<< Chapter < Page
  Signal theory   Page 1 / 1
Chapter >> Page >
Description of the Gram-Schmidt procedure to formulate orthonormal bases.

The Gram-Schmidt algorithm or procedure is used to find an orthonormal basis for the subspace [ S ] , even if S is not linearly independent. The algorithm is formally defined as follows:

  • Inputs : Set of vectors S = s 1 , , s n .
  • Outputs : Orthonormal basis elements W = w 1 , , w n that span the same space: [ W ] = [ S ] .
  • Procedure:
    1. Take the first element of the set and divide it by its norm (that is, normalize the element):
      w 1 = s 1 s 1 .
    2. Take the second element and subtract the projection into the first basis element:
      t 2 = s 2 - s 2 , w 1 w 1 .
      Normalize the result t 2 :
      w 2 = t 2 t 2 .
      It is easy to check that w 1 and w 2 are orthonormal:
      w 1 , w 2 = w 1 , t 2 t 2 = 1 t 2 w 1 , s 2 - s 2 , w 1 w 1 = 1 t 2 w 1 , s 2 - s 2 , w 1 ¯ w 1 , w 1 = 1 t 2 w 1 , s 2 - s 2 , w 1 ¯ = 1 t 2 · 0 = 0 .
      Thus, w 1 and w 2 are orthonormal.
    3. The second step is repeated for each additional element; i t h element follows the following formula:
      t i = s i - j = 1 i - 1 s i , w j w j , w i = t i t i .

When the set S includes linearly dependent vectors, some of the unnormalized vectors t i = 0 , as the projections will cancel out with some elements of S . As a result, the number of vectors needed will be higher than the dimensionality of the space; in other words, the dimensionality of [ S ] will be smaller than the cardinality of the set | S | .

Example 1 Let S = 1 , t , t 2 , where s 1 = 1 , s 2 = t and s 3 = t 2 . We can therefore write the set of quadratic functions as Q = [ S ] , and S C ( T ) , where T = [ 0 , 1 ] . Recall that for this space, the inner product is written as x , y = 0 1 x ( t ) y ( t ) d t . We obtain a basis for Q using the Gram-Schmidt procedure: it requires us to compute several norms and inner products on the way.

  1. Solve for w 1 : s 1 = s 1 , s 1 = 0 1 1 · 1 d t = 1 , and so w 1 ( t ) = s 1 ( t ) s 1 = 1 1 = 1 .
  2. Solve for w 2 :
    s 2 , w 1 = 0 1 t · 1 d t = t 2 2 t = [ 0 , 1 ] = 1 2 , t 2 ( t ) = s 2 ( t ) - s 2 , w 1 w 1 ( t ) = t - 1 2 · 1 = t - 1 2 , t 2 = t 2 , t 2 = 0 1 t - 1 2 · t - 1 2 d t = 0 1 t 2 - t + 1 4 d t = t 3 3 - t 2 2 + t 4 | t = [ 0 , 1 ] = 1 12 , w 2 = t - 1 2 1 12 = 12 t - 12 2 = 2 3 t - 3 .
    It is easy to check that w 2 has unit norm and is orthogonal to w 1 .
  3. Solve for w 3 :
    s 3 , w 1 = 0 1 t 2 · 1 d t = t 3 3 | t = [ 0 , 1 ] = 1 3 , s 3 , w 2 = 0 1 t 2 · 2 3 t - 3 d t = 3 2 t 4 - 1 3 t 3 | t = [ 0 , 1 ] = 1 2 3 , t 3 = s 3 - s 3 , w 1 w 1 - s 3 , w 2 w 2 = t 2 - 1 3 - 1 2 3 · 2 t 3 - 3 = t 2 - t - 1 6 , t 3 = t 3 , t 3 = 0 1 t 2 - t - 1 6 2 d t , w 3 ( t ) = t 3 t 3 .
    We can check that [ W ] = Q .

Get Jobilize Job Search Mobile App in your pocket Now!

Get it on Google Play Download on the App Store Now




Source:  OpenStax, Signal theory. OpenStax CNX. Oct 18, 2013 Download for free at http://legacy.cnx.org/content/col11542/1.3
Google Play and the Google Play logo are trademarks of Google Inc.

Notification Switch

Would you like to follow the 'Signal theory' conversation and receive update notifications?

Ask