-
Deep unrolling for learning optimal spatially varying regularisation parameters for Total Generalised Variation
Authors:
Thanh Trung Vu,
Andreas Kofler,
Kostas Papafitsoros
Abstract:
We extend a recently introduced deep unrolling framework for learning spatially varying regularisation parameters in inverse imaging problems to the case of Total Generalised Variation (TGV). The framework combines a deep convolutional neural network (CNN) inferring the two spatially varying TGV parameters with an unrolled algorithmic scheme that solves the corresponding variational problem. The t…
▽ More
We extend a recently introduced deep unrolling framework for learning spatially varying regularisation parameters in inverse imaging problems to the case of Total Generalised Variation (TGV). The framework combines a deep convolutional neural network (CNN) inferring the two spatially varying TGV parameters with an unrolled algorithmic scheme that solves the corresponding variational problem. The two subnetworks are jointly trained end-to-end in a supervised fashion and as such the CNN learns to compute those parameters that drive the reconstructed images as close to the ground truth as possible. Numerical results in image denoising and MRI reconstruction show a significant qualitative and quantitative improvement compared to the best TGV scalar parameter case as well as to other approaches employing spatially varying parameters computed by unsupervised methods. We also observe that the inferred spatially varying parameter maps have a consistent structure near the image edges, asking for further theoretical investigations. In particular, the parameter that weighs the first-order TGV term has a triple-edge structure with alternating high-low-high values whereas the one that weighs the second-order term attains small values in a large neighbourhood around the edges.
△ Less
Submitted 6 March, 2025; v1 submitted 23 February, 2025;
originally announced February 2025.
-
A global approach for the inverse scattering problem using a Carleman contraction map
Authors:
Phuong M. Nguyen,
Loc H. Nguyen,
Huong T. T. Vu
Abstract:
This paper addresses the inverse scattering problem in the domain Omega. The input data, measured outside Omega, involve the waves generated by the interaction of plane waves with various directions and unknown scatterers fully occluded inside Omega. The output of this problem is the spatially dielectric constant of these scatterers. Our approach to solving this problem consists of two primary sta…
▽ More
This paper addresses the inverse scattering problem in the domain Omega. The input data, measured outside Omega, involve the waves generated by the interaction of plane waves with various directions and unknown scatterers fully occluded inside Omega. The output of this problem is the spatially dielectric constant of these scatterers. Our approach to solving this problem consists of two primary stages. Initially, we eliminate the unknown dielectric constant from the governing equation, resulting in a system of partial differential equations. Subsequently, we develop the Carleman contraction mapping method to effectively tackle this system. It is noteworthy to highlight this method's robustness. It does not request a precise initial guess of the true solution, and its computational cost is not expensive. Some numerical examples are presented.
△ Less
Submitted 21 June, 2024; v1 submitted 5 April, 2024;
originally announced April 2024.
-
Determining initial conditions for nonlinear hyperbolic equations with time dimensional reduction and the Carleman contraction
Authors:
Trong D. Dang,
Loc H. Nguyen,
Huong T. T. Vu
Abstract:
This paper aims to determine the initial conditions for quasi-linear hyperbolic equations that include nonlocal elements. We suggest a method where we approximate the solution of the hyperbolic equation by truncating its Fourier series in the time domain with a polynomial-exponential basis. This truncation effectively removes the time variable, transforming the problem into a system of quasi-linea…
▽ More
This paper aims to determine the initial conditions for quasi-linear hyperbolic equations that include nonlocal elements. We suggest a method where we approximate the solution of the hyperbolic equation by truncating its Fourier series in the time domain with a polynomial-exponential basis. This truncation effectively removes the time variable, transforming the problem into a system of quasi-linear elliptic equations. We refer to this technique as the "time dimensional reduction method." To numerically solve this system comprehensively without the need for an accurate initial estimate, we used the newly developed Carleman contraction principle. We show the efficiency of our method through various numerical examples. The time dimensional reduction method stands out not only for its precise solutions but also for its remarkable speed in computation.
△ Less
Submitted 11 June, 2024; v1 submitted 2 December, 2023;
originally announced December 2023.
-
Large-time behavior of one-phase Stefan-type problems with anisotropic diffusion in periodic media
Authors:
Norbert Požár,
Giang Thi Thu Vu
Abstract:
We study the large-time behavior of solutions of a one-phase Stefan-type problem with anisotropic diffusion in periodic media on an exterior domain in a dimension $n \geq 3$. By a rescaling transformation that matches the expansion of the free boundary, we deduce the homogenization of the free boundary velocity together with the homogenization of the anisotropic operator. Moreover, we obtain the c…
▽ More
We study the large-time behavior of solutions of a one-phase Stefan-type problem with anisotropic diffusion in periodic media on an exterior domain in a dimension $n \geq 3$. By a rescaling transformation that matches the expansion of the free boundary, we deduce the homogenization of the free boundary velocity together with the homogenization of the anisotropic operator. Moreover, we obtain the convergence of the rescaled solution to the solution of the homogenized Hele-Shaw-type problem with a point source and the convergence of the rescaled free boundary to a self-similar profile with respect to the Hausdorff distance.
△ Less
Submitted 3 June, 2018;
originally announced June 2018.
-
Long-time behavior of the one-phase Stefan problem in periodic and random media
Authors:
Norbert Požár,
Giang Thi Thu Vu
Abstract:
We study the long-time behavior of solutions of the one-phase Stefan problem in inhomogeneous media in dimensions $n \geq 2$. Using the technique of rescaling which is consistent with the evolution of the free boundary, we are able to show the homogenization of the free boundary velocity as well as the locally uniform convergence of the rescaled solution to a self-similar solution of the homogeneo…
▽ More
We study the long-time behavior of solutions of the one-phase Stefan problem in inhomogeneous media in dimensions $n \geq 2$. Using the technique of rescaling which is consistent with the evolution of the free boundary, we are able to show the homogenization of the free boundary velocity as well as the locally uniform convergence of the rescaled solution to a self-similar solution of the homogeneous Hele-Shaw problem with a point source. Moreover, by viscosity solution methods, we also deduce that the rescaled free boundary uniformly approaches a sphere with respect to Hausdorff distance.
△ Less
Submitted 23 February, 2017;
originally announced February 2017.
-
Cyclic tridiagonal pairs, higher order Onsager algebras and orthogonal polynomials
Authors:
P. Baseilhac,
A. M. Gainutdinov,
T. T. Vu
Abstract:
The concept of cyclic tridiagonal pairs is introduced, and explicit examples are given. For a fairly general class of cyclic tridiagonal pairs with cyclicity N, we associate a pair of `divided polynomials'. The properties of this pair generalize the ones of tridiagonal pairs of Racah type. The algebra generated by the pair of divided polynomials is identified as a higher-order generalization of th…
▽ More
The concept of cyclic tridiagonal pairs is introduced, and explicit examples are given. For a fairly general class of cyclic tridiagonal pairs with cyclicity N, we associate a pair of `divided polynomials'. The properties of this pair generalize the ones of tridiagonal pairs of Racah type. The algebra generated by the pair of divided polynomials is identified as a higher-order generalization of the Onsager algebra. It can be viewed as a subalgebra of the q-Onsager algebra for a proper specialization at q the primitive 2Nth root of unity. Orthogonal polynomials beyond the Leonard duality are revisited in light of this framework. In particular, certain second-order Dunkl shift operators provide a realization of the divided polynomials at N=2 or q=i.
△ Less
Submitted 5 February, 2017; v1 submitted 3 July, 2016;
originally announced July 2016.
-
Interference Alignment Designs for Secure Multiuser MIMO Systems: Rank Constrained Rank Minimization Approach
Authors:
Tung T. Vu,
Ha Hoang Kha,
Trung Q. Duong
Abstract:
In this paper, we formulate the interference alignment (IA) problem for a multiuser multiple-input multiple-output (MIMO) system in the presence of an eavesdropper as a rank constrained rank minimization (RCRM) problem. The aim of the proposed rank minimization IA schemes is to find the precoding and receiver subspace matrices to align interference and wiretapped signals into the lowest dimension…
▽ More
In this paper, we formulate the interference alignment (IA) problem for a multiuser multiple-input multiple-output (MIMO) system in the presence of an eavesdropper as a rank constrained rank minimization (RCRM) problem. The aim of the proposed rank minimization IA schemes is to find the precoding and receiver subspace matrices to align interference and wiretapped signals into the lowest dimension subspaces while keeping the desired signal subspace spanning full available spatial dimensions. To deal with the nonconvexity of the rank function, we present two convex relaxations of the RCRM problem, namely nuclear norm (NN) and reweighted nuclear norm (RNN), and transform the rank constraints to equivalent and tractable ones. We then derive a coordinate decent approach to obtain the solutions for IA schemes. The simulation results show that our proposed IA designs outperform the conventional IA design in terms of average secrecy sum rate. On the other hand, our proposed designs perform the same or better than other secure IA schemes which account for low interference and wiretapped signal power rather than for low dimensions of interference and wiretapped signal matrices in the systems which achieve the perfect IA.
△ Less
Submitted 3 August, 2015;
originally announced August 2015.
-
Particle Swarm Optimization for Weighted Sum Rate Maximization in MIMO Broadcast Channels
Authors:
Tung T. Vu,
Ha Hoang Kha,
Trung Q. Duong,
Nguyen-Son Vo
Abstract:
In this paper, we investigate the downlink multiple-input-multipleoutput (MIMO) broadcast channels in which a base transceiver station (BTS) broadcasts multiple data streams to K MIMO mobile stations (MSs) simultaneously. In order to maximize the weighted sum-rate (WSR) of the system subject to the transmitted power constraint, the design problem is to find the pre-coding matrices at BTS and the d…
▽ More
In this paper, we investigate the downlink multiple-input-multipleoutput (MIMO) broadcast channels in which a base transceiver station (BTS) broadcasts multiple data streams to K MIMO mobile stations (MSs) simultaneously. In order to maximize the weighted sum-rate (WSR) of the system subject to the transmitted power constraint, the design problem is to find the pre-coding matrices at BTS and the decoding matrices at MSs. However, such a design problem is typically a nonlinear and nonconvex optimization and, thus, it is quite hard to obtain the analytical solutions. To tackle with the mathematical difficulties, we propose an efficient stochastic optimization algorithm to optimize the transceiver matrices. Specifically, we utilize the linear minimum mean square error (MMSE) Wiener filters at MSs. Then, we introduce the constrained particle swarm optimization (PSO) algorithm to jointly optimize the precoding and decoding matrices. Numerical experiments are exhibited to validate the effectiveness of the proposed algorithm in terms of convergence, computational complexity and total WSR.
△ Less
Submitted 3 August, 2015;
originally announced August 2015.
-
Wiretapped Signal Leakage Minimization for Secure Multiuser MIMO Systems via Interference Alignment
Authors:
Tung T. Vu,
Ha Hoang Kha,
Trung Q. Duong,
Nguyen-Son Vo
Abstract:
In this paper, interference alignment (IA) is designed for secure multiuser multiple-input multiple-output systems in the presence of an eavesdropper. The proposed IA technique designs the transmit precoding and receiving subspace matrices to minimize both the total inter-main-link interference and the wiretapped signals. With perfect channel state information is known at the transmitters (Txs), t…
▽ More
In this paper, interference alignment (IA) is designed for secure multiuser multiple-input multiple-output systems in the presence of an eavesdropper. The proposed IA technique designs the transmit precoding and receiving subspace matrices to minimize both the total inter-main-link interference and the wiretapped signals. With perfect channel state information is known at the transmitters (Txs), the cost function of the optimization problem is alternatively minimized over the precoding matrices at the Txs and the receiving subspace matrices at the receivers (Rxs) and the eavesdropper. The feasible condition, the proof of convergence of the proposed IA approach are provided. Numerical results reveal a significant improvement in terms of average secrecy sum rate of our IA algorithm as compared to the conventional IA design.
△ Less
Submitted 3 August, 2015;
originally announced August 2015.
-
On the Interference Alignment Designs for Secure Multiuser MIMO Systems
Authors:
Kha Hoang Ha,
Thanh Tung Vu,
Trung Quang Duong,
Nguyen-Son Vo
Abstract:
In this paper, we propose two secure multiuser multiple-input multiple-output transmission approaches based on interference alignment (IA) in the presence of an eavesdropper. To deal with the information leakage to the eavesdropper as well as the interference signals from undesired transmitters (Txs) at desired receivers (Rxs), our approaches aim to design the transmit precoding and receive subspa…
▽ More
In this paper, we propose two secure multiuser multiple-input multiple-output transmission approaches based on interference alignment (IA) in the presence of an eavesdropper. To deal with the information leakage to the eavesdropper as well as the interference signals from undesired transmitters (Txs) at desired receivers (Rxs), our approaches aim to design the transmit precoding and receive subspace matrices to minimize both the total inter-main-link interference and the wiretapped signals (WSs). The first proposed IA scheme focuses on aligning the WSs into proper subspaces while the second one imposes a new structure on the precoding matrices to force the WSs to zero. When the channel state information is perfectly known at all Txs, in each proposed IA scheme, the precoding matrices at Txs and the receive subspaces at Rxs or the eavesdropper are alternatively selected to minimize the cost function of an convex optimization problem for every iteration. We provide the feasible conditions and the proofs of convergence for both IA approaches. The simulation results indicate that our two IA approaches outperform the conventional IA algorithm in terms of average secrecy sum rate.
△ Less
Submitted 15 November, 2016; v1 submitted 3 August, 2015;
originally announced August 2015.
-
Higher order relations for ADE-type generalized q-Onsager algebras
Authors:
P. Baseilhac,
T. T. Vu
Abstract:
Let $\{A_j|j=0,1,...,rank(g)\}$ be the fundamental generators of the generalized $q-$Onsager algebra $\cal O_{q}(\widehat{g})$ introduced in \cite{BB1}, where $\widehat{g}$ is a simply-laced affine Lie algebra. New relations between certain monomials of the fundamental generators - indexed by the integer $r\in\mathbb{Z}^{+}$ - are conjectured. These relations can be seen as deformed analogues of L…
▽ More
Let $\{A_j|j=0,1,...,rank(g)\}$ be the fundamental generators of the generalized $q-$Onsager algebra $\cal O_{q}(\widehat{g})$ introduced in \cite{BB1}, where $\widehat{g}$ is a simply-laced affine Lie algebra. New relations between certain monomials of the fundamental generators - indexed by the integer $r\in\mathbb{Z}^{+}$ - are conjectured. These relations can be seen as deformed analogues of Lusztig's $r-$th higher order $q-$Serre relations associated with ${\cal U}_q({\widehat g})$, which are recovered as special cases. The relations are proven for $r\leq 5$. For $r$ generic, several supporting evidences are presented.
△ Less
Submitted 19 June, 2015; v1 submitted 20 December, 2013;
originally announced December 2013.
-
Analogues of Lusztig's higher order relations for the q-Onsager algebra
Authors:
P. Baseilhac,
T. T. Vu
Abstract:
Let $A,A^*$ be the generators of the $q-$Onsager algebra. Analogues of Lusztig's $r-th$ higher order relations are proposed. In a first part, based on the properties of tridiagonal pairs of $q-$Racah type which satisfy the defining relations of the $q-$Onsager algebra, higher order relations are derived for $r$ generic. The coefficients entering in the relations are determined from a two-variable…
▽ More
Let $A,A^*$ be the generators of the $q-$Onsager algebra. Analogues of Lusztig's $r-th$ higher order relations are proposed. In a first part, based on the properties of tridiagonal pairs of $q-$Racah type which satisfy the defining relations of the $q-$Onsager algebra, higher order relations are derived for $r$ generic. The coefficients entering in the relations are determined from a two-variable polynomial generating function. In a second part, it is conjectured that $A,A^*$ satisfy the higher order relations previously obtained. The conjecture is proven for $r=2,3$. For $r$ generic, using an inductive argument recursive formulae for the coefficients are derived. The conjecture is checked for several values of $r\geq 4$. Consequences for coideal subalgebras and integrable systems with boundaries at $q$ a root of unity are pointed out.
△ Less
Submitted 18 December, 2013; v1 submitted 12 December, 2013;
originally announced December 2013.
