
eLABa objektas: "Lygčių ir nelygybių simbolinio sprendimo lygiagretusis metodas", 2011,D:20110831:14040469461 

E. dokumentai 

URL nuoroda 
http://vddb.laba.lt/obj/LTeLABa0001:E.02~2011~D_20110831_14040469461 
Dokumentas 
Magistro darbas

Prieigos teisės 
Laisvai prieinamas internete.

Institucija 
Kauno technologijos universitetas

Mokslo kryptis 
07 T  Informatikos inžinerija

Atsakomybė 
Neverauskas, Aurimas  Magistro baigiamojo darbo autorius Marcinkevičius, Romas  Magistro baigiamojo darbo vadovas Ostreika, Armantas  Magistro baigiamojo darbo recenzentas Blažauskas, Tomas  Magistro baigiamojo darbo vertinimo komisijos narys Kauno technologijos universitetas  Mokslinį laipsnį teikianti institucija

Antraštė (ės) 
Lygčių ir nelygybių simbolinio sprendimo lygiagretusis metodas Parallel method for symbolic solution of equations and inequalities

Santrauka [EN] 
I have presented an effective way to solve symbolic systems of equations and inequalities using parallel processes and compared it to ineffective method. Also, I have performed analysis of presented algorithm, determining its performance dependencies and comparing its performance to existing software.

Also, this paper discusses architectural solutions for the application system: MVC design pattern, "Onion" architecture and Dependency Injection. These architectural patterns benefit more than standard layered architecture, software, based on these patterns, is more maintainable and changeable.

These days, computers usually have multicore processors, but not all software use them efficiently. The main problem is to create algorithm for solving symbolic systems of equations and inequalities using parallel processes, using calculation power and decreasing calculation time. Such application system was created and analyzed in this paper.

It was determined that created software is superior to Maple CAS when task is small by input but requires a lot of calculating power (systems of inequalities). On the other hand, results differ when task consist of plenty of equations (4050 equations in system, same number of unknowns). Created software falls behind Maple CAS in performance. The main reason, for this, is that created software spends too much time to analyze task and strings in it.


Raktažodžiai: parallel, algorithm, equation, inequality, algebra 
