setTime [-intervals <timeIntervals> | -ticks <ticksPerInterval> | -history <historyLength> | -dtfixed]
With no arguments this returns the values of the active network's timeIntervals, ticksPerInterval, and historyLength parameters.
Otherwise, this changes one or more of these values and returns the new values. All values are integers. If ticksPerInterval is set to a value other than 1, the network will not be set to CONTINUOUS type as it would be in old versions of Lens.
historyLength determines the size of the history arrays in each unit and thus the values displayed in the Unit Viewer. A full history is needed for continuous networks but not for feed-forward or simple-recurrent nets or deterministic Boltzmann machines. If not specified, historyLength will be set to the maximum useful value, which is the same as the network's maxTicks, or timeIntervals * ticksPerInterval, when this command is invoked. A historyLength of zero will prevent the unit history arrays from being allocated or used.
If ticksPerInterval is set, the default behavior is for the
network's dt to be recalculated to its inverse. However, the
-dtfixed option will prevent dt from being changed.
To get the time parameters for the active network:
lens> setTime 3 5 16
To change the time parameters for the active network:
lens> setTime -i 4 4 5 21
To save space when training a Boltzmann machine, you might set the history length to 0:
lens> setTime -h 0 4 5 0