Solution

ID: past-exam-of-the-mathematics-course-of-the-university-of-cambridge/2016/iii/paper-111/4/i/solution

Use , the necessary parameter range for the bounds involving . The construction comes from the singular value decomposition of matrix Fourier blocks. For each inequivalent unitary irreducible representation of dimension , consider the Hilbert space with the Hilbert-Schmidt inner product and the operator
Its operator norm is at most one, since is a unitary matrix and . Thus all its singular values lie in .
Two identities control their weighted moments:
Here the expectation in is uniform on the solutions of the constraint, and the trace on used to compute each moment is the ordinary operator trace. We justify the identities explicitly to fix their normalizations and the noncommutative order.
The adjoint operator is , so
For rectangular matrices, the operator has trace , as is seen on the matrix-unit basis. Squaring the previous operator therefore gives
The required representation theory consists of unitarization of a finite-group representation, the Schur orthogonality relations, and the regular representation decomposition. The last gives the character expansion of the identity delta
Summing the two operator traces with weights selects in the first, and in the second. Multiplication by converts each independent-variable expectation to its conditional expectation. This proves . In particular is real and nonnegative, despite the apparently complex summands in its original formula. Since , the nonvacuous hypothesis has .
Select every singular value satisfying . Index the selected pairs by , set , , and choose unit right singular vectors and unit left singular vectors with . Define
Both are matrices, proving (i). Repeated group representations in this list are literally the same chosen representative, rather than different equivalent realizations.

New to topics? Read the docs here!