Let θ = 1 9 9 6 2 π . Evaluate the determinant of the 1 9 9 6 × 1 9 9 6 matrix I + A , where I is the 1 9 9 6 × 1 9 9 6 identity matrix and A = ( a j k ) has entries a j k = c o s ( j θ + k θ ) for all j , k .
This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try
refreshing the page, (b) enabling javascript if it is disabled on your browser and,
finally, (c)
loading the
non-javascript version of this page
. We're sorry about the hassle.
nice solution. we can directly use the result about circulant matrices. A is a circulant matrix. we also know the eigenvalues of a circulant matrix. ones we know the eigenvalues of A we know the eigenvalues of I+A. Determinant of I+A is equal to the product of the eigenvalues of I+A.
Log in to reply
Very true, but this is not a circulant matrix, which is why the eigenvalue/eigenvector system is more complicated. If A were circulant, then the v ( k ) would be the eigenvectors. However A is not circulant. However P A is a circulant matrix, where P is the matrix with 1 s on the top-right/bottom-left diagonal and 0 s everywhere else. This is why A v ( k ) is a multiple of v ( − k ) , and not simply of v ( k ) .
@Mark Hennings, have you ever written the Putnam Exam? I'm just curious because you seem so good at math. Also, btw, I'm not posting any more of these problems because I'm not really gaining anything from posting them.
Problem Loading...
Note Loading...
Set Loading...
Suppose that N ≥ 3 and consider the N × N matrix A with components a j k = cos ( N 2 π ( j + k ) ) 1 ≤ j , k ≤ N If we define X r = cos ( N 2 π ( r + 2 ) ) r ∈ Z then X r + N = X r for all r , and we have a j k = X j + k − 2 1 ≤ j , k ≤ N Let us define ζ = e N 2 π i and define the vector v ( k ) ∈ C N by v ( k ) j = ζ k ( j − 1 ) 1 ≤ j ≤ N It is well-known that { v ( k ) : 1 ≤ k ≤ N } is an orthogonal basis for C N . If we define Λ j = k = 1 ∑ N X k ζ k j 1 ≤ j ≤ N then we can show that A v ( k ) = { Λ k v ( N − k ) Λ N v ( N ) 1 ≤ k ≤ N − 1 k = N Thus C N can be split into A -invariant subspaces:
and hence there exists a basis for eigenvectors for A , and hence for I + A . Thus the eigenvalues of I + A are
1 + Λ N and (if N is even) 1 + Λ 2 1 N ,
1 + Λ k Λ N − k and 1 − Λ k Λ N − k for all 1 ≤ k < 2 1 N .
Thus we deduce that d e t ( I + A ) = ⎩ ⎪ ⎪ ⎪ ⎨ ⎪ ⎪ ⎪ ⎧ ( 1 + Λ N ) 1 ≤ k < 2 1 N ∏ ( 1 − Λ k Λ N − k ) ( 1 + Λ N ) ( 1 + Λ 2 1 N ) 1 ≤ k < 2 1 N ∏ ( 1 − Λ k Λ N − k ) N o d d N e v e n We now calculate that Λ k = j = 1 ∑ N X j ζ k j = 2 1 j = 1 ∑ N ( ζ j + 2 + ζ − j − 2 ) ζ k j = 2 1 ζ 2 j = 1 ∑ N ζ ( k + 1 ) j + 2 1 ζ − 2 j = 1 ∑ N ζ ( k − 1 ) j = ⎩ ⎨ ⎧ 2 1 N ζ − 2 2 1 N ζ 2 0 k = 1 k = N − 1 o . w . from which we deduce that d e t ( I + A ) = 1 − 4 1 N 2 N ≥ 3 irrespective of whether N is even or odd. With N = 1 9 9 6 we obtain the answer is − 9 9 6 0 0 3 .