-
Multi-Level GNN Preconditioner for Solving Large Scale Problems
Authors:
Matthieu Nastorg,
Jean-Marc Gratien,
Thibault Faney,
Michele Alessandro Bucci,
Guillaume Charpiat,
Marc Schoenauer
Abstract:
Large-scale numerical simulations often come at the expense of daunting computations. High-Performance Computing has enhanced the process, but adapting legacy codes to leverage parallel GPU computations remains challenging. Meanwhile, Machine Learning models can harness GPU computations effectively but often struggle with generalization and accuracy. Graph Neural Networks (GNNs), in particular, ar…
▽ More
Large-scale numerical simulations often come at the expense of daunting computations. High-Performance Computing has enhanced the process, but adapting legacy codes to leverage parallel GPU computations remains challenging. Meanwhile, Machine Learning models can harness GPU computations effectively but often struggle with generalization and accuracy. Graph Neural Networks (GNNs), in particular, are great for learning from unstructured data like meshes but are often limited to small-scale problems. Moreover, the capabilities of the trained model usually restrict the accuracy of the data-driven solution. To benefit from both worlds, this paper introduces a novel preconditioner integrating a GNN model within a multi-level Domain Decomposition framework. The proposed GNN-based preconditioner is used to enhance the efficiency of a Krylov method, resulting in a hybrid solver that can converge with any desired level of accuracy. The efficiency of the Krylov method greatly benefits from the GNN preconditioner, which is adaptable to meshes of any size and shape, is executed on GPUs, and features a multi-level approach to enforce the scalability of the entire process. Several experiments are conducted to validate the numerical behavior of the hybrid solver, and an in-depth analysis of its performance is proposed to assess its competitiveness against a C++ legacy solver.
△ Less
Submitted 13 February, 2024;
originally announced February 2024.
-
Enhancing Data-Assimilation in CFD using Graph Neural Networks
Authors:
Michele Quattromini,
Michele Alessandro Bucci,
Stefania Cherubini,
Onofrio Semeraro
Abstract:
We present a novel machine learning approach for data assimilation applied in fluid mechanics, based on adjoint-optimization augmented by Graph Neural Networks (GNNs) models. We consider as baseline the Reynolds-Averaged Navier-Stokes (RANS) equations, where the unknown is the meanflow and a closure model based on the Reynolds-stress tensor is required for correctly computing the solution. An end-…
▽ More
We present a novel machine learning approach for data assimilation applied in fluid mechanics, based on adjoint-optimization augmented by Graph Neural Networks (GNNs) models. We consider as baseline the Reynolds-Averaged Navier-Stokes (RANS) equations, where the unknown is the meanflow and a closure model based on the Reynolds-stress tensor is required for correctly computing the solution. An end-to-end process is cast; first, we train a GNN model for the closure term. Second, the GNN model is introduced in the training process of data assimilation, where the RANS equations act as a physics constraint for a consistent prediction. We obtain our results using direct numerical simulations based on a Finite Element Method (FEM) solver; a two-fold interface between the GNN model and the solver allows the GNN's predictions to be incorporated into post-processing steps of the FEM analysis. The proposed scheme provides an excellent reconstruction of the meanflow without any features selection; preliminary results show promising generalization properties over unseen flow configurations.
△ Less
Submitted 29 November, 2023;
originally announced November 2023.
-
An Implicit GNN Solver for Poisson-like problems
Authors:
Matthieu Nastorg,
Michele Alessandro Bucci,
Thibault Faney,
Jean-Marc Gratien,
Guillaume Charpiat,
Marc Schoenauer
Abstract:
This paper presents $Ψ$-GNN, a novel Graph Neural Network (GNN) approach for solving the ubiquitous Poisson PDE problems with mixed boundary conditions. By leveraging the Implicit Layer Theory, $Ψ$-GNN models an "infinitely" deep network, thus avoiding the empirical tuning of the number of required Message Passing layers to attain the solution. Its original architecture explicitly takes into accou…
▽ More
This paper presents $Ψ$-GNN, a novel Graph Neural Network (GNN) approach for solving the ubiquitous Poisson PDE problems with mixed boundary conditions. By leveraging the Implicit Layer Theory, $Ψ$-GNN models an "infinitely" deep network, thus avoiding the empirical tuning of the number of required Message Passing layers to attain the solution. Its original architecture explicitly takes into account the boundary conditions, a critical prerequisite for physical applications, and is able to adapt to any initially provided solution. $Ψ$-GNN is trained using a "physics-informed" loss, and the training process is stable by design, and insensitive to its initialization. Furthermore, the consistency of the approach is theoretically proven, and its flexibility and generalization efficiency are experimentally demonstrated: the same learned model can accurately handle unstructured meshes of various sizes, as well as different boundary conditions. To the best of our knowledge, $Ψ$-GNN is the first physics-informed GNN-based method that can handle various unstructured domains, boundary conditions and initial solutions while also providing convergence guarantees.
△ Less
Submitted 26 March, 2024; v1 submitted 6 February, 2023;
originally announced February 2023.
-
Continuous Methods : Adaptively intrusive reduced order model closure
Authors:
Emmanuel Menier,
Michele Alessandro Bucci,
Mouadh Yagoubi,
Lionel Mathelin,
Thibault Dairay,
Raphael Meunier,
Marc Schoenauer
Abstract:
Reduced order modeling methods are often used as a mean to reduce simulation costs in industrial applications. Despite their computational advantages, reduced order models (ROMs) often fail to accurately reproduce complex dynamics encountered in real life applications. To address this challenge, we leverage NeuralODEs to propose a novel ROM correction approach based on a time-continuous memory for…
▽ More
Reduced order modeling methods are often used as a mean to reduce simulation costs in industrial applications. Despite their computational advantages, reduced order models (ROMs) often fail to accurately reproduce complex dynamics encountered in real life applications. To address this challenge, we leverage NeuralODEs to propose a novel ROM correction approach based on a time-continuous memory formulation. Finally, experimental results show that our proposed method provides a high level of accuracy while retaining the low computational costs inherent to reduced models.
△ Less
Submitted 30 November, 2022;
originally announced November 2022.
-
DS-GPS : A Deep Statistical Graph Poisson Solver (for faster CFD simulations)
Authors:
Matthieu Nastorg,
Marc Schoenauer,
Guillaume Charpiat,
Thibault Faney,
Jean-Marc Gratien,
Michele-Alessandro Bucci
Abstract:
This paper proposes a novel Machine Learning-based approach to solve a Poisson problem with mixed boundary conditions. Leveraging Graph Neural Networks, we develop a model able to process unstructured grids with the advantage of enforcing boundary conditions by design. By directly minimizing the residual of the Poisson equation, the model attempts to learn the physics of the problem without the ne…
▽ More
This paper proposes a novel Machine Learning-based approach to solve a Poisson problem with mixed boundary conditions. Leveraging Graph Neural Networks, we develop a model able to process unstructured grids with the advantage of enforcing boundary conditions by design. By directly minimizing the residual of the Poisson equation, the model attempts to learn the physics of the problem without the need for exact solutions, in contrast to most previous data-driven processes where the distance with the available solutions is minimized.
△ Less
Submitted 21 November, 2022;
originally announced November 2022.
-
Machine Learning model for gas-liquid interface reconstruction in CFD numerical simulations
Authors:
Tamon Nakano,
Alessandro Michele Bucci,
Jean-Marc Gratien,
Thibault Faney,
Guillaume Charpiat
Abstract:
The volume of fluid (VoF) method is widely used in multi-phase flow simulations to track and locate the interface between two immiscible fluids. A major bottleneck of the VoF method is the interface reconstruction step due to its high computational cost and low accuracy on unstructured grids. We propose a machine learning enhanced VoF method based on Graph Neural Networks (GNN) to accelerate the i…
▽ More
The volume of fluid (VoF) method is widely used in multi-phase flow simulations to track and locate the interface between two immiscible fluids. A major bottleneck of the VoF method is the interface reconstruction step due to its high computational cost and low accuracy on unstructured grids. We propose a machine learning enhanced VoF method based on Graph Neural Networks (GNN) to accelerate the interface reconstruction on general unstructured meshes. We first develop a methodology to generate a synthetic dataset based on paraboloid surfaces discretized on unstructured meshes. We then train a GNN based model and perform generalization tests. Our results demonstrate the efficiency of a GNN based approach for interface reconstruction in multi-phase flow simulations in the industrial context.
△ Less
Submitted 12 July, 2022;
originally announced July 2022.
-
Continuous Methods : Hamiltonian Domain Translation
Authors:
Emmanuel Menier,
Michele Alessandro Bucci,
Mouadh Yagoubi,
Lionel Mathelin,
Marc Schoenauer
Abstract:
This paper proposes a novel approach to domain translation. Leveraging established parallels between generative models and dynamical systems, we propose a reformulation of the Cycle-GAN architecture. By embedding our model with a Hamiltonian structure, we obtain a continuous, expressive and most importantly invertible generative model for domain translation.
This paper proposes a novel approach to domain translation. Leveraging established parallels between generative models and dynamical systems, we propose a reformulation of the Cycle-GAN architecture. By embedding our model with a Hamiltonian structure, we obtain a continuous, expressive and most importantly invertible generative model for domain translation.
△ Less
Submitted 8 July, 2022;
originally announced July 2022.
-
CD-ROM: Complemented Deep-Reduced Order Model
Authors:
Emmanuel Menier,
Michele Alessandro Bucci,
Mouadh Yagoubi,
Lionel Mathelin,
Marc Schoenauer
Abstract:
Model order reduction through the POD-Galerkin method can lead to dramatic gains in terms of computational efficiency in solving physical problems. However, the applicability of the method to non linear high-dimensional dynamical systems such as the Navier-Stokes equations has been shown to be limited, producing inaccurate and sometimes unstable models. This paper proposes a deep learning based cl…
▽ More
Model order reduction through the POD-Galerkin method can lead to dramatic gains in terms of computational efficiency in solving physical problems. However, the applicability of the method to non linear high-dimensional dynamical systems such as the Navier-Stokes equations has been shown to be limited, producing inaccurate and sometimes unstable models. This paper proposes a deep learning based closure modeling approach for classical POD-Galerkin reduced order models (ROM). The proposed approach is theoretically grounded, using neural networks to approximate well studied operators. In contrast with most previous works, the present CD-ROM approach is based on an interpretable continuous memory formulation, derived from simple hypotheses on the behavior of partially observed dynamical systems. The final corrected models can hence be simulated using most classical time step** schemes. The capabilities of the CD-ROM approach are demonstrated on two classical examples from Computational Fluid Dynamics, as well as a parametric case, the Kuramoto-Sivashinsky equation.
△ Less
Submitted 2 May, 2023; v1 submitted 22 February, 2022;
originally announced February 2022.
-
Greedy palindromic lengths
Authors:
Michelangelo Bucci,
Gwenaël Richomme
Abstract:
In [A. Frid, S. Puzynina, L.Q. Zamboni, \textit{On palindromic factorization of words}, Adv. in Appl. Math. 50 (2013), 737-748], it was conjectured that any infinite word whose palindromic lengths of factors are bounded is ultimately periodic. We introduce variants of this conjecture and prove this conjecture in particular cases. Especially we introduce left and right greedy palindromic lengths. T…
▽ More
In [A. Frid, S. Puzynina, L.Q. Zamboni, \textit{On palindromic factorization of words}, Adv. in Appl. Math. 50 (2013), 737-748], it was conjectured that any infinite word whose palindromic lengths of factors are bounded is ultimately periodic. We introduce variants of this conjecture and prove this conjecture in particular cases. Especially we introduce left and right greedy palindromic lengths. These lengths are always greater than or equals to the initial palindromic length. When the greedy left (or right) palindromic lengths of prefixes of a word are bounded then this word is ultimately periodic.
△ Less
Submitted 17 June, 2016;
originally announced June 2016.
-
Aperiodic pseudorandom number generators based on infinite words
Authors:
Lubomira Balkova,
Michelangelo Bucci,
Alessandro De Luca,
Jiri Hladky,
Svetlana Puzynina
Abstract:
In this paper we study how certain families of aperiodic infinite words can be used to produce aperiodic pseudorandom number generators (PRNGs) with good statistical behavior. We introduce the \emph{well distributed occurrences} (WELLDOC) combinatorial property for infinite words, which guarantees absence of the lattice structure defect in related pseudorandom number generators. An infinite word…
▽ More
In this paper we study how certain families of aperiodic infinite words can be used to produce aperiodic pseudorandom number generators (PRNGs) with good statistical behavior. We introduce the \emph{well distributed occurrences} (WELLDOC) combinatorial property for infinite words, which guarantees absence of the lattice structure defect in related pseudorandom number generators. An infinite word $u$ on a $d$-ary alphabet has the WELLDOC property if, for each factor $w$ of $u$, positive integer $m$, and vector $\mathbf v\in\mathbb Z_{m}^{d}$, there is an occurrence of $w$ such that the Parikh vector of the prefix of $u$ preceding such occurrence is congruent to $\mathbf v$ modulo $m$. (The Parikh vector of a finite word $v$ over an alphabet $\mathcal A$ has its $i$-th component equal to the number of occurrences of the $i$-th letter of $\mathcal A$ in $v$.) We prove that Sturmian words, and more generally Arnoux-Rauzy words and some morphic images of them, have the WELLDOC property. Using the TestU01 and PractRand statistical tests, we moreover show that not only the lattice structure is absent, but also other important properties of PRNGs are improved when linear congruential generators are combined using infinite words having the WELLDOC property.
△ Less
Submitted 1 December, 2014; v1 submitted 23 November, 2013;
originally announced November 2013.
-
Reversible Christoffel factorizations
Authors:
Michelangelo Bucci,
Alessandro De Luca,
Luca Q. Zamboni
Abstract:
We define a family of natural decompositions of Sturmian words in Christoffel words, called *reversible Christoffel* (RC) factorizations. They arise from the observation that two Sturmian words with the same language have (almost always) arbitrarily long Abelian equivalent prefixes. Using the three gap theorem, we prove that in each RC factorization, only 2 or 3 distinct Christoffel words may occu…
▽ More
We define a family of natural decompositions of Sturmian words in Christoffel words, called *reversible Christoffel* (RC) factorizations. They arise from the observation that two Sturmian words with the same language have (almost always) arbitrarily long Abelian equivalent prefixes. Using the three gap theorem, we prove that in each RC factorization, only 2 or 3 distinct Christoffel words may occur. We begin the study of such factorizations, considered as infinite words over 2 or 3 letters, and show that in the general case they are either Sturmian words, or obtained by a three-interval exchange transformation.
△ Less
Submitted 11 July, 2013; v1 submitted 13 November, 2012;
originally announced November 2012.
-
Some characterizations of Sturmian words in terms of the lexicographic order
Authors:
Michelangelo Bucci,
Alessandro De Luca,
Luca Q. Zamboni
Abstract:
In this paper we present three new characterizations of Sturmian words based on the lexicographic ordering of their factors.
In this paper we present three new characterizations of Sturmian words based on the lexicographic ordering of their factors.
△ Less
Submitted 27 May, 2012;
originally announced May 2012.
-
Enumeration and Structure of Trapezoidal Words
Authors:
Michelangelo Bucci,
Alessandro De Luca,
Gabriele Fici
Abstract:
Trapezoidal words are words having at most $n+1$ distinct factors of length $n$ for every $n\ge 0$. They therefore encompass finite Sturmian words. We give combinatorial characterizations of trapezoidal words and exhibit a formula for their enumeration. We then separate trapezoidal words into two disjoint classes: open and closed. A trapezoidal word is closed if it has a factor that occurs only as…
▽ More
Trapezoidal words are words having at most $n+1$ distinct factors of length $n$ for every $n\ge 0$. They therefore encompass finite Sturmian words. We give combinatorial characterizations of trapezoidal words and exhibit a formula for their enumeration. We then separate trapezoidal words into two disjoint classes: open and closed. A trapezoidal word is closed if it has a factor that occurs only as a prefix and as a suffix; otherwise it is open. We investigate open and closed trapezoidal words, in relation with their special factors. We prove that Sturmian palindromes are closed trapezoidal words and that a closed trapezoidal word is a Sturmian palindrome if and only if its longest repeated prefix is a palindrome. We also define a new class of words, \emph{semicentral words}, and show that they are characterized by the property that they can be written as $uxyu$, for a central word $u$ and two different letters $x,y$. Finally, we investigate the prefixes of the Fibonacci word with respect to the property of being open or closed trapezoidal words, and show that the sequence of open and closed prefixes of the Fibonacci word follows the Fibonacci sequence.
△ Less
Submitted 29 October, 2012; v1 submitted 6 March, 2012;
originally announced March 2012.
-
A new characteristic property of rich words
Authors:
Michelangelo Bucci,
Alessandro De Luca,
Amy Glen,
Luca Q. Zamboni
Abstract:
Originally introduced and studied by the third and fourth authors together with J. Justin and S. Widmer in arXiv:0801.1656, rich words constitute a new class of finite and infinite words characterized by containing the maximal number of distinct palindromes. Several characterizations of rich words have already been established. A particularly nice characteristic property is that all 'complete re…
▽ More
Originally introduced and studied by the third and fourth authors together with J. Justin and S. Widmer in arXiv:0801.1656, rich words constitute a new class of finite and infinite words characterized by containing the maximal number of distinct palindromes. Several characterizations of rich words have already been established. A particularly nice characteristic property is that all 'complete returns' to palindromes are palindromes. In this note, we prove that rich words are also characterized by the property that each factor is uniquely determined by its longest palindromic prefix and its longest palindromic suffix.
△ Less
Submitted 15 July, 2008;
originally announced July 2008.
-
A connection between palindromic and factor complexity using return words
Authors:
Michelangelo Bucci,
Alessandro De Luca,
Amy Glen,
Luca Q. Zamboni
Abstract:
In this paper we prove that for any infinite word W whose set of factors is closed under reversal, the following conditions are equivalent:
(I) all complete returns to palindromes are palindromes;
(II) P(n) + P(n+1) = C(n+1) - C(n) + 2 for all n, where P (resp. C) denotes the palindromic complexity (resp. factor complexity) function of W, which counts the number of distinct palindromic facto…
▽ More
In this paper we prove that for any infinite word W whose set of factors is closed under reversal, the following conditions are equivalent:
(I) all complete returns to palindromes are palindromes;
(II) P(n) + P(n+1) = C(n+1) - C(n) + 2 for all n, where P (resp. C) denotes the palindromic complexity (resp. factor complexity) function of W, which counts the number of distinct palindromic factors (resp. factors) of each length in W.
△ Less
Submitted 11 April, 2008; v1 submitted 10 February, 2008;
originally announced February 2008.
