math

Show that if x is a nonzero column vector in R^n, then the nxn matrix:

A = I - 2/||x||^2 * xx^T
is orthogonal.

Notation key:
||x|| = norm of x
x^T = transpose of x
I = identity matrix.


Let me try to convince a math student to use "physics" notations that many mathematicians don't like.

Let's work with the elements of matrices and use a notation for that instead of the matrix itself. What we do in physics is we attach indices to the matrix. We write:

A_{i,j}

to denote the matrix element in the i-th row and j-th column (starting from the upper left corner go i steps down and j steps to the right).

The unit matrix is denoted by the Kronecker delta symbol:

delta_{i,j}

which is zero if i is not equal to j and is one if they are equal.

Finally we use the Einstein summation convention for repeated indices.If
C = A B, you can write this as:

C_{i,j} = A_{i,k}B_{k,j}

the repeated index k is summed over.

You can write the square of the norm of x as the inner product with itself, so:

||x||^2 = x_{k}x_{k}


The matrix A in this problem can be written in the "index notation" as:

A_{i,j} =
delta_{i,j}-2/||x||^2 x_{i}x_{j}

You see that we now don't need that stupid T symbol on the x. We don't need to indicate that the index j runs from left to right so that you would have to rotate the column vector x so that it lies on his side.

For an orthogonal matrix the matrix product with its transpose is the unit matrix. The transpose of A is A itself in this case, as you easily see by interchanging i with j:
A_{i,j} = A_{j,i}.

In the general case you need to show that:

A_{i,k}A_{j,k} = delta_{i,j}

In our case A_{j,k} = A_{k,j} so this reduces to an ordinary matrix product, but that doesn't really matter.

If you work out the product (and sum over the repeated index) you obtain four terms.
The product of the deltas gives you:

delta_{i,j}

The three other terms cancel. You obtain twice a product with a delta and the second term in A_{i,j} which after summation of the repeated index gives you:

-4x_{i}x_{j}/||x||^2

The product of the two seciond terms in A gives you

4/||x||^4x_{i}x_{k}x_{k}x_{j}

Summation over the repeated index k gives a factor ||x||^2 and this then cancels the other term.

Thanks Count. I finally had time to go through and understand this proof. I now understand, but this wasn't simple. I really appreciate the help.

  1. 👍 0
  2. 👎 0
  3. 👁 66

Respond to this Question

First Name

Your Response

Similar Questions

  1. math

    If A^TA is an invertible matrix, prove that the column vectors of A are linearly independent. You know that if statement X implies statement Y then that is equivalent to Not(Y) implies Not(X). You can start by taking the column

    asked by mathstudent on January 8, 2007
  2. precalc

    Given a square matrix M, we say that a nonzero vector v is an eigenvector of M if Mv=kv for some real number k. The real number k is called the eigenvalue of v with respect to M. 1. Let v be an eigenvector of the matrix M with

    asked by katie on August 7, 2016
  3. linear(hw check)

    determine if v1= [ 2 1 0] v2=[ -1 1 3] v3=[ 0 -1 6] spans the vector space of rows with three real entries which has dimension 3. so I wanted to make sure I did this correct. First I created a matrix with v1,v2,v3 as the columns

    asked by kirsten on March 4, 2015
  4. Algebra II-Please check calcs

    Could someone check this matrix calculation The first matrix dimension is 1 by 3 row 1 = 1 row 2 = 7 row 3 =3 Second matrix is 1 by 3 Row 1 column one =2 row 1 column two = -5 row one column three = 5 my calculation is that it

    asked by Ray Scholz on January 26, 2011
  5. calculus

    A vector field with a vector potential has zero flux through every closed surface in its domain. it is observed that although the inverse-square radial vector field F = (e^r)/(p^2) satisfies div(F) = 0, F cannot have a vector

    asked by Tom on October 25, 2009
  6. algebra

    When i have a matrix 4 X 4 and i have to multiply it by a 4 X 3 i know that the product size has to be a 4x 3 so do i do a row times column or how ? It's easiest to learn using the notations for tensors. For any matrix X denote

    asked by student on June 17, 2007
  7. Math

    The vector a1=(1,1,0) and a2=(1,1,1) span plane in R^3. Find the projection matrix P onto the plane, and find nonzero vector b that is projected to zero.

    asked by Melvin on January 2, 2011
  8. Algebra

    How would you prove this theorem: The column space of an m x n matrix A is a subspace of R^m by using this definition: A subspace of a vector space V is a subset H of V that has three properties: a) the zero vector of V is in H.

    asked by Min on June 19, 2007
  9. Java programming

    can anybody help me.. Multidimensional Array Use a two-dimensional array (for example of 3 by 3) to simulate the addition operation of two matrices. Sample output: the first matrix: 1 2 3 1 0 1 1 2 1 the second matrix: 1 1 1 2 2 2

    asked by noh on April 14, 2013
  10. math , help

    can someone show me how to solve this: directions: pivot once as indicated in each simplex tableau. Read the solution from the result. the number that is highlighted is row 2 column 3 which is the 5 the matrix is a 4 X 8 under

    asked by jas on June 13, 2007

More Similar Questions