens et contribuer non seulement changer notre façon de calculer les métriques de sûreté de fonctionnement dans de grands modèles, mais aussi impacter les applications industrielles supportées par ces méthodes. Mots clés-Calcul Quantique, Fiabilité, Etudes probabilistes de sûreté Probabilistic Safety Studies seem to be among the combinatorial problems that can be a good candidate for quantum computing. In this paper, we give a perspective of what could become of the algorithms in the field of PSA. Different classes of quantum algorithms could be envisaged in this direction and may contribute not only to change the way we calculate the main safety metrics using large models but also have an impact on the industrial applications supported by these methods.
HAL (Le Centre pour la Communication Scientifique Directe), Oct 11, 2022
Plusieurs processus d'ingenierie et de supports de connaissance étant maintenant digitalisés, les... more Plusieurs processus d'ingenierie et de supports de connaissance étant maintenant digitalisés, les données numériques ou textuelles peuvent être extraites de la documentation avec une assez bonne qualité. Dans cette perspective, EDF et Aristé ont décidé de développer un automate pour importer les données des schémas mécaniques pour une utilisation directe dans un outil expert pour les études de fiabilité Knowledge Based Trees (KB3) dans le but d'effectuer des études de fiabilité système. L'algorithme que nous proposons, peut importer des schémas de deux outils de Conception Assitée par Ordinateur (CAO) utilisés dans l'industrie nucléaire pour représenter les systèmes thermohydrauliques. Une étude en cours considère un outil CAO pour les systèmes de controle commande (CC). Les premiers tests sont très encourageants, ce programme permet d'améliorer l a qualité des études de fiabilité mais surtout les effors d'ingénierie pour la saisie des systèmes en question. D'autre part, on peut utiliser des ajustments pour tenir compte des besoins spécifiques des projets. Une nouvelle étude est prévue pour envisager les systèmes électriques après les systèmes CC. Mots Clés-Etudes probabilistes de sûreté, KB3, Système Expert, Data Mining, Automatisation. Abstract-Now, that many engineering processes and documents are digitalized, numeric data can be extracted from design documentation with quite good quality. In this new environment, EDF and Aristé decided to develop an automaton for system design import and enrichment in KB3, directly from the design information system to get relevant data and different topological elements. We propose an algorithm to import system design from two CAO tools used in the Nuclear Power Point industry for hydraulic systems. An ongoing study is considering Computer-Aided Design (CAD) tool for Intrumentation and Control systems (I&C). Several systems have been successfully tested but a need emerged to refine the sensitivity to data to keep versus data to erase regarding different assumptions (e.g. for some hazards). This program will hopefully enhance the quality of reliability studies in addition to a significant reduction of engineering efforts. Different adjustments and customization are performed depending on the projects before going into production. A new study is planned to examine the feasibility on electrical systems after I&C.
HAL (Le Centre pour la Communication Scientifique Directe), Oct 11, 2022
HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific re... more HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. Mise à jour de l'état de l'art sur les méthodes et outils innovants pour le traitement des systèmes complexes et benchmarking
Ce travail propose une modélisation adéquate en graphes pour les systèmes et séquences accidentel... more Ce travail propose une modélisation adéquate en graphes pour les systèmes et séquences accidentelles de sûreté nucléaire. Ces systèmes et séquences proviennent des "Etudes Probabilistes de Sûreté" (EPS) qui consistent à analyser de façon exhaustive tous les scénarios accidentels envisageables, d'estimer leurs probabilités d'occurrence (en les regroupant par famille) et les conséquences associées. Ensuite une analyse des réseaux complexes résultants est effectuée par des mesures de centralités. Une première application consiste à la prédiction du Facteur d'Accroissement du Risque nucléaire en utilisant les algorithmes d'apprentissages supervisé : méthodes à base d'arbre de classification, régression logistique et méthodes ensemblistes, sur des données déséquilibrées. Par ailleurs, un nouveau coefficient synthétique de centralité et une mesure de similarité sont conçus pour comparer les structures de réseaux, indépendamment de leurs caractéristiques topologiques, en se basant sur les interdépendances entre leurs vecteurs de centralités. Cette nouvelle approche utilise des techniques statistiques (échantillonnage, corrélation et homogénéité). La pertinence et l'efficacité de cette nouvelle mesure de similarité sont validées sur le clustering de graphes théoriques classiques et la prédiction du type de graphes. Enfin, une application de cette approche est réalisée pour le clustering des réseaux complexes des systèmes de sûreté nucléaire.
HAL (Le Centre pour la Communication Scientifique Directe), Oct 11, 2022
préparer une version quantique qui permet de résoudre ce problème dans l'attente de la miseà l'éc... more préparer une version quantique qui permet de résoudre ce problème dans l'attente de la miseà l'échelle des ordinateurs quantiques. Abstract-Finding the failure scenarios of a system is a wellknown problem in the industrial world, and it can be solved by searching for the paths in the state graph of this system. The difficulty is that we do not search for only one or a part of these paths, we search for all of them. In this paper, we propose our quantum approach to find all the paths in this state graph between a source vertex and the marker vertices. Our approach is based on quantum walks to determine all the possible paths instead of treating them one by one classically, also our approach is able to determine all these paths in the same time with only ⌈ K×n ln 2 ⌉ + n qubits and n+2 oracles for the case where we try to determine only the scenarios with failures and the scenarios with the possibility of only one repair during the time period. We tested our approach on the state graph of a small system of four components and were able to find all the failure scenarios of this system. Unfortunately, we cannot test on large systems due to the number of qubits available in quantum computers today, which limits this work in the use case. The goal of this paper is to prepare a quantum version that solves this problem while waiting for the scaling up of quantum computers.
Proceedings of the 3rd International Conference on Complexity, Future Information Systems and Risk, 2018
We explore applying network based metrics to predict safety metrics of components in Nuclear Powe... more We explore applying network based metrics to predict safety metrics of components in Nuclear Power Plants (NPP). We first show how to model accident sequences as complex networks, then we conduct a statistical study over the main network metrics to show that these are highly correlated with the RIF (Risk Increase Factor) which is a very popular metric in nuclear safety studies.
Decidabilite et complexite de systemes de contraintes ensemblistes
Les resultats de recherche presentes dans cette these concernent l'etude de la decidabilite e... more Les resultats de recherche presentes dans cette these concernent l'etude de la decidabilite et de la complexite du probleme de satisfiabilite d'un systeme de contraintes ensemblistes. Cette these est composee de deux parties. Dans la premiere nous etudions le probleme de satisfiabilite des contraintes ensemblistes dans un univers heterogene d'ensembles hereditairement finis. Nous donnons certains resultats de complexite qui montrent 1) le caractere np-dur du probleme en question meme pour un fragment tres reduit, 2) le caractere np pour les fragments etudies, en utilisant des techniques connues mais avec une nette amelioration de la complexite calculatoire. Nous proposons ensuite deux approches pour resoudre le probleme de satisfiabilite. La premiere est basee sur une reduction polynomiale du probleme de satisfiabilite des formules ensemblistes a sat. La deuxieme est une reduction du probleme de satisfiabilite pour les formules conjonctives a un probleme equivalent de pr...
2022 Asia Conference on Algorithms, Computing and Machine Learning (CACML)
Finding the failure scenarios of a system is a very complex problem in the field of Probabilistic... more Finding the failure scenarios of a system is a very complex problem in the field of Probabilistic Safety Assessment (PSA). In order to solve this problem we will use the Hidden Quantum Markov Models (HQMMs) to create a generative model. Therefore, in this paper, we will study and compare the results of HQMMs and classical Hidden Markov Models HMM on a real datasets generated from real small systems in the field of PSA. As a quality metric we will use Description accuracy DA and we will show that the quantum approach gives better results compared with the classical approach, and we will give a strategy to identify the probable and noprobable failure scenarios of a system.
2021 IEEE 23rd Int Conf on High Performance Computing & Communications; 7th Int Conf on Data Science & Systems; 19th Int Conf on Smart City; 7th Int Conf on Dependability in Sensor, Cloud & Big Data Systems & Application (HPCC/DSS/SmartCity/DependSys)
In this paper we provide the quantum version of the Convex Non-negative Matrix Factorization algo... more In this paper we provide the quantum version of the Convex Non-negative Matrix Factorization algorithm (Convex-NMF) by using the D-wave quantum annealer. More precisely, we use D-wave 2000Q to find the low rank approximation of a fixed real-valued matrix X by the product of two non-negative matrices factors W and G such that the Frobenius norm of the difference X − XW G is minimized. In order to solve this optimization problem we proceed in two steps. In the first step we transform the global real optimization problem depending on W, G into two quadratic unconstrained binary optimization problems (QUBO) depending on W and G respectively. In the second step we use an alternative strategy between the two QUBO problems corresponding to W and G to find the global solution. The running of these two QUBO problems on D-wave 2000Q need to use an embedding to the chimera graph of D-wave 2000Q, this embedding is limited by the number of qubits of D-wave 2000Q. We perform a study on the maximum number of real data to be used by our approach on D-wave 2000Q. The proposed study is based on the number of qubits used to represent each real variable. We also tested our approach on D-Wave 2000Q with several randomly generated data sets to prove that our approach is faster than the classical approach and also to prove that it gets the best results.
Distance Estimation for Quantum Prototypes Based Clustering
Neural Information Processing, 2019
Quantum machine learning is a new area of research with the recent work on quantum versions of su... more Quantum machine learning is a new area of research with the recent work on quantum versions of supervised and unsupervised algorithms. In recent years, many quantum machine learning algorithms have been proposed providing a speed-up over the classical algorithms. In this paper, we propose an analysis and a comparison of three quantum distances for protoptypes-based clustering techniques. As an application of this work, we present a quantum K-means version which gives a good classification just like its classical version, the difference resides in the complexity: while the classical version of K-means takes polynomial time, the quantum version takes only logarithmic time especially in large datasets. Finally, we validate the benefits of the proposed approach by performing a series of empirical evaluations regarding the quantum distance estimation and its behavior versus the stability of finding the nearest centers in the right order.
Deciding in HFS-theory via linear integer programming
Logic Programming and Automated …, 1993
We give a reduction, in polynomialy bounded time, of the satisfiability problem in the set langua... more We give a reduction, in polynomialy bounded time, of the satisfiability problem in the set language MPLS, to an equivalent linear integer programming problem. We show that this provides an algorithm to decide the Set Unification Problem. The procedure performed ...
Uploads
Papers by Mohamed HIBTI