Search results for "Matrix multiplication"
showing 10 items of 20 documents
Experimental Study of Six Different Implementations of Parallel Matrix Multiplication on Heterogeneous Computational Clusters of Multicore Processors
2010
Two strategies of distribution of computations can be used to implement parallel solvers for dense linear algebra problems for Heterogeneous Computational Clusters of Multicore Processors (HCoMs). These strategies are called Heterogeneous Process Distribution Strategy (HPS) and Heterogeneous Data Distribution Strategy (HDS). They are not novel and have been researched thoroughly. However, the advent of multicores necessitates enhancements to them. In this paper, we present these enhancements. Our study is based on experiments using six applications to perform Parallel Matrix-matrix Multiplication (PMM) on an HCoM employing the two distribution strategies.
Multi-label classification using boolean matrix decomposition
2012
This paper introduces a new multi-label classifier based on Boolean matrix decomposition. Boolean matrix decomposition is used to extract, from the full label matrix, latent labels representing useful Boolean combinations of the original labels. Base level models predict latent labels, which are subsequently transformed into the actual labels by Boolean matrix multiplication with the second matrix from the decomposition. The new method is tested on six publicly available datasets with varying numbers of labels. The experimental evaluation shows that the new method works particularly well on datasets with a large number of labels and strong dependencies among them.
Response determination of linear dynamical systems with singular matrices: A polynomial matrix theory approach
2017
Abstract An approach is developed based on polynomial matrix theory for formulating the equations of motion and for determining the response of multi-degree-of-freedom (MDOF) linear dynamical systems with singular matrices and subject to linear constraints. This system modeling may appear for reasons such as utilizing redundant DOFs, and can be advantageous from a computational cost perspective, especially for complex (multi-body) systems. The herein developed approach can be construed as an alternative to the recently proposed methodology by Udwadia and coworkers, and has the significant advantage that it circumvents the use of pseudoinverses in determining the system response. In fact, ba…
Drude weight increase by orbital and repulsive interactions in fermionic ladders
2019
In strictly one-dimensional systems, repulsive interactions tend to reduce particle mobility on a lattice. Therefore, the Drude weight, controlling the divergence at zero-frequency of optical conductivities in perfect conductors, is lower than in non-interacting cases. We show that this is not the case when extending to quasi one-dimensional ladder systems. Relying on bosonization, perturbative and matrix product states (MPS) calculations, we show that nearest-neighbor interactions and magnetic fluxes provide a bias between back- and forward-scattering processes, leading to linear corrections to the Drude weight in the interaction strength. As a consequence, Drude weights counter-intuitivel…
Renormalization group flows for Wilson-Hubbard matter and the topological Hamiltonian
2019
Understanding the robustness of topological phases of matter in the presence of interactions poses a difficult challenge in modern condensed matter, showing interesting connections to high energy physics. In this work, we leverage these connections to present a complete analysis of the continuum long-wavelength description of a generic class of correlated topological insulators: Wilson-Hubbard topological matter. We show that a Wilsonian renormalization group (RG) approach, combined with the so-called topological Hamiltonian, provide a quantitative route to understand interaction-induced topological phase transitions that occur in Wilson-Hubbard matter. We benchmark two-loop RG predictions …
Entanglement in Gaussian matrix-product states
2006
Gaussian matrix product states are obtained as the outputs of projection operations from an ancillary space of M infinitely entangled bonds connecting neighboring sites, applied at each of N sites of an harmonic chain. Replacing the projections by associated Gaussian states, the 'building blocks', we show that the entanglement range in translationally-invariant Gaussian matrix product states depends on how entangled the building blocks are. In particular, infinite entanglement in the building blocks produces fully symmetric Gaussian states with maximum entanglement range. From their peculiar properties of entanglement sharing, a basic difference with spin chains is revealed: Gaussian matrix…
Entanglement continuous unitary transformations
2016
Continuous unitary transformations are a powerful tool to extract valuable information out of quantum many-body Hamiltonians, in which the so-called flow equation transforms the Hamiltonian to a diagonal or block-diagonal form in second quantization. Yet, one of their main challenges is how to approximate the infinitely-many coupled differential equations that are produced throughout this flow. Here we show that tensor networks offer a natural and non-perturbative truncation scheme in terms of entanglement. The corresponding scheme is called "entanglement-CUT" or eCUT. It can be used to extract the low-energy physics of quantum many-body Hamiltonians, including quasiparticle energy gaps. We…
Simulation of matrix product states for dissipation and thermalization dynamics of open quantum systems
2020
Abstract We transform the system/reservoir coupling model into a one-dimensional semi-infinite discrete chain through unitary transformation to simulate the open quantum system numerically with the help of time evolving block decimation (TEBD) algorithm. We apply the method to study the dynamics of dissipative systems. We also generate the thermal state of a multimode bath using minimally entangled typical thermal state (METTS) algorithm, and investigate the impact of the thermal bath on an empty system. For both cases, we give an extensive analysis of the impact of the modeling and simulation parameters, and compare the numerics with the analytics.
Higher order matrix differential equations with singular coefficient matrices
2015
In this article, the class of higher order linear matrix differential equations with constant coefficient matrices and stochastic process terms is studied. The coefficient of the highest order is considered to be singular; thus, rendering the response determination of such systems in a straightforward manner a difficult task. In this regard, the notion of the generalized inverse of a singular matrix is used for determining response statistics. Further, an application relevant to engineering dynamics problems is included.
Empirical Autotuning of Two-level Parallel Linear Algebra Routines on Large cc-NUMA Systems
2012
In large cc-NUMA systems the efficient use of the different levels of the memory hierarchy is not an easy task, and the performance of multithreading implementations of the libraries decreases when the number of cores used increases, so producing an important lost of efficiency. To alleviate this problem, routines with multilevel parallelism can be developed by combining OpenMP and BLAS parallelism. In that way, higher performance can be achieved, but it is necessary to develop some autotuning technique for the appropriate selection of the number of threads to use at each level. The selection can be made through theoretical models of the execution time or some installation methodology. This…