Enhanced Types of Algorithms for Sparse-Matrix LU Factorization With Suppressed Fill-In and Improved Accuracy
The result's identifiers
Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F12%3A00204680" target="_blank" >RIV/68407700:21230/12:00204680 - isvavai.cz</a>
Result on the web
<a href="http://radio.feld.cvut.cz/personal/dobes2/dobes.en.pdf" target="_blank" >http://radio.feld.cvut.cz/personal/dobes2/dobes.en.pdf</a>
DOI - Digital Object Identifier
—
Alternative languages
Result language
angličtina
Original language name
Enhanced Types of Algorithms for Sparse-Matrix LU Factorization With Suppressed Fill-In and Improved Accuracy
Original language description
The computational efficiency of the sparse-matrix LU factorization represents the key point of the computer-aided design of electronic circuits. The LU factorization must be performed thousand times during the numerical solution of systems of nonlinear differential-algebraic equations, which further enhances importance of this problem. Moreover, the sparse matrix of the system is always irregular here - hence, the methods that are generally available for regular ones cannot be used. We created a novel modification of the Markowitz criterion, which is compatible with the so-called fast modes of the LU factorization. The modified criterion consists in an estimation of probabilities of the fill-in enlargement. The probabilities are determined for all columns of the system matrix before the LU factorization, where the column probability is calculated as the average value of the probabilities of its elements. Finally, the columns are reordered so that first and last ones should be those wit
Czech name
—
Czech description
—
Classification
Type
R - Software
CEP classification
JA - Electronics and optoelectronics
OECD FORD branch
—
Result continuities
Project
<a href="/en/project/GAP102%2F10%2F1665" target="_blank" >GAP102/10/1665: Symbolic and Semisymbolic Methods for Power and Mechatronic Applications</a><br>
Continuities
S - Specificky vyzkum na vysokych skolach
Others
Publication year
2012
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Data specific for result type
Internal product ID
SLUF Version 3 '12 (FTN95, LISP)
Technical parameters
Sparse-matrix LU factorization with suppressed fill-In and improved accuracy. Obecně použitelný algoritmus vhodný pro řešení jakýchkoliv velmi rozsáhlých soustav lineárních rovnic s nepravidelnou řídkou maticí soustavy. Licenci na zdrojový kód lze získat: ČVUT v Praze, K13137, Josef Dobeš, Technická 2, 16627 Praha 6, tel. 224352207, dobes@fel.cvut.cz
Economical parameters
Při řešení rozsáhlých úloh v elektrotechnice, ekonomice, astronomii apod. s využitím LU faktorizace úspora paměti, ale především strojního času v řádu desítek až stovek procent oproti dostupným algoritmům.
Owner IČO
68407700
Owner name
ČVUT