Rotations and complex eigenvalues Math 130 Linear Algebra

D Joyce, Fall 2012

The eld C of complex numbers is what is called an algebraically closed eld which means every polynomial of degree n has all n roots (when multiplicities are counted). Eigenvalues for linear operators are so important that well extend our scalars from R to C to ensure there are enough eigenvalues.

Example 2. Eigenvalues of a general rotation in 2 Rotations are important linear operators, but R . Recall that the matrix transformation x Ax, they dont have real eigenvalues. They will, howwhere ever, have complex eigenvalues. cos sin A= , Example 1. Well look at general notations next, sin cos but lets warm up with a counterclockwise rotation by 90 . Thats the matrix transformation x Ax, describes a rotation of the plane by an angle of . Lets nd the eigenvalues of this generic rotawhere 0 1 tion of the plane. The characteristic polynomial A= , 1 0 is det(A I) which equals Its characteristic polynomial is det(A I) which equals 1 = 2 + 1. 1 There are no real roots of this polynomial 2 + 1, only the imaginary roots i. Thus this rotation has no real eigenvalues and no real eigenvectors. How can we continue on? We can treat the matrix as a matrix over the complex numbers C instead of just the real numbers R. Now it describes a linear transformation C2 C2 . It has two complex eigenvalues, i, that is, the spectrum for a 90 counterclockwise rotation is the set {i, 1}. Lets nd the eigenvalues for the eigenvalue 1 = i. Well row-reduce the matrix A 1 I. A 1 I = i 1 1 i 1 i 0 0 cos sin = (cos )2 + sin2 . sin cos Well set that to 0 and solve for . We quickly run into problems, as (cos )2 = sin2 has no real solutions. Thus, there are no real eigenvalues for rotations (except when is a multiple of , that is the rotation is a half turn or the identity). To get the missing eigenvalues, well treat the matrix as a matrix over the complex numbers C instead of just the real numbers R. Then it describes a linear transformation C2 C2 , and we can continue on. cos = i sin = cos i sin = e i We get two complex eigenvalues. Each of these will have an associated eigenspace. Lets nd the eigenspace for 1 = cos + i sin . Well solve the equation (A1 )x = 0 by row-reducing the matrix 1

Thus, the solutions to this system, that is, the 1 eigenspace, is the set of vectors in C2 of the form (z, w) = (iw, w) where w is an arbitrary complex number. Likewise, you can show that the 2 -eigenspace, where 2 = i, consists of vectors (z, w) = (iw, w) where w is arbitrary.

A 1 I. A 1 I = = cos sin (cos + i sin )I sin cos i sin sin sin i sin i 1 1 i 1 i 0 0

Thus, the generic solution to this system is (z, w) = (iw, w) where w is an arbitrary complex number. Generally speaking, nding the complex eigenspaces for a rotation isnt as important as nding the eigenvalues. Math 130 Home Page at http://math.clarku.edu/~djoyce/ma130/