Quantum channels are trace-preserving completely positive linear maps between Banach spaces of trace-class operators (Schatten classes of order 1); these are noncommutative analogs of Markov operators in classical probability theory. They also play the role of dynamical maps in quantum theory [1, Chap. 6]. The main characteristics determining the information properties of a quantum channel include its classical entanglement-assisted and unassisted capacities. The classical (unassisted) capacity C(Φ) of a channel Φ determines the limit rate of classical information transmission through Φ with any block coding at the input and the corresponding measurement at the output, and the classical entanglement-assisted channel capacity Cea(Φ) supposes, in addition, the presence of an entangled state between the input and the output of the channel Φ (a detailed description of transmission protocols can be found in [1, Chap. 8]). Since entanglement is an additional resource, it follows that Cea(Φ) ≥ C(Φ) for any channel Φ. LetH be a separable Hilbert space. By T(H) we denote the Banach space of all trace-class operators onH and byS(H), the subset of T(H) consisting of all positive operators with trace 1; we refer to such operators as quantum states and denote them by Greek letters ρ, σ, . . . . We denote a set of quantum states {ρi} with probability distributions {πi} by {πi, ρi} and call it an ensemble of states; the state ρ = ∑ i πiρi is called the average state of the ensemble {πi, ρi}. A quantum channel is a trace-preserving completely positive linear map Φ: T(HA) → T(HB) [1, Chap. 6]. Let H(ρ) be the von Neumann entropy of a state ρ, and let H(ρ‖σ) be the quantum relative entropy of states ρ and σ. For a given channel Φ and any ensemble {πi, ρi} of input quantum states, the output χ-quantity is determined by the expression