[mmaimcal] PS:

[Mark Holdaway]

I forgot to add:

opacity varies fairly smoothly, a 1-2 hour delta_t in the dynamic
scheduling is not too far from optimal integrated sensitivity (ie,
infinitesimal delta_t).  A correlary of the short distance variations
seen in the phase stability is that phase stability can be highly time
variable; observational sensitivity would benefit from a shorter time
scale on phase stability grounds;  30 minutes?  At some point, the
overhead of switching programs begins to be appreciable, cutting into the
sensitivity.

Again: if this is all automated, there is no major problem keeping track
of these small pieces.

My picture had always been: software indicates if a change in program
should be made (ie, conditions in this 30 minute chunk have either
improved or deteriorated), and suggests the new program.  Barry Clark
looks up from his newspaper and hits "return" or selects his own program.

	-Mark