Solving the Linear Algebraic Equation Systems with Double Precision of Computations in C Language
| Authors: Manichev V.B., Glazkova V.V., Kozhevnikov D.Yu., Kir, Sakharov M.K. | Published: 01.09.2013 |
| Published in issue: #4(85)/2011 | |
| DOI: | |
| Category: Informatics & Computing Technology | |
| Keywords: simulation, ordinary differential equations, differential-algebraic equations, integration methods, time domain, linear algebraic equations | |
A problem of obtaining a numerical solution to the linear algebraic equation systems with a precision of 15 correct significant figures (double precision in C language) for all elements of the solution vector is considered. It is shown that the known iteration methods for refining the vector of solution to the linear algebraic equation systems make it possible to estimate the precision of the obtained solution for a single iteration. To achieve the required double precision for values of the solution vector elements, it is necessary to use computations with the increased number of digits in mantissa of real numbers. Results of solving of test problems are given.