Matrix Inverse
You need to remember that the
Inverse may not exist:
This matrix does not have an
a-(} r
00
348 imase
‘an nxn (square) matric ke
invari tan oni if
there is another nxn
attic say B, such that
AE Man
‘Such matrix Bis
BeEMnn
ABS BA=In*
caledthelmeneota = AT =B
bn by
Proc asumertat Bethe ( nt *) a
ba, bap
pit)
Now, we test: az~(j ay (
‘condo
by )
baFinding Matrix Inverse
Minter Rectan igs ons Stans)
Example!
Bere)
[Exuraneiminsion 2250 10
‘smal = [oaof2 oa
oo 2|-t 24
Uae” Sommancvofnaesee/ 1 0 of. 2
ever, FF = [o1ofi2 0 0
Incase of 2 matics there ica direct formulaand a
cleect tes forinvertibity
fen ay) MA¥KA invert and oni t
AS ( " )
ess a Met A= aut ~ oun £0
Insuenease At
wea
‘There a semi-ivect analogue of this formula fr larger matrices
‘We wit cuss when we wil study determinant tern this
Example 2
1a 12zsfioo
a-(2a4 234}o10
345, sasfoon
Wi
eeMatrix Transpose
A=(ax) and AT = (bj)
AEMmn + AT= Mam
Acs,
bie = Kj, F= 1...) K=1,...,m
ou 110
A=|1 2] > AT=
5 92-1Properties of Matrix Operations
“Transpose and Addition Transpose ane Mullion by Scalar
ma :
(A4B) = ATL BT, ADEM ay Cs
Transpose and Matrix Multiplication Inyserse and Multiplication by Scalar
(AB) = BP-AT, Ae Mop, BEMm (44) T= 2144, ACM ACC
Matric Inverse and Matrix Multiplication
‘Transpose and Matrix Invserse
(ey (8)", Aen (AB)-"= B44, AM, BeMan perensprere at,Special Matrices
AE Mn
Symmetric Matrix Skew-symmetric matrixExample
oo 123 023 ‘00 8
Find old a 7 oo4 es
oot oo1 ooo 000
ARN AM = (ty +N) = S Chalks N" = Chats + ChyaN + Cha? = Ty +2022 x N+ 1011 x 2021 x N?
«eo 2021, Reference to Binomial Theorem in
Binomial Formula (a+8)"= Scher tot be Chg ae eae cea
‘Theorem: Binomial Formula (for
square matrices) holds if matrices
commute!
A,BE My, AB= BADefinition
Vector space isa set of objects equipped with two operations: ‘addition’ and “multiplication by
scalar" ('scaling’ that are subject to vector space axioms
(Wt F) voastststeston F=R or C__ Fieldofscalars
$:VxKV3V re ae “muripcation by scalar" ule or combing
abcVioatbey mmomunmwstsne NER ae Vey ae Y marmnencatineanaiesnaer
FAVE ag vets HEV LRG
Vector space axioms:
A(ua) = (Aya, YaeV, Ver
atb=bta, va,beV gmmtsorvoycensusparcuns eso ott seni youn sae
(a+b) +e=a+(b+c), Vab,ceV la=a, YaeV "Scaling by!" axiom
‘essociativty: you ean carry out
‘alton in ary order
Aa+b)=Aat+Ab, VAEF, vabeV
SOEV: atO=a, YacV
vacV, 3-acV: a+(-a)=0 (Q+u)a=Aatpa, VAweR, YaeVPrimary Examples of Vector
Spaces rep atau Nac supa
Primary Example 1; =7Feete="
n-tuples (tnt) nen
Feflnnler, eF,
sekitonrule: (6h) +(H,
rubisiconts a,
Primary Example 3:
Space of mXn matrices
Mat®)={A= (oa) en en €F
then onan menon bys
In this course, you can assume without proof
that every primary example is a confirmed
vector space satisfying every axiom,
Primary Example 2:
Polynomials of degree n
or less
‘There are infinitely many examples of vector spaces, Within them we single out a few special
‘examples primary examples Special feature of primary vector spaces ithe fact that they are not
‘embedded into alarger vector space
(A048)
RO) =foa tort tons, CF 50m
Pla) = op Fare tent ey2 and gle) = byt Bet Ba
addition rule: (p+ a)l a) + a: (ag + by) + (a + bi) +... + (+ bee
Sear (ea) = Ante) = Aay + Aare t+ Ange
Primary Example 4: Space of
piece-wise continuous functions
Colot)= (Fl8|-+8, Ff pnceise continuous}
addtionue (+90) =) +02) f,9¢ Glad]
mytpicton’” —anay=Mte) Fe Glad AEF
‘Tuaisonectthe mr att andthe mot
‘rampisont on sot fow ocesacneVector Subspaces Uniike primary vector spaces, vector subspaces ARE embedded into a larger KNOWN
vector space, often a primary vector space. Ths sa significant advantage.
(V,+15F) known larger vector space ‘Asubset of a vector space Vis a vector
subspace if this subset with addition and
multiplication-by-scalar "borrowed from this
larger vector space is a vector space in its own
right, Le, satisfy all of the vector space axioms
W CV sector sub-space iff
(W,+.5F) isavectorspace
Subspace Theorem: Zi S GAH vere ee ans
WV levector sub-space if The advantage of working with vector
subspaces lies in the Subspace
1 OW nonempty Theorem. You do not have to check
every axiom if you use this theorem.
Instead, you only check that the subset
) AWEW, VACR, YacW closedwrtscaling 5 closed with respect to addition and
multiplication-by-scalar.
(2) atbeW YabeW — closed wrt additionExamplel wert
We {xeRt: 42-32 +20} xeR: x
convaton fr arte abi"
(21,22, 24,24)
By Subspace Theorem:
xyeW > day —Sey+2y=0 and 4y,— Sy) +H
X+¥= (ti +o,22+ potatos tet)
Aa + 14) ~ 8(es + a) + (2a + My) = Ar + 4 — Bes —Byy +4
= (401 — 829 + 24) + (Ayr — 8y +44) = 0+0
=o =Gaew> condition (3) satisfied
(0,0,0,0): 4-0-3-040=...=0 + condition () satisfied
7 ee Ax = (Ari, Ara, Ats, 04)
4(r,) — 300) + (ni) =
(421 — 825 +24) =Example 2
wort w {xeR!: 21=0 or zs =o}
x = (1,0,0,0) y = (0,0,1,0) = xeW and yew
‘on the other hand: x+y = (1,0,1,0) ¢W condition (2) failed
By Subspace Theorem: W is not a subspace
Note: conditions (1) and (3) are satisfied in this example
O¢W truestatementMatrix Nullspace, Matrix Col(umn) space, Span
We associate two types of subspaces with every rectangular
‘matinc Matrix Nullspace and Matrix Column) space
Ae Moo(F) mC and mutta = fer
cS FY wd ot= {yer fon se
‘Theorem: Matrix nullspace and Matrix Col-space are vector subspaces.
‘The proof ofthis result is base
lonthe Subspace Theorem
We can describe matric col space alternatively using the language of span
and linear combinations.
A=( nf fe [a Je aemeennes ext umden gen ema
columns of matrix
EF and x= (020.0084)
A= rim} nnae b ot 2
cee
‘The advantage of SPAN language is that can be usedin the context
OF ANY vector space nat spacinealy the vector space of tuples
(48) ads
s}SY Sactoroneecar spe
spins (as toot Aote Me
net}
Theorem: Span ofset Sis
vector subspace of V
Most ofthe vector subspaces you worked with in
Mathematics 16 falls into one of the categories: null
space of a matrix or col space of a matrix span, more
‘general.