The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of...

33
The QR iteration for eigenvalues

Transcript of The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of...

Page 1: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

The QR iteration for eigenvalues

Page 2: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

. . .

The intention of the algorithm is to perform a sequence of similarity

transformations on a real matrix so that the limit is a triangular matrix.

Page 3: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

. . .

The intention of the algorithm is to perform a sequence of similarity

transformations on a real matrix so that the limit is a triangular matrix.

If this were possible then the eigenvalues would be exactly the

diagonal elements.

Page 4: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

But it may not be possible:

Page 5: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

But it may not be possible:since • Real matrices may have complex eigenvaluesand• All of the arithmetic in the algorithm is real

Page 6: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

But it may not be possible:since • Real matrices may have complex eigenvaluesand• All of the arithmetic in the algorithm is real

There is no way the real numbers can converge to anything other than real numbers.

Page 7: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

But it may not be possible:since • Real matrices may have complex eigenvaluesand• All of the arithmetic in the algorithm is real

There is no way the real numbers can converge to anything other than real numbers.

That is: It is impossible for the limit to have numbers with non-zero imaginary parts.

Page 8: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

But it may not be possible:since • Real matrices may have complex eigenvaluesand• All of the arithmetic in the algorithm is real

There is no way the real numbers can converge to anything other than real numbers.

That is: It is impossible for the limit to have numbers with non-zero imaginary parts.

If any eigenvalues have non-zero imaginary parts, the sequence will not converge to them.

Page 9: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

If any eigenvalues have non-zero imaginary parts, the sequence will not converge to them.

Page 10: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

If any eigenvalues have non-zero imaginary parts, the sequence will not converge to them.

Are we dead?

Page 11: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

If any eigenvalues have non-zero imaginary parts, the sequence will not converge to them.

Are we dead?

Nope, but we have to modify our expectations.

Page 12: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

. . .

Instead of the limit being an upper triangular matrix

Page 13: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

. . .

Instead of the limit being an upper triangular matrix

it is block upper triangular

Page 14: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

. . .

Instead of the limit being an upper triangular matrix

it is block upper triangular

Page 15: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

. . .

Instead of the limit being an upper triangular matrix

it is block upper triangular

The blocks are 2 by 2 and…

Page 16: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

. . .

Instead of the limit being an upper triangular matrix

it is block upper triangular

The blocks are 2 by 2 and…the eigenvalues we want are the

complex conjugate pairs of eigenvalues of the blocks

Page 17: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

. . .

This actually presents no major troubles.

The blocks are 2 by 2 and…the eigenvalues we want are the

complex conjugate pairs of eigenvalues of the blocks

Page 18: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

So this is the algorithm in a mathematical form(as opposed to form representing what happens in

storage):

Page 19: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

So this is the algorithm in a mathematical form(as opposed to form representing what happens in

storage):

0. Set A1 = AFor k = 1, 2, …

1. Do a QR factorization of Ak: Ak = QkRk

2. Set Ak+1 = RkQk

Page 20: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

This is the algorithm in a programming form:

For k = 1, 2, …1. Do a QR factorization of A: A → QR2. Set A ← RQ

Page 21: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Since Ak = QkRk

QkTAk = Qk

TQkRk= Rk

Page 22: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Since Ak = QkRk

QkTAk = Qk

TQkRk= Rk

but then

Ak+1 = RkQk= QkTAkQk

Page 23: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Since Ak = QkRk

QkTAk = Qk

TQkRk= Rk

but then

Ak+1 = RkQk= QkTAkQk

and since Qk is orthogonal, QkT = Qk

-1 and

Page 24: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Since Ak = QkRk

QkTAk = Qk

TQkRk= Rk

but then

Ak+1 = RkQk= QkTAkQk

and since Qk is orthogonal, QkT = Qk

-1 and

Ak+1 = Qk-1AkQk

Page 25: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Since Ak = QkRk

QkTAk = Qk

TQkRk= Rk

but then

Ak+1 = RkQk= QkTAkQk

and since Qk is orthogonal, QkT = Qk

-1 and

Ak+1 = Qk-1AkQk

Ak+1 is similar to Ak

Page 26: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Ak+1 is similar to Ak

Page 27: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Ak+1 is similar to Ak

is similar to Ak-1

Page 28: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Ak+1 is similar to Ak

is similar to Ak-1

is similar to Ak-2

Page 29: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Ak+1 is similar to Ak

is similar to Ak-1

is similar to Ak-2

. . .

is similar to A1 =A

Page 30: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Ak+1 is similar to Ak

is similar to Ak-1

is similar to Ak-2

. . .

is similar to A1 =A

We have a sequence of similar matricesA1, A2, A3, … tending to a block triangular matrix

whose eigenvalues are easy to obtain.

Page 31: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Not only are the matrices in the sequence similar they are

orthogonally similar - the similarity transformation is orthogonal

Page 32: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Not only are the matrices in the sequence similar they are

orthogonally similar - the similarity transformation is orthogonal

Since orthogonal matrices preserve lengths, this means:

• The matrices of the sequence do not get very large or very small, and

• The computations are done more accurately.

Page 33: The QR iteration for eigenvalues. ... The intention of the algorithm is to perform a sequence of similarity transformations on a real matrix so that the.

Let’s see the algorithm in action.

The sizes will be indicated by color.

Since, what will be interesting is seeing the subdiagonal components get smaller, we will use

a logarithmic scale that emphasizes small numbers.

1. (Unshifted) QR2. Corner shifted QR3. Double shift QR