How do I optimize the coupling strength between a qubit and a readout resonator?

Optimizing the coupling strength g between a qubit and a readout resonator involves balancing several competing requirements: strong enough coupling for fast, high-fidelity readout (the dispersive shift chi = g^2/delta, where delta is the qubit-resonator detuning, determines the readout signal strength; larger g gives larger chi and better measurement contrast), but weak enough to avoid: Purcell-limited T1 decay (the qubit spontaneously emits through the resonator at rate gamma_Purcell = (g/delta)^2 x kappa; too large g reduces T1), measurement-induced dephasing (the readout photons in the resonator perturb the qubit frequency by chi/n_readout per photon, causing dephasing during measurement; too large chi makes the qubit sensitive to photon number fluctuations), and qubit-resonator hybridization (when g/delta approaches 1, the qubit and resonator become strongly hybridized, and the qubit loses its identity as a clean two-level system; the dispersive approximation breaks down). The optimal coupling is determined by: setting g to achieve a dispersive shift chi of 1-10 MHz (sufficient for single-shot readout in 100-500 ns), with a detuning delta of 1-3 GHz (keeping g/delta approximately 0.05-0.15 in the dispersive regime), and a corresponding g of approximately 50-300 MHz. The coupling is physically implemented through: a coupling capacitor between the qubit and resonator (the capacitance C_g determines g: g = C_g x omega_r / sqrt(C_q x C_r) approximately, where C_q and C_r are the qubit and resonator capacitances), or galvanic (direct electrical) connection through a shared inductor or transmission line section.