advances
Returns the pre-run sub-stream advance for each point of inputs, in request order, and moves the tape forward. See the class documentation for the independent vs. CRN semantics.
Return
the advance to apply (as numberOfStreamAdvancesPriorToRunning) for each point
Parameters
the points of the request, in order
true for common random numbers (all points share the current block), false for independent, non-overlapping stream blocks
Returns the pre-run sub-stream advance for each of a sequence of runs described only by how many replications each consumes, in order, and moves the tape forward. See the class documentation for the independent vs. CRN semantics.
This is the whole of the arithmetic: positioning a run on the tape depends on nothing about a point except its replication count. The ModelInputs overload above reads exactly that from each input and delegates here, so a harness outside ksl.simopt that positions replications absolutely on a sub-stream tape can reuse this policy rather than manufacture ModelInputs values it has no other use for, or hand-roll a weaker version of the same rule.
Overloaded on the Kotlin side only: a List<ModelInputs> and a List<Int> erase to the same JVM signature, so this carries a distinct @JvmName and Java callers reach it as advancesForReplicationCounts.
Return
the advance to apply (as numberOfStreamAdvancesPriorToRunning) for each run
Parameters
the replications each run will consume, in order; each must be non-negative
true for common random numbers (every run shares the current block), false for independent, non-overlapping blocks