The repository contains a MatLab code that is implemented to reconstruct the unknown spectrum demonstrated in our submitted work "Miniaturized Spectrometers with a Tunable van der Waals Junction". This code was archived in this link during the submission process and will not be updated here. Further updates on the code can be found at the following link: https://github.com/fonig/Reconstruction The learning process allows obtaining a response matrix with elements that are defined by the gate voltage and the wavelength. The response matrix and input data (photocurrent as a function of the gate voltage) are connected with a matrix equation with a vector of weight coefficients which determine a desired spectral function. For the solution of the matrix equation, one needs to minimize the residual norm (squared error) by solving the non-negative least square problem. For the stabilization of the minimization procedure, Tikhonov regularization is used. A vector of coefficients that provides a global minimum of the modified residual norm allows reconstructing the considered spectrum by substituting it in the initial expansion in basis functions. More details can be found in our manuscript (Miniaturized Spectrometers with a Tunable van der Waals Junction).