timecov(covariate, minimum = 1, maximum = length(covariate),
transform = function(t) 1 + (0 * t) + (0 * t^2),
onlytime = FALSE)1-1-1-1-1. A linearly ionlytime = TRUE is set, the function returns a list of matrices where all entries within a matrix are identical and reflect the transformation of time. This is useful when a function of time per se should be included in the model as a covatimecov function takes a list of matrices or networks (a varying dyadic covariate) and creates an interaction term with time. For example, if the rules of network formation are expected to change after some time steps (say, after the third out of six time steps), one can model the effect of the covariate for the first three time points and for the remaining three time points using separate model terms. To achieve this, the covariate matrix at each time step is multiplied by zeros or ones, depending on whether the time step should be incorporated in that model term. In this situation, one would create two dyadic covariates, one where the first three time periods are present and the remaining three are set to zero, and the other one where the first three time periods are set to zero and the remaining ones are present.Another usage scenario is that time per se may have a polynomial effect (of any shape) on the probability of forming a tie. In this case, the timecov function can be used to create a covariate list of matrices where the entries of the matrix correspond to polynomial functions of time. For example, if tie formation becomes increasingly likely (as expressed by a linear relationship), a list of matrices with linearly increasing entries over time can be created. Such time effects can also be interacted with other covariates. For example, the further time progresses, the more (or less) the property captured by the covariate becomes important for tie formation.