Answers are grounded in this paper and cite their sources.
We study coherent transport controlled through an accessible boundary of a fixed finite network. Eliminating the constrained boundary gives the exact interior Hamiltonian L R = L ΩΩ -L ΩB (L BB + R) -1 L BΩ . For affine Hermitian response families, cospectrality of two sites is either persistent or confined to a proper real algebraic set; along a one-parameter direction H, the nonpersistent case has at most rank(H)(|Ω|-1) admissible values. The contact-resolvent gain κ R = ∥L ΩB (L BB + R) -1 ∥ controls differential and finite calibration errors. Modulo physically irrelevant scalar shifts, the sharper phase-insensitive error ε ph bounds the implemented transfer probability P at an ideal perfect-transfer time τ by P ≥ cos 2 (τ ε ph ). Its local counterpart is a positive semidefinite control susceptibility: the quadratic loss vanishes on kernel directions, for which the fourth-order coefficient is explicit. At a boundary pole, the active-pole rank counts the diverging ordered interior eigenvalues and determines the inverse-distance growth of κ α , whereas dark poles leave the contact gain bounded. A complete-graph family realizes exact routing from one sender to any number of selectable receivers at a common readout time. An unbalanced double star gives a pole-controlled switch whose off-regime maximum transfer probability, first-maximum time, and Hamiltonian derivative norm scale as O(η 2 ), O(η), and O(η -2 ), respectively, with η = |1 + α| the pole margin.
No takes yet. Share an insight, caveat, or question.
Xing-Kun Song (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: