0% found this document useful (0 votes)
4 views6 pages

Strassen's Algorithm

The document discusses an algorithm for matrix multiplication, specifically focusing on a recursive approach that partitions matrices to optimize the computation process. It outlines steps for implementing the algorithm, including creating submatrices and performing operations recursively. The complexity of the algorithm is analyzed, indicating improvements over traditional methods.

Uploaded by

07.sista
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
4 views6 pages

Strassen's Algorithm

The document discusses an algorithm for matrix multiplication, specifically focusing on a recursive approach that partitions matrices to optimize the computation process. It outlines steps for implementing the algorithm, including creating submatrices and performing operations recursively. The complexity of the algorithm is analyzed, indicating improvements over traditional methods.

Uploaded by

07.sista
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 6
Ree § ¢/ & nybetbe Matyi Pe Re » Frere won 4 Ajgortthe A» Ay B* bt one sqpoane watt ( nen) ce AB c= Ce Pex €,§°4,9,° n Cty . 2 Ate beg S quoare madnge mul Hipiiead ron algo rithm Sees 4 Ne A Bows 9. keé ¢ be a new (nxm ) madi 8 fe t24 ton - ae je n(ne) uu bey Je 4 ten © Cat ob ‘ Neg (naye” Ff Kd to ( b j> Den en # Cag = Cegt Ate® Ks : t relunn C Teny= On!) Sty « hivtde & Comqiren alyrcitin 1 € cape alivicle & Conquer BUyovit hen te Compute Mattox produces Cork OE —— »; MCIMme 1 he OI MN eC rmed Poew enc a 9 thm eacl, of qhe (nxn) nom diciy An each aliviole Step qe till Aivide vio foun (2 x 4 ) moticicey. ancl boy aatoming Mat OI. an excited power of 9 a-(e" Y 0 ALR Nyy No» fh : By ) ° ( a (4, Bos) C= Ay «By, + Ain Bo \ C= (3 os Cio= Au Bi + Ay Bar un € do " Coy Any Bn + Ars Boy Coo = Aa Bya + Ads Boo Sqpvaste Matex Meltiply Recursive (a, B) Sep = f\e rows naw mabrice 19 9 Le Che @ new MN Maitix 3. if need ue ch = Qy by r CeVgive & else peurtt ton A) + ggpnee RTL e( Ay» Bu ; (giv nel > ) (Ais »Po) Gs Cue sqmartrce eer y ative 4 sqmadrcrreecon’ , grayrcectere ve (All Bro) +54 “1. Cy = SY (Aya » 8 . ; recor) eh PX EE CELE vel Ay Bu) 4 sqmati x eC yy © SY (As) Bay q.¢ cqrnentiticwceco reve (A, Br) 4 apratitd rere 227 sy we 23)83,) Io, sedosn C et ay fe i ) hi ‘ > We can partition the matrices without copyin ent co > a 7 ee drick Ps to Use inolex Cotcutaty, J : 7 © fdentHy a Sib madre} | On rcix hy QQ Kron of Mow indices and a range of - orciginal rnceteedx ? column Fndices of the Let TCn) be the time +o multiply kwo (nxn ) matrices . when nz4 we pertfoarn just one scelare mulkiplfoation in step Yo tr Ac ahave | atqorithm. “Fen TD) gl) 34n Vine [Step 6 to4 in he above algoninn eevee recursively — eatl Square roagsrax multiply ecavse each fonction a. totat of BLmes: B | recerresive cane multiplies two (Plz x02) mae TON) to overcatt nonning by all Qi recursive call 1 oe 20 thereby contrtibutin Hine, tre Hime token % 8 TO%2) + The total Hine of the parcti tioning Deve, dhe reeceursive cathy enol dhe cime $0 alot from the Mecermaive & Dope dhe receeresive cate ta the som time for atl dh« tho Madyere Ey resulting tty. Se T(n)= 8(4) +8TO) + e(n2) @(4)i6 neglected So Ti) = @(4) yee ) BTM) LOC) OL y rm Stk 4 7 orrassen's Algo 4 a Stra Ay rithm petted: . SArassen's method ocd ie dn make lhe enn enreaion ATER prehik feos hereky . : a ‘I Herne tH perforin Bali 4 dy marti ply : : plication veeethye 61 innlead 3 pec ceresive ceercgive omutlipl tea tan. f mca), ned dT Cp \ 4 O(n’) , ny4 Then ane | Mepa in his, method! cred) cyivide The imped matrcices A&B & op" it mater CG! inte (> K "I, ) sub matt ces This step taxes O( 1) Kime by index catcviado® , re "71h)} = Srep-2 Create JO matUces- a ea 840 each of which (M2 x 2) and the sen aiffesene & uoe Cen of two meetecices onentect ws Stee , create att 4o matrices in o(?) Lime Step 3 Using tre Sobmatictces eneated tn step se ee compute in Stepd mec orawely Ap motaces eneatect Pa - ef patti preoctesc# ye Py Pp Ene motrox Pe % (%a* 9/2) . bmatrices Ci Cy Cai on Ache ing Be ahors a th q gobmatyee desinecl Se Compute toe ore of the result matt X by - sphanae tt O79 ya stoos coms t Pi matriees: " Ol ry) Roe bee Can compote an [| Z=KY . (§ F es) le Ha | Vette #Py> Ps Py tPs < ‘| P#Py bP feeP a) TRick 5, - ANED f Diagonal bast CR + Fig CR * Revo opertaston (+) ” Coferon Operation (-) core = cot Als © 7 i Py = A(F -H) eqperditf — Gr = mew value “Cr corrcesho i Po =H (A+B) (nc ee akin g A. eres PO PeeE(c+D) Et 4 weapon d “HI” pa cei ae it D' cox Pu = D(G-€) at Ps = Ard) *(€+H) - Pe - B-D) xe (Gt) & last tom os X K fagd «vee of Y p \A-C) ¥(E+F) (Firert col’ & Fired YG ew YY yo = + . Valo e' 4 €'8 colume ding i Ea a) GH / 7 jt pee UG eS oe? Cy t aneeemmmmmen pene p< ACF“) = 2(3-4) 2 (4 —o. | m= A(ASB)= 4 (244) 2 en “* | Px, - E(c+DD = 1(B+s)= 28 = i py = D(G-E)= CC yt) = Bx37 IS Ps * (PxdD) ¥ (extn) =(2+ 5) (ita x8 96 pg = (B-D) + ( Gs) = (u-s)x44) =E0 aye cll pa 7 BOX (evr) = (2-8) «C1 43) 2 -Y Ze Po ts t Pa ~Pe pprPs. PaxPy p,—Ps, +P ~ Pm) ; j-nesense —@ FUX . : +3 KIS ~3-% + S6-C) “e Sa [ ‘8 3y = lagg YY,

You might also like