0% found this document useful (0 votes)
6 views17 pages

Optimal Matrix Multiplication Techniques

The document discusses optimal matrix multiplication strategies, focusing on minimizing computational costs through various algorithms and methods. It emphasizes the importance of structuring subproblems effectively to achieve optimal solutions. Additionally, it outlines specific steps and considerations for implementing these strategies in practical scenarios.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
6 views17 pages

Optimal Matrix Multiplication Techniques

The document discusses optimal matrix multiplication strategies, focusing on minimizing computational costs through various algorithms and methods. It emphasizes the importance of structuring subproblems effectively to achieve optimal solutions. Additionally, it outlines specific steps and considerations for implementing these strategies in practical scenarios.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Dursmic bofsmmig

dngmG phoPamwnng A Ulce a


divide and anguenmatheds b u t e
i a dilfence in DAC mctads
Subphselems eere an dndopondet dao
dmgmiC psAaSmmlg Subablams
an
atgusp om 3hane Subpelekem)

mDAC nethed e 2abe opeardlt


Comims ubpnallaim Colhla nd4nemic
peamalng alocithm 2oles Le
S nee ery Subphalglem t once
and A SAres asLEA
h
a dalke,
9as OS RALCo ame

Dnsmic e7nm19goenaly applied


d ophmi2atpn p2lloms 3uched
phmln e Can lecmgny p obille
Ond we aLalh do fird a aionu
Ot Dpima mgalmum a

minimum)
Page No
-krisHNA
Date

Dsmic-agsmming alocHmm Csn le


ekon ido 3equone of-fom alep

D Chehceize sbuctuma of Cm
Optimal Seludion
2) Reauasively detine- h yalze ofan
OpimsL suludion
- e vdue of an opima sdution
3 Comphe
ma butomp faadaion
) Con3tOuct cn opimaLsoudian-fSn
e Camputed iMfhmedion

SlepSlep eoniille if on t e velus


Can
a n opmmel ssludion
of an Sasigud
hcdbixCGaine Mulipliachion
Hea u e ana given
seguae hain)
a

CA An) of nmetoi(e do be
muliplieds and e Loi do ampde
h phoduct Aa An

Audu of mstaices i Sy pabothaiud


t i edtha a single madai
oduc of wsf panuthesjned
nedaik oducb S ndad
Denetheser
Medi easin muldipicedion s
and. alll panmcthealzedt.n
akoclcie
yieldH

Mldhiplss huo modaices


A pq wodaix(gien)
= qn moàzia Gieh)
CP (J be
Compued)
Alont
MATRINMULTIPLY (A, B)
Hham e o , nCmpdhadle dimowion
doth Aeth al (
ds cli]<o
kelth CalumnAJ
1 do C[AJ<c[ij+A(KJ4

Te ineedo Campule Cis demingled. y


thnumben oSalen rauldiplicotion in
Jin , ohidh w p9

To gehe didfa Cot incanod


[Link] a
ad phoduc odiden He publem
Ofa chsun

DEmplo n-2 m m
M(M ma) (M1-ML).ma
I o Io loo = loox 5
M SS6

M-(mm)
m-m)asutn looS5o= 25o0o
Slan muldiph
yauhs My jee5o
AbLaM MyaguLn olad450
=Sooo0
mulKplidinn
mlGipliedin= 25ooo+50 00015,003
OR
Ma).m
(Mr duguusus 1oxloo 5= 5000
&oa mudilich

Ms-m3 ues 1o5So E2500


SGl mustiplidti
Tale nJdipliction, Soo+25s0 7S00
=

Copuiva pheduce atcodsrg o


ord pandi2af sn lo tis fodor Han
Padhenadion of modaices
houg be
ofa Sooned inany tau
csn
Coldh M1, MMy atth eeb.
m M

M1-(M Ma)

mg my m mg

M13 my my
n a Defirsdin otte maolam
mem neblam an he seleo as

Caiven a chatn ({A,, A)


O matice cohOLL fn =1,
2- mataip Ai hab dumonabn
Pa- p f panDthe212e t h
phducs h A n in auay fhat
nmmizes}e numbe ah
mhplicciond
mednp chain mthpliel on phablan
e ara n
actua muldiply in9
motnice Ou go only t
dakemina an hdtn fn mulilyin
matmie thott has Joaet os

C r9 numbe
or Loays o
POnRshslzati6n
Pcn) dansles ha Yumbe 0E
sa atengdiv panetha-picinr of
a squanie of n mgdi ces.
.
Date
NPSHNA
Cohen m=1,

Chn
A-fIl pnuthecbizea
paduct y paduct fwo incdoid
f y ponuxthoizeo nodaio 3ubphaduds
CAnd Spld betoeantte-os
ubpoducb
and
mgy occuJ
Jeaun fa kh
(F)Et msdniices fsn a
S--Y4. s e
K=1,2,
dletin e shecuaLNe
Pol
PCKP(n-k) if n7

leplESbuchne oam optal


panestheh 2cdion,

tCneniene a t odnpts a

hadcdi sn
s1, faheanstanis
Ai- LohenR
Ola- m evaludiia-h

anductAL, Ja A
-krisHNA
Page No.
Date

thep°alonn nodaivla ie Is
than any praxteaizatiom orf
Rhoduc AL,[Link] aplat ta
hoductr betuen fAK and A+I Sa
Sone inlegen k inte ange isk
os i fo Aome valuo of k, ue finut
CesmpubeAhe mgtdices A..k and AtkH
and Henmuliplythan thgethan do
prnduce t h finsd phoductr As

Ces g s panoekizadion H e cast


o

Cmpdng lemadhix Ask plubs


He Cas fGenpuhip9
Aki PLs
Hhe Cot of mipl tham dngade

palamu.
Suppoeht an einal penudhei2th
cf Ai,A An 3pldth paeduc
cuen Ar and HyHl,
Tey-h panadthezizodisn of-He
a i ubchairm AiAim k oHhin
AA must be an 0etima
Ai
pmudh20àon ef AiAt k
[Link]
Page No. risHNAn
Date

csenuction halda fd He
A3imsh
otho2shon erke 3ubchadn At Ata2A
inthe o@timalu panatkarzedion of Al,Ast
loeiam oetimak pooudteizhon
A mut

Nots ue e suh oztimal 3uataucrun


shau t h t e Cam Ceshuct a n opdimal

Saltion o t e alem um tima


galuions to Subyphablams

dein he cest of C a n
H e
elution aauaively in dems
optimal
h optimal udion s to ubpallavr
6
mininum numbe o
Leb mX,jI loe-He
nuedaal tdo Campe
3(alm multiplucation
matnisx ALi fhom ALAL l
Sesuhe.
t h Cesti of-fhe
ehthefAL prusblon, CompuceAns)
coouds
chrapess Ci
hus be ml,n
Pege No. krisHNA
Date
ocunie
fincdio 0f m1 as allau

ifL-i hlom i tivial chsin


Cesibts on on motix Asi= Ar

m[ij=0 fon N=\,2,--n


Lshen
Hetwe use stbuchunoste othnma
Solubion mAep1
Los us as um thads opima panatheirdist-
aplits i (oducs Ai Atn---j afa
AK and. k Ask
S mi11 eualldoHeinimum
Go t
Compusting he subreducdss
Ai.K and
Aktl1 pus casyt o
muzhplpinA th ws mathices
tagadhe
[Link] eimtrlin of esdh mstanic
A Pi-Alt o Compucins
madoi hadu e
trAAth dakes
hus ue, otatn

Te ae Orlr -l pokible. Valua off K

Sne e ophima panutaj2disn mub


had on oheck em all, do find-He
hest
OL nacuaaie soldHon alLe

m ak
min mClJHm[K+LJ+Ph

o eoptaadk o-heud to Astuct an


optimal aution, I t s dofire
ohch
SO=k doaley at ue

Split He prada ct AA-A


e p oltain an optimal pamutheianhi
sp1 Cmpctina - e opimal, cast
bnasted of Compthyqe oldhton t
ecuonce O Jadoyive e
perfaim
he thisd shee ofte y9m.C
hspemig pstadiam and ompte 4e
opfima a t aing tabrls),leatam
up appnach.

fAatsna R3eudneda aSUmeL ett nsnijy


AL ha Amoaisn PtXPe fo l=,2.n
e input a
esuonte p<fo,Pl,--sPn
Lohonlcrgth CPJ=
cedma uSe am auallicoy ailo
ml.n,l. -nf Suhra hm
Cota
An auaiJlany Jobla 31.-n1..n) t
e lendg cohtch indos o K_achieed
he Dptinmol ca in Ganmpuding mL
Gse tsl use dala Sdo Coslbuct
Cin ophimal mldtion
MATKI eHAIN-ORDER (P)
ne Jorgth [e1-1
ds mi,i]<o
C h hasn loadh)
S
da i -
m(j]<oo

o f q<mtJ
thn mA 2

1 nuhum mamd

Saliet i a n s spoidtm
nine.0-3 algeithm [Link]
mi 0i J=\,2,m
n lt Compultb n , IfA=1,2
hain of
n-(mimimum Co f
ngth J=2) dning flny
execuis of f J a p in lnt .
e ecard tima neug he Joop
Compules mi, fi=y2,3
h-2h minimum Caxts heh aina af
ogth =3) and fodh

At esh 3lep he. mj cot oinpuded,


n i e q-2 deponds on on
dalle exdoies m,1andm(kjJ
alnaady Coimpudede

oIS7So 7,875 37 875e


2 .262407S7125o,So
7So2So S375
3Soo
S 0to

2 3

S
S
Page No. -krisHNA
Date

dtimodbn

AL 30 35
35 ISP
I5S
5Klo f
As 10X20
A 20X25 f

n224 .5+hAP=13 000


inas= min mf25]+m4J+A6f=125
Yn(2+mtsj+htaPs= 11375

=125

We Coksthuctan optimal goludtian


Srpy m h e Camputkd inrfmsdim

ghbud in h Jalle l n , 1..n.

ueds t value o
Eadh evday 8 1 1 panethobadtian
Kdht te optmal
ALAIAslts the paaduct
be Con Ak qnd Al
PageNo. -krisHNA
Date

dinal mathi maAkiplicdisn An


Computsng An optiralt 13 ELOn]
A-S1 Ast+1n

Cn oima panoy-thoszodion of<AL,AL


piDub alynrtim ana e inda cey- tand
Te Inidialu coll eie
PRiNT-0PT1OAL-PARENS(8.h
gs
RINT-6PTImAL-PARENS s.i)
then pirt *Ai
ese i t
u
RINT-OPTUMAL-PARENS(8,
RINT-OP i,3 (L)
Pit
TMmAL-PARENS (8, SijHI,

You might also like