Multiobjective genetic algorithm approach to optimize beam matching and beam transport in high-intensity hadron linacs

In this article we demonstrate how a multiobjective genetic algorithm, like the nondominated sorting genetic algorithm II (NSGAII), and a selection tool, like the technique for order preference by similarity to ideal solution (TOPSIS), can be employed for beam matching and for optimizing the beam transport in the low and medium energy section of a modern hadron linac including space charge effect. Combining NSGAII with the particle tracking code TRAVEL v4.07, we determine the Pareto optimal front, and then apply TOPSIS for the final selection using the example of the 160 MeV H$^−$ LINAC4 at CERN. We first determine the matching parameters yielding the optimum transport of the 45 keV H$^−$ beam from the ion source to the radio-frequency quadrupole (RFQ) and then onwards at the entrance of the drift tube linac (DTL). Next, we optimize six parameters of the beam phase space to maximize the beam transmission from the exit of the RFQ to the DTL through the medium energy transport section. Finally, we benchmark our predictions with simulations based on the independent TRACE 3-D code and also against a beam experiment, in which the transverse emittance was obtained from a temporary slit-and-grid emittance diagnostic device behind the medium energy beam transport line.