A SIP is a Stochastic Information Packet that is a standard library of data (see ProbabilityManagement.org for more details on this standard). DiscoverSim treats this as a Custom Discrete distribution with unweighted data and is sampled sequentially (unrandomized).
A SIP cannot be included in Specified Input Correlations, nor can it be truncated. The number of rows (nrows) in the SIP data should match the number of replicates (nreps), or nreps/nrows should be evenly divisible.