Subject: 4Q1995 Schedule Comments

From: Ed Turner

Submitted: Thu, 14 Sep 95 17:30:39 EDT

Message number: 16 (previous: 15, next: 17 up: Index)

Dear Schedulers and anyone else interested,

I have heard from NMSU, PU and UW about 4Q scheduling problems.  May I take
silence for happiness, or at least resignation, on your parts UC and WSU?

UW seems to have drawn a major portion of the scheduling problems for 4Q,
and I have written a detailed response to the difficulties they reported.
The more general parts of this response may be of interest to all, so I
attach them and one additional commnet below.  I would urge all institutional
schedulers to read this material and to share it with any of their users
who might be interested.

Let me emphasize that I do not consider our current system of scheduling
the 3.5m to be very satisfactory.  I am anxious to receive any comments
or suggestions and willing to consider experimenting with alternative
methods, including radically different ones, different scheduling
strategies, different mechanisms (from computers to committees), whatever.

Cheers,

Ed Turner

----------------------------------------------------------------------------

>1) I will do what I can to alleviate the problems when a revised schedule
>   is issued.  This will be a few weeks; it seems to work better to accumulate
>   a bunch of problems and fix them all at once than to try to make a 
>   of incremental changes.  I hope to be able to make improvements for all of
>   the problems but cannot promise anything.  Since the quarter is fully
>   schedules, changing the time for one of your programs means forcing a
>   change on some other program.
>2) You can help your own cause by finding a fix for your problem on your   
>   own steam and then letting me know.  In other words, if you can find
>   someone (at UW or elsewhere) who has time you would prefer and is  
>   willing to swap, it can almost certainly be arranged.  Formally, it
>   requires the agreement of the affected PI's, of the relevant institutional
>   schedulers, and mine...but no one is likely to object if the PIs's agree,
>   of course.  This may not be so easy to do, I should warn you, as lots of
>   people requested/needed very specific times and so may not be willing to
>   switch.  However, it is a lot easier for you to do it than for me to do
>   it for a lot of different programs.
>3) The scheduling exercise is very complicated and (for me at least) difficult.    I am sure that I am making mistakes and that I end up far from an
>   optimum schedule.  I think it is becoming more difficult in that users
>   seem to be submitting more and more constrained requests.  We will probably
>   eventually have to find a better way to schedule the telescope, but that
>   is another story.
>4) Most (but not all) of the problems you reported arise from a combination   
>   of relatively low priority ("scientific rating" in George's words) 
>   assigned by the UW TAC plus some sort of restrictive scheduling requests.
>   Thus, when your program came to the front of the queue for scheduling, 
>   your constraints may have excluded all remaining unscheduled times.  In
>   such cases, I tried to use my judgement as to what might still be useful
>   in some way or to get as close as possible.  To be more specific about
>   priorities, UW05 and UW06 were the two lowest priority dark/grey time
>   UW proposals to receive any time.  UW11-13 were 2-4 among the 5 UW bright
>   time programs that received any time; they were also 8-10 out of the  
>   total of 11 UW programs receiving time.  (Because UW did not provide a
>   single prioritized list, as requested, but only separate dark, grey and
>   bright lists, I arbitrarily concatenated the 3 in that order; this may
>   have resulted in unduly low priority for UW bright time programs.) 
>5) These requests and a large fraction of all 4Q requests asked for minimum
>   blocks of time, typically of half a night or a bit more (this is about
>   how long you can follow a typical well placed field at low air mass on
>   a given night).  This is perfectly understandable but makes for great
>   scheduling difficulties since putting one such block, or perhaps one
>   large block and a small monitoring observation, in a given night typically
>   leaves only one medium sized block or two smaller pieces to be handed
>   out to other programs.  Thus, as one proceeds through the scheduling
>   queue (in priority order), all available large blocks which access a
>   given part of the sky get used up relatively early leaving only smaller
>   blocks for later programs.  When a program requests a larger block than
>   any still available, I typically still give it time on the requested
>   part of the sky in hopes that something can still be done with it.
6) Certain parts of the sky were also particularly difficult for lower
   priority programs because they were heavily oversubscribed.  As
   Telescope Operators at any observatory will tell you, astronomers spend
   a lot of time chasing objects as they set in the west; the same seems
   to apply to observing quarters.  The first halves of early October
   nights were the most requested hours of the 4Q (chasing 3Q targets),
   just as the first halves of early July nights were the most in demand
   in the 3Q (chasing 2Q targets).  I suspect that there is some deep fact
   of human nature lurking in this pattern.  :-)
APO APO APO APO APO  Apache Point Observatory 3.5m  APO APO APO
APO
APO  This is message 16 in the apo35-general archive. You can find
APO  the archive in /u/strauss/apo/mailer/apo35-general on astro.princeton.edu
APO  To join/leave the list, send mail to apo35-request@astro.princeton.edu
APO  To post a message, mail it to apo35-general@astro.princeton.edu
APO
APO APO APO APO APO APO APO APO APO APO APO APO APO APO APO APO