In neutron spectrometry, it is not possible to measure certain physical quantities directly. For example, the neutron spectral flux density (neutron spectrum) is measured by means of protons, i.e. we obtain a proton spectrum. Knowing the detector response function and the proton spectrum, we may employ a sophisticated technique to calculate the original neutron spectrum. One of the possible techniques is Maximum Likelihood Estimation, which has been successfully used in our team for many years.

The aim of this thesis is to provide a new, user-friendly, platform independent implementation of the MLE-EM algorithm that would be able to process data from proportional detectors, which is currently not possible. The response functions for proportional detectors of different parameters should be calculated.

The software must be usable for data from different kinds of detectors, i.e. with different response matrices and different pre-processing functions. It also has to allow different normalization functions to be applicable to the result.

The text of the thesis should include an introduction to neutron spectrometry, description of the algorithm and underlying statistical model, description of the implementation (high level design, class diagram), and verification of the results (for simulated and real data). A brief user manual must be included.