Searching for the optimal parameters of nanoelectronic devices is a primal problem in modeling. We solve this problem through the example of the electron ballistic switch in a quantum network model. For this purpose, we use a computing scheme in which closed channels are taken into account. It allows calculating correctly a scattering matrix of the switch and, consequently, the electric currents flowing through it. Without loss of generality, we consider a model of a two-junction switch at room temperature. This device is characterized by localization of the controlling electric field in the domain before branching. We optimize switch parameters using a genetic algorithm. At the expense of optimization, the switch efficiency for InP, GaAs and GaSb reached 77–78%. It is established that, for the considered materials, the volt–ampere characteristics of the device are close to the linear ones at bias voltages of 0–50 mV. It allowed describing with good accuracy electron transport in the switch by means of a 3 × 3 matrix of approximate conductivity. Finally, based on the parameter optimization of the two-junction switch, we formulate the general scheme of modeling nanoelectronic devices in the framework of a quantum network formalism.