摘要: Dipole magnetic confinement has demonstrated high confinement quality in experiments such as the Levitated Dipole Experiment (LDX), but requires a levitated superconducting coil whose suspension, cryogenics, and position control remain difficult to engineer reliably. The magnetic mirror dipole field scheme removes the need for levitation: the coil is fixed and enclosed in a superconducting shell, which compresses and enhances the field and produces mirror confinement while keeping the field lines closed and dipole-like. We use a Boris orbit-tracking algorithm to scan particle energy, injection position, and pitch angle in four configurations (two with a central field peak, Models I and III; two with an outer-ring peak, Models II and IV) and compare them by three metrics: the maximum confinement energy, the confinement rate in perpendicular-injection scans, and the energy evolution of the loss-cone boundary, including the topology-dependent adiabaticity-breaking threshold. Model II outperforms the other three on all three: its maximum confinement energy is 2.75 MeV, its confinement rate reaches 36.9% in perpendicular-injection scans, and its adiabaticity-breaking threshold (~100 eV) is an order of magnitude above that of the center-peaked configurations (~10 eV). Above a topology-dependent threshold energy, the numerical loss-cone boundary departs from the classical adiabatic prediction, and no simple analytical correction captures the departure; confinement performance must therefore be evaluated with orbit-following simulations. The comparison identifies magnetic field topology as the primary factor governing confinement and favors the outer-ring-peaked configuration for magnetic mirror dipole field fusion devices.