Scribd
Upload
53 views26 pages

Cogsci 131 Linear Algebra: Tom Griffiths

This document provides an outline for a linear algebra course, beginning with an introduction to vectors and matrices. Key concepts covered include: - Vectors represent sets of values and can be added, scaled, and combined linearly - Matrices allow multiplying vectors to transform their direction and length - Inner and outer products generalize the dot product to matrices and vectors - Matrix operations like addition and multiplication follow rules of linearity - Eigenvectors and eigenvalues describe vectors that change only in length when multiplied by matrices - Special matrices like the identity and diagonal matrices have important properties

Uploaded by

Brian
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
53 views26 pages

Cogsci 131 Linear Algebra: Tom Griffiths

This document provides an outline for a linear algebra course, beginning with an introduction to vectors and matrices. Key concepts covered include: - Vectors represent sets of values and can be added, scaled, and combined linearly - Matrices allow multiplying vectors to transform their direction and length - Inner and outer products generalize the dot product to matrices and vectors - Matrix operations like addition and multiplication follow rules of linearity - Eigenvectors and eigenvalues describe vectors that change only in length when multiplied by matrices - Special matrices like the identity and diagonal matrices have important properties

Uploaded by

Brian
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
You are on page 1/ 26

CogSci 131

Linear algebra

Tom Griffiths

Outline
Vectors
Break
Matrices

Vectors
A vector is an efficient way of expressing a set
of values of different features or dimensions
" x1 %
#1&
x =$ '
e.g. x = % (
# x2 &
$ 1'
Often, it is helpful to think of a vector as
corresponding to a point in space
x2

x1
x

Multiplying a vector by a scalar


" ax1 %
ax = $ '
# ax 2 &
# 1 & # 1.5 1 & # 1.5 &
e.g.1.5x = 1.5% ( = %
(=%
(
1
1.5

1
1.5
$
'
$
'
$
'

x2

x1
x
ax

Adding vectors
" x1 + y1 %
x + y =$
'
# x2 + y2 &
# 1 & # 2 & # 1+ 2 & # 3 &
% (+% (=%
(=% (
1
1
1+
1
2
$
'
$
'
$
'
$
'

y
x

x+y

Linear combinations
"1%
v1 = $ '
#2&

" 3%
v2 = $ '
# 2&

"9%
u=$ '
#10&

Can we find c1 and c2 such that u = c1v1+c2v2?

c1,c2 > 0

any c1,c2

Linear combinations
A vector v is a linear combination of v1, , vn if
there are c1, , cn s.t. v = c1v1 + c2v2 + + cnvn
The set of all linear combinations is the span
of v1, , vn
e.g. the span of

"1%
$ '
#0&

and

"0%
$ 'is
#1&

all vectors in 2D space

A set of vectors spanning a space provides a

basis for that space


e.g.

"1%
$ '
#0&

and

"0%
$ ' form
#1&

a basis for 2D space

Inner products

" y1 % T
y = $ ' y = ( y1 y 2 )
# y2 &
transpose
(y in Matlab)
" x1 % T
x = $ ' x = ( x1 x 2 )
x
#
2&

inner
"
%
x
1
y T x = ( y1 y 2 )$ ' = y1 x1 + y 2 x 2
# x2 &

Length

x T x = x12 + x 22

Based on the Pythagorean theorem, we can


define the length (or norm) of a vector to be
T

2
1

x = x x= x +x

2
2

Angle
The cosine of the angle between two vectors
is given by
xT y
cos( ) =
x y
So

x T y = x y cos( )

Two vectors x and y are orthogonal when


xTy = 0, corresponding to cos() = 0, meaning
thatthe vectors form a right-angle

Algebra of inner products


xT y = yT x

commutativity

c(x T y) = (cx)T y = x T (cy)


T

x (y1 + y 2 ) = x y1 + x y 2

linearity

Break

Up next:
Matrices

Matrices
y = Wx
y2

y
y1

" y1 %
y =$ '
# y2 &

x2

x1
x

" x1 %
x =$ '
# x2 &

" w11 w12 %


W =$
'
# w 21 w 22 &

Matrix multiplication
y = Wx
y2

" x1 %
" y1 %
" w11 w12 %
y =$ ' W =$
' x =$ '
# x2 &
# y2 &
# w 21 w 22 &

y1

y = Wx
x

x1
x

y1 = w11 x1 + w12 x 2
y 2 = w 21 x1 + w 22 x 2

Matrix multiplication

Inner and outer products


" y1 % T
y = Wx y = $ ' y = ( y1 y 2 )
y2 &
#
transpose
y
y
(y in Matlab)
" x1 % T
x
=
x
=
x
x
(
$
'
1
2)
y
x
#
2&

inner

"
%
x
1
T
y
x = ( y1 y 2 )$ ' = y1 x1 + y 2 x 2
x
# x2 &

outer

" y1 %
" y1 x1 y1 x 2 %
x
T
yx = $ '( x1 x 2 ) = $
'
x
# y2 &
# y 2 x1 y 2 x 2 &

Basic operations on matrices


Multiplication by a scalar

" aw11
aW = $
# aw 21

aw12 %
'
aw 22 &

Addition of matrices

" w11 + m11


W +M =$
# w 21 + m21

w12 + m12 %
'
w 22 + m22 &

Algebra of matrices and vectors


W(ax) = aWx
W(x + y) = Wx + Wy

(W + M)x = Wx + Mx

Matrix multiplication
Basic rules follow from multiplication of a
vector by a matrix

Algebra of matrix multiplication


W(aM) = aWM

W(M + N) = WM + WN

(M + N)W = MW + NW

Linearity
A function f is linear if:
f(cx) = cf(x)
f(x + y) = f(x) + f(y)
A function using matrix multiplication (with
inner products as a special case) is linear
In fact, any linear function can be represented
using matrix multiplication
Linear algebra: the math of linear functions

Eigenvectors and eigenvalues


Just think about square matrices mapping
from a space back to itself
Most vectors will change direction and length,
but some will just change length
these are the eigenvectors

Wv = v
eigenvector

eigenvalue

Eigenvectors and eigenvalues


Examples:

Each matrix can have many eigenvectors and


corresponding eigenvalues
e.g. the second matrix shown above has
"1%
"0%
eigenvectors $0' and $1' , with eigenvalues 3 and 4
# &
# &
any scalar multiple of an eigenvector is an
eigenvector, so typically the length is fixed

Some special kinds of matrices


The identity matrix maps every vector to itself
" 1 0%
I=$
'
# 0 1&
A diagonal matrix only has non-zero entries
along the diagonal
eigenvalues
are diagonal entries, eigenvectors

"1%
"0%
are $0' and $1'
# &
# &
" 3 0%
e.g.

D=$
'
#0 4 &

Thursday
Semantic networks and spreading activation
We will use some linear algebra

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy