Constructing the red-black graph for a finite task system.

Consider a finite task sytem S.  We assume that S has an optimizer O.
Our example is the ice-cream problem.  T = {v,c}, and the NFA is shown
in Figure 1.  The optimizer is shown as the DFA in Figure 2.  Note that
the states of optimizer are the offset functions.

Any NFA is augmented into a red-black graph in two ways.  Figure 3 shows
the augmentation of the optimizer using dummy red nodes and dummy red
edges, where the original nodes and edges of the optimizer are black.
Figure 4 shows the augmentation of the original task system using dummy
black nodes and dummy black edges, where the original nodes and edges of
the task system are red.

As the game is played, a pebble moves in each graph.  Figure 1 is the
adversary's private graph.  Figure 2 is the online observer's graph,
ignoring the algorithm.  The online observer does not know where the
adversary's pebble is, but does know the tasks.  Figure 3 is the online
observer's graph, taking into account the algorithm, but where the
online observer has no knowledge of the algorithm's movements.  Figure
4 is the algorithm's private graph, if the algorithm has no memory except
its own state and the current request.

Figure 5 is obtained by taking a product of Figure 3 and Figure 4.  This
is the algorithm's private graph, if the algorithms keeps track of the
position of the pebble in the optimizer.  We call this the red-black
graph for the task system.
