0% found this document useful (0 votes)
91 views11 pages

MA-310 Linear Algebra: Nazar Khan

The document defines inner products on vector spaces and discusses their properties and applications. It introduces inner products and their axioms. Specific examples of inner products discussed include the standard Euclidean inner product using dot products, weighted Euclidean inner products, and matrix inner products. Angles and orthogonality are also generalized from Rn to general inner product spaces using the inner products. An example computes the angle between two square matrices using a matrix inner product.

Uploaded by

Malik Hammad
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)
91 views11 pages

MA-310 Linear Algebra: Nazar Khan

The document defines inner products on vector spaces and discusses their properties and applications. It introduces inner products and their axioms. Specific examples of inner products discussed include the standard Euclidean inner product using dot products, weighted Euclidean inner products, and matrix inner products. Angles and orthogonality are also generalized from Rn to general inner product spaces using the inner products. An example computes the angle between two square matrices using a matrix inner product.

Uploaded by

Malik Hammad
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/ 11

MA-310 Linear Algebra

Nazar Khan

PUCIT

10. Inner Product Spaces


Inner Product Angles & Orthogonality

Inner Product

We used the dot product of vectors in Rn to define notions


of
I length,
I angle,
I distance, and
I orthogonality.
Now we generalize those ideas to any vector space, not just
Rn .

Nazar Khan Linear Algebra


Inner Product Angles & Orthogonality

Inner Product

An inner product on a real vector space V is a function that


associates a real number < u, v > with each pair of vectors
in V in such a way that the following 4 axioms are satisfied
for all vectors u, v, and w in V and all scalars k.
1. < u, v >=< v, u > [Symmetry]
2. < u + v, w >=< u, w > + < v, w > [Additivity]
3. < ku, v >= k < u, v > [Homogeneity]
4. < v, v >≥ 0 and < v, v >= 0 if and only if v = 0
[Positivity]
A real vector space with an inner product is called a real inner
product space.

Nazar Khan Linear Algebra


Inner Product Angles & Orthogonality

Inner Product
Standard

I Inner product of two vectors u and v in Rn was earlier defined


using the dot product

< u, v >= u · v = u1 v1 + u2 v2 + · · · + un vn

I This is commonly known as the Euclidean inner product or


standard inner product.
I Inner product can be defined in other ways as well – as long as
the defined function satisfies the 4 axioms in the last slide.

Nazar Khan Linear Algebra


Inner Product Angles & Orthogonality

Weighted Euclidean Inner Product

I Defined as

< u, v >= w1 u1 v1 + w2 u2 v2 + · · · + wn un vn

with weights w1 , w2 , . . . , wn .
I Setting all weights to 1 yields the standard Euclidean inner
product.

Nazar Khan Linear Algebra


Inner Product Angles & Orthogonality

Weighted Euclidean Inner Product

I Left figure: Set of points at distance 1 from origin w.r.t


standard Euclidean inner product < u, v >= u1 v1 + u2 v2 .
I Right figure: Set of points at distance 1 from origin w.r.t
weighted Euclidean inner product < u, v >= 19 u1 v1 + 14 u2 v2 .

Nazar Khan Linear Algebra


Inner Product Angles & Orthogonality

Weighted Euclidean Inner Product

I Sketch the unit circle in R2 w.r.t weighted Euclidean inner


1 1
product < u, v >= 25 u1 v1 + 49 u2 v2 .
I Find weighted Euclidean inner products on R2 for which the
"unit circles" are the ellipses shown in the following figures.

Nazar Khan Linear Algebra


Inner Product Angles & Orthogonality

Matrix inner product

I Defined as

< u, v >= Au · Av = (Au)T Av = uT AT Av

I Also called the inner product on Rn generated by A.


I Setting A = I yields the standard Euclidean inner product.
I Setting A as a diagonal matrix yields the weighted Euclidean
inner product. Find A for
< u, v >= w1 u1 v1 + w2 u2 v2 + · · · + wn un vn .
I Can be viewed as standard inner product but after
transforming by A.
I Plays a big role in Machine Learning, Image Processing, and
Computer Vision.

Nazar Khan Linear Algebra


Inner Product Angles & Orthogonality

Angles & Orthogonality


In General inner product spaces

I We have already seen that angle between two vectors in Rn


can be computed using the dot product as
 
−1 u·v
θ = cos
||u||||v||
I Recall that dot product is a specialized form of inner product
which is more general.
I Angle between two vectors in a general inner product space
can be computed using the inner product as
 
−1 < u, v >
θ = cos
||u||||v||

Nazar Khan Linear Algebra


Inner Product Angles & Orthogonality

Angles & Orthogonality


In General inner product spaces
u·v
I Recall that −1 ≤ ||u||||v|| ≤ 1.
<u,v>
I For general inner products −1 ≤ ||u||||v|| ≤ 1 also holds.

I Norm (or length) is defined by ||v|| = < v, v >.
I Distance between two vectors becomes

d(u, v) = ||u − v|| = < u − v, u − v >.
I Properties of length and distance also carry over in general
spaces.
I ||u + v|| ≤ ||u|| + ||v|| (Triangle inequality for vectors)
I d(u, v) ≤ d(u, w) + d(w, v) (Triangle inequality for distances)
I < u, v >= 0 implies orthogonality.
I Note that orthogonality depends on the definition of the inner
product.
I Compute < u, v > for u = (1, 1) and v = (1, −1) using
standard and weighted Euclidean inner product definitions.

Nazar Khan Linear Algebra


Inner Product Angles & Orthogonality

Example
Angle between square matrices

I We have seen that matrices satisfy the 10 axioms of vector


spaces.
I For n × n matrices, an inner product can be defined as
< u, v >= trace(U T V ) = u11 v11 + u22 v22 + · · · + unn vnn .
I Find the cosine of the angle between the vectors
   
1 2 −1 0
u=U= and v = V =
3 4 3 2

I This gives us a method for computing similarities between


objects in general vector spaces. Prerequisite: inner product
needs to be defined first.

Nazar Khan Linear Algebra

You might also like