Given a unitary U with eigenvalue e^{2πiφ}, QPE estimates φ using n ancilla qubits. Each ancilla qubit j applies controlled-U^{2^j} to the eigenstate, accumulating phase 2πiφ·2^j. An inverse QFT then converts these phases into a binary fraction approximation of φ. With n qubits, precision is 1/2^n. The probability distribution over measurement outcomes |k⟩ is P(k) ∝ |sin(π(k/2^n - φ)·2^n)|² / |sin(π(k/2^n - φ))|² — a Dirichlet kernel peaked at k = ⌊φ·2^n⌉.