Abstract: The paper reviews the iterative Image Space Reconstruction Algorithm (ISRA) for solving Linear Inverse Problems with Positive Constraints. The development follows that for the EM algorithm in Vardi and Lee (1993). The algorithm is set down, a range of special cases for particular contexts are listed, convergence issues are discussed, and there is a concluding discussion. The speeds of convergence of EM and ISRA are comparable, although the latter often needs noticeably fewer operations per iteration.
Key words and phrases: EM algorithm, emission tomography, image analysis, image space reconstruction algorithm, integral equations, inverse problems, linear equations, mixtures, motion blurring, portfolio optimization.