I. Introduction
In the design of magnetic systems with purposes of varied nature, axial symmetry is a feature common to be found. It offers a simplicity that is advantageous both from the constructive and the theoretical perspectives, as it has been shown for the construction of magnetic lenses [1], construction of an electromechanical actuator [2], containment of low-density plasma [3], calculation of the self-energy in a sphere with no volume charge in the magnetization [4], and design of thick solenoids as sources of constant magnetic field [5]. Particularly, magnetic resonance imaging (MRI) has driven the demand for spatially homogeneous magnetic fields, leading to the designs of magnet sources with this kind of symmetry: starting with four thick solenoids [6] for the very first whole-body MRI scanner at 0.15 T, through a combination of coils, permanent magnets, and iron yokes [7] until to a well-known Helmholtz pair of coils [8]. Still within the realm of MRI, the Maxwell pair of coils is, on the other hand, another example of designs with this symmetry, but with the goal of producing spatially uniform gradients of magnetic fields [9]. Both pairs of coils are also the result of an optimization. In general terms, a procedure of optimization is the search of a minimum (global or local) in a set of variables over an objective function, which complies with several constraints imposed upon it. Within the realm of the type of magnetic sources hereby concerned, such a procedure has been executed over solutions for the magnetic field obtained through finite-element analysis (FEA) [2], [10] or through a series of zonal harmonics for the magnetic field [5]. From a more theoretical perspective, an optimization method was proposed by Jensen [12] for the same type of source based on a generalized family of orthonormal functions, which allows him to derive both the Halbach structure [11] and a tightly assembled arrangement of magnet rings of counteracting remanence out of the same formalism. The variety of methods, aiming either at the optimization itself or at the calculation of the field, is under the current state of the research noticeable.