An analog signal having a value A is converted into an n-bit digital signal. An optical calculation part performs a part of a calculation shown below for an A/D conversion as below.U.sub.i =[{{.SIGMA.(W.sub.ij .times.X.sub.j)+h.sub.i }.times.V}+A].times.Si . . . (1)The calculation of the equation (1) is performed fori=0, 1, . . . , n-1 respectively;the W.sub.ij is 0 if i.gtoreq.j.gtoreq.0, or -(2**j) if j>i.gtoreq.0; hi is -(2**i), or -{(2**i)-.epsilon.}(.vertline..epsilon..vertline..ltoreq.1);V and S.sub.i respectively have any desired positive values and the said .SIGMA. represents a summation of each expression following thereto for j=0, 1, . . . , n-1. The thresholding compares the result U.sub.i of the calculation of each equation (1) to a threshold value, and then 1 or 0 is selected. The result of the selection is then assigned to X.sub.i. The calculation of the equation (1) is then performed repeatedly until X.sub.i converge on solutions. The solution of each X.sub.i is then provided as digital values of the n-bit digital signal.