In this pWer,we propose a new lteratlve correcting type ofunstructured algeï¼ braicçš¿ltipid(AMG)mthod which has。re extensl陀 annlicatlon nrosnects and better stability on the background of*theound existent AMG method for eL liptic PDE.Applying the new AMG method to solve the Inverse ofsuåœbio & matrix weget the corresponding bio出 Gaussï¼Seldel iteration based on Vï¼Cycle mthodmd preconditioned coqugate gradient(PCG)mthod for the discrete mdels on Lattice materials.Using the operator mtrlx)矾tending tecncnolo欧 wegmerali沈 the AMG metbd to the case ofequatlon groups and欧t a tWe 。f AMG method called AMV nd APCG method aboutthe discrete modes of lattice materials.A great unç¿er ofexperlments show thæ—¶ when the imperï¼ tant paramter a E(0.l,lï¼½the iteration number ofcorresponding AMV and APCG method is on the whole Indå°endent of*the scale ofquestlons and the parameter a for thetatlce materlajs discrete modets with qï¼l,3 azld 4,that when a Is very small the iteration unå’–er ofAPCG method Is independent of*the scale ofquestlons anddauges little s a becomes smaller In the case ofq二1.3.TherefOrethehlghefficlenCyandrobustness oftheALIG algorithm havebeen testd.Its superlorltyls shown In thelargesscale scientific compuï¼ tatlons about lattice materials二Moreoverby m冰ingfurther analysis forthe approximately continuous models associated wlththe discrete models on fatï¼ tlce materials with qï¼lwe haveproved the result that the condition number of PCG method,whose preconditioner B Is chosen as the Inverse of the block dlagonalmatrlx,Is Independent oftheparameter a.Accordingl。,we verify the correctness of*thetness corresponding numerical results In theory....
|