Development of Monte Carlo Algorithms for Eigenvalue – Calculations

As a first stage in the development of new algorithms to reduce the variance in eigenvalue calculations with the source-iteration approach, a "superhistory" approach has been introduced into MCNP5. An adjustment of the v-value to ensure a supply of children at each normalization in the case that keff differs appreciably from unity has not yet been inserted. Following this, the development will be articulated in two steps - firstly a search for the trade-off between an optimum population within each cycle (or superhistory) and at the next normalization point (i.e. end of superhistory); secondly a search for the trade-off between cycle size and number of normalization points.