spectrum
The spectrum of a matrix or operator is the complete set of its eigenvalues — the collection of all numbers lambda for which A - lambda I fails to be invertible. Think of it as the operator's fingerprint: the precise list of scaling factors it applies along its special directions.
In finite dimensions the spectrum is a finite set, written spec(A) or sigma(A). It consists of exactly the roots of the characteristic polynomial. Note that the spectrum is a set, so it forgets multiplicity: the matrix [3, 0; 0, 3] and the number 3 alone both have spectrum {3}, even though one is 2x2.
The word spectrum becomes far richer in infinite dimensions, where operators can fail to be invertible without having any eigenvector at all. There the spectrum splits into point, continuous, and residual parts. For now, in finite dimensions, spectrum = eigenvalues — but keep that broader meaning in your pocket for the operators track.
Two of the diagonal entries are equal, but the spectrum is a set, so 2 is listed once.
When eigenvalues are listed with their algebraic multiplicities (a multiset rather than a set), the result is sometimes called the spectrum 'with multiplicity'. The plain spectrum is the underlying set.