When a light intensity upon a perturbation is detected, an error calculation/correction unit (85) in a control unit (8) corrects and updates the above-described initial manipulated variables based on perturbation manipulated variables and manipulated variables, i.e., operation manipulated variables to obtain the maximum light intensity from the light intensity value at each perturbation manipulated variable, thereby adjusting the tilt angle of a mirror. More specifically, assuming that the time series data of an acquired output light intensity can be approximated to a cosine function, the error calculation/correction unit (85) calculates a phase difference θ between the cosine function and a sine or cosine function used to set x- and y-axis perturbation patterns for a circular trajectory perturbation. Manipulated variables at coordinates defined by the phase difference θ and polar coordinates of a radius voltage to perturb the mirror are calculated. Voltage values at coordinates defined by a function for setting the driving voltages of the mirror (230) are calculated and set as the driving voltages for one output port.