scholarly journals Extremely Scalable Algorithm for 108-atom Quantum Material Simulation on the Full System of the K Computer

2016 ◽  
Author(s):  
Takeo Hoshi ◽  
Hiroto Imachi ◽  
Kiyoshi Kumahata ◽  
Masaaki Terai ◽  
Kengo Miyamoto ◽  
...  
Keyword(s):  
Full System ◽  
Scalable Algorithm

scholarly journals SiCaSMA: An Alternative Stochastic Description via Concatenation of Markov Processes for a Class of Catalytic Systems

Mathematics ◽  
2021 ◽  
Vol 9 (10) ◽  
pp. 1074
Author(s):  
Vincent Wagner ◽  
Nicole Erika Radde
Keyword(s):  
Reaction System ◽  
Stochastic Simulation Algorithm ◽  
Test Bed ◽  
Biochemical Reaction Networks ◽  
Full System ◽  
Catalytic Systems ◽  
Reaction Systems ◽  
Sample Paths ◽  
Alternative Description ◽  
Linear Differential

The Chemical Master Equation is a standard approach to model biochemical reaction networks. It consists of a system of linear differential equations, in which each state corresponds to a possible configuration of the reaction system, and the solution describes a time-dependent probability distribution over all configurations. The Stochastic Simulation Algorithm (SSA) is a method to simulate sample paths from this stochastic process. Both approaches are only applicable for small systems, characterized by few reactions and small numbers of molecules. For larger systems, the CME is computationally intractable due to a large number of possible configurations, and the SSA suffers from large reaction propensities. In our study, we focus on catalytic reaction systems, in which substrates are converted by catalytic molecules. We present an alternative description of these systems, called SiCaSMA, in which the full system is subdivided into smaller subsystems with one catalyst molecule each. These single catalyst subsystems can be analyzed individually, and their solutions are concatenated to give the solution of the full system. We show the validity of our approach by applying it to two test-bed reaction systems, a reversible switch of a molecule and methyltransferase-mediated DNA methylation.

scholarly journals FastSK: fast sequence analysis with gapped string kernels

2020 ◽  
Vol 36 (Supplement_2) ◽  
pp. i857-i865
Author(s):  
Derrick Blakely ◽  
Eamon Collins ◽  
Ritambhara Singh ◽  
Andrew Norton ◽  
Jack Lanchantin ◽  
...  
Keyword(s):  
Sequence Analysis ◽  
Dna Sequences ◽  
English Language ◽  
Computation Time ◽  
Entity Recognition ◽  
Supplementary Information ◽  
Support Vector ◽  
Homology Detection ◽  
Scalable Algorithm ◽  
String Kernels

Abstract Motivation Gapped k-mer kernels with support vector machines (gkm-SVMs) have achieved strong predictive performance on regulatory DNA sequences on modestly sized training sets. However, existing gkm-SVM algorithms suffer from slow kernel computation time, as they depend exponentially on the sub-sequence feature length, number of mismatch positions, and the task’s alphabet size. Results In this work, we introduce a fast and scalable algorithm for calculating gapped k-mer string kernels. Our method, named FastSK, uses a simplified kernel formulation that decomposes the kernel calculation into a set of independent counting operations over the possible mismatch positions. This simplified decomposition allows us to devise a fast Monte Carlo approximation that rapidly converges. FastSK can scale to much greater feature lengths, allows us to consider more mismatches, and is performant on a variety of sequence analysis tasks. On multiple DNA transcription factor binding site prediction datasets, FastSK consistently matches or outperforms the state-of-the-art gkmSVM-2.0 algorithms in area under the ROC curve, while achieving average speedups in kernel computation of ∼100× and speedups of ∼800× for large feature lengths. We further show that FastSK outperforms character-level recurrent and convolutional neural networks while achieving low variance. We then extend FastSK to 7 English-language medical named entity recognition datasets and 10 protein remote homology detection datasets. FastSK consistently matches or outperforms these baselines. Availability and implementation Our algorithm is available as a Python package and as C++ source code at https://github.com/QData/FastSK Supplementary information Supplementary data are available at Bioinformatics online.

scholarly journals Scalable algorithm for solving 3D contact problems with friction

PAMM ◽  
2007 ◽  
Vol 7 (1) ◽  
pp. 1025201-1025202
Author(s):  
Radek KucÌŒera ◽  
Jaroslav Haslinger ◽  
Zdeněk Dostál
Keyword(s):  
Contact Problems ◽  
Scalable Algorithm

A Full-System Approach of the Elastohydrodynamic Line/Point Contact Problem

2008 ◽  
Vol 130 (2) ◽  
Author(s):  
W. Habchi ◽  
D. Eyheramendy ◽  
P. Vergne ◽  
G. Morales-Espejel
Keyword(s):  
Finite Element ◽  
Nonlinear System ◽  
System Approach ◽  
Jacobian Matrix ◽  
Point Contact ◽  
System Of Equations ◽  
Nonlinear System Of Equations ◽  
Full System ◽  
Element Discretization ◽  
Fully Coupled

The solution of the elastohydrodynamic lubrication (EHL) problem involves the simultaneous resolution of the hydrodynamic (Reynolds equation) and elastic problems (elastic deformation of the contacting surfaces). Up to now, most of the numerical works dealing with the modeling of the isothermal EHL problem were based on a weak coupling resolution of the Reynolds and elasticity equations (semi-system approach). The latter were solved separately using iterative schemes and a finite difference discretization. Very few authors attempted to solve the problem in a fully coupled way, thus solving both equations simultaneously (full-system approach). These attempts suffered from a major drawback which is the almost full Jacobian matrix of the nonlinear system of equations. This work presents a new approach for solving the fully coupled isothermal elastohydrodynamic problem using a finite element discretization of the corresponding equations. The use of the finite element method allows the use of variable unstructured meshing and different types of elements within the same model which leads to a reduced size of the problem. The nonlinear system of equations is solved using a Newton procedure which provides faster convergence rates. Suitable stabilization techniques are used to extend the solution to the case of highly loaded contacts. The complexity is the same as for classical algorithms, but an improved convergence rate, a reduced size of the problem and a sparse Jacobian matrix are obtained. Thus, the computational effort, time and memory usage are considerably reduced.

scholarly journals A scalable algorithm for the optimization of neural network architectures

2021 ◽  
pp. 102788
Author(s):  
Massimiliano Lupo Pasini ◽  
Junqi Yin ◽  
Ying Wai Li ◽  
Markus Eisenbach
Keyword(s):  
Neural Network ◽  
Network Architectures ◽  
Scalable Algorithm ◽  
Neural Network Architectures

Sliding Mode Analysis of a Counterbalance Valve Induced Instability in an Electrohydraulic Drive

2021 ◽  
Author(s):  
Philipp Zagar ◽  
Rudolf Scheidl
Keyword(s):  
System Dynamics ◽  
Sliding Mode ◽  
Hydraulic Drive ◽  
Mode Analysis ◽  
Order System ◽  
Full System ◽  
Reduced Order ◽  
Valve Dynamics ◽  
Electrohydraulic Drive ◽  
Saturation Effects

Abstract This paper analyzes dynamic effects of an electro-hydraulic drive which uses a counter-balance valve for rod volume compensation. It shows that local stability analysis is not sufficient in this particular case to get general statements of the system's chattering properties. A reduced-order switched system is proposed to gain deeper insights in system dynamics with saturation effects such as the end-stop of a valve poppet and solutions are compared numerically to the full-system dynamics which incorporates pressure built-up, piston and valve dynamics as well as motor dynamics. It is shown that in cases of e.g. fast valves with small cracking pressures undesirable chattering of the full system exists which can be easily understood in terms of the reduced-order system in form of sliding mode solutions. The paper also describes under which conditions such sliding modes exist, how they behave and how they can be interpreted in terms of the full system.

scholarly journals Fully Parameterized Reduced Order Models Using Hyper-Dual Numbers and Component Mode Synthesis

2015 ◽  
Author(s):  
Matthew S. Bonney ◽  
Daniel C. Kammer ◽  
Matthew R. W. Brake
Keyword(s):  
Truncation Error ◽  
Sampling Methods ◽  
Latin Hypercube Sampling ◽  
Component Mode Synthesis ◽  
Reduced Order Models ◽  
Dual Numbers ◽  
Order Model ◽  
Full System ◽  
Reduced Order ◽  
Computational Burden

The uncertainty of a system is usually quantified with the use of sampling methods such as Monte-Carlo or Latin hypercube sampling. These sampling methods require many computations of the model and may include re-meshing. The re-solving and re-meshing of the model is a very large computational burden. One way to greatly reduce this computational burden is to use a parameterized reduced order model. This is a model that contains the sensitivities of the desired results with respect to changing parameters such as Young’s modulus. The typical method of computing these sensitivities is the use of finite difference technique that gives an approximation that is subject to truncation error and subtractive cancellation due to the precision of the computer. One way of eliminating this error is to use hyperdual numbers, which are able to generate exact sensitivities that are not subject to the precision of the computer. This paper uses the concept of hyper-dual numbers to parameterize a system that is composed of two substructures in the form of Craig-Bampton substructure representations, and combine them using component mode synthesis. The synthesis transformations using other techniques require the use of a nominal transformation while this approach allows for exact transformations when a perturbation is applied. This paper presents this technique for a planar motion frame and compares the use and accuracy of the approach against the true full system. This work lays the groundwork for performing component mode synthesis using hyper-dual numbers.

A Scalable Algorithm for Constructing Frequent Pattern Tree

2014 ◽  
Vol 10 (1) ◽  
pp. 42-56 ◽  
Author(s):  
Zailani Abdullah ◽  
Tutut Herawan ◽  
A. Noraziah ◽  
Mustafa Mat Deris
Keyword(s):  
Data Structure ◽  
Frequent Pattern ◽  
Frequent Patterns ◽  
Scalable Algorithm ◽  
Tree Construction ◽  
Frequent Pattern Tree ◽  
Support Threshold ◽  
Benchmark Datasets ◽  
Tree Data ◽  
Tree Data Structure

Frequent Pattern Tree (FP-Tree) is a compact data structure of representing frequent itemsets. The construction of FP-Tree is very important prior to frequent patterns mining. However, there have been too limited efforts specifically focused on constructing FP-Tree data structure beyond from its original database. In typical FP-Tree construction, besides the prior knowledge on support threshold, it also requires two database scans; first to build and sort the frequent patterns and second to build its prefix paths. Thus, twice database scanning is a key and major limitation in completing the construction of FP-Tree. Therefore, this paper suggests scalable Trie Transformation Technique Algorithm (T3A) to convert our predefined tree data structure, Disorder Support Trie Itemset (DOSTrieIT) into FP-Tree. Experiment results through two UCI benchmark datasets show that the proposed T3A generates FP-Tree up to 3 magnitudes faster than that the benchmarked FP-Growth.

Sign in / Sign up

Export Citation Format

Share Document