A guide to setting up Complex configurations
Recent years have seen a rise in the number of programs that require complex configuration strategies to maximize the yield of an AAOmega observing run. This guide is intended to outline some of the strategies that have been adopted in the past. It is not exhaustive, and we welcome comments and suggestions.
Large scale survey programs have their own special requirements with regard to target allocation priorities. The GAMA survey project (Driver et al. 2009 A&G 50 12) implemented a very detailed, multi-year, survey strategy which is documented in Robotham et al. 2010 PASA 27 76. While this strategy is likely more complex than most program will need, many of the issues of concern are discussed.
This guide is not intended to replace the Configure manual
AAOmega has only ~350-400 fibres once sky fibres and the current status of fibres is taken into account. Therefore the ideal AAOmega program has of the order of 400 targets per 2degree diameter field and a uniform target distribution with no closely spaced targets. Unfortunately, nature does not work like that. This results in extended target lists, with multiple priority levels and strongly clustered source distributions. The Configure software is necessarily generic, requiring the observer to carefully define input target lists.
The main points to consider are:
Number of targets. In most cases, and particularly so for the default Simulated Annealing Configure algorithm, simply passing a list of >1,500 targets to Configure will not produce optimal results. With around 350-400 fibres available per configuration, target lists should normally be re-sampled to include only a small excess of targets to fibres in the input file. 800-1000 targets works well for relatively uniform fields, lower numbers are needed for more compact fields.
Repeat observations. In surveys where the same (or overlapping) fields will be targeted multiple times, it is often advantageous to reallocate targets between observations. This can increase the total target yield by rejecting targets that are confirm to have unwanted spectral types, or replacing objects for which the spectral quality obtained is already sufficient.
Locking a sub-set of allocations. You may wish to force the repeat observation of a sub-set of high priority targets, at the expense of a large number of lower ones, but also include these lower priority targets in a new configuration where possible. There is a facility to lock fibres in place between configurations.
A simple way to cover multiple targets in a field, using repeat observations, is outlined in section 5.4 of the configure manual "Multiple configurations to cover a target list". The process uses the option to save the unallocated targets from a configuration to a file, using the menu option File->list... in Configure. While this technique certainly works, and is discussed below, it is limited in that it can be rather tedious to undertake, and also cannot be done efficiently in advance of a run, since the 2dF astrometric solution will change before the run and so many allocations will not be valid during the obseravtions. A better solution is outlined in Example 2 below.
Example 1 - Simple multiple configurations
Starting from a large master catalogue:
Allocate the field using configure
Save the .sds file.
Also save a .lis file (File->List...) which contains the unallocated objects (this is an option in the pop-up that will appear then you select File->List...). The .lis file format for the unallocated objects is the same as that for an input .fld file. If you select one of the other save options, the formatting will be a little different.
Load this .lis file into configure and rerun the configuration, saving a new .lis file of the outstanding unallocated objects each time.
Example 2 - Configuration using a master catalogue, and current status list
For the reasons below, Example 1 above is not the preferred approach for most programs:
For the first run in any AAOmega observing block, the astrometric files will not be available prior to the run, and hence the configurations would be invalid if prepared in advance. This is typically only a small effect, but is compounded with each new iteration.
The fibre availability of the field plates will change with time during a run, due to the slow rate of fibre attrition.
No accounting is made for data quality once observations have begun.
There is little flexibility between the interchange of the two 2dF field plates during observations.
The solution is to prepare a master catalogue, with target observation priorities, for your full observing region and to create a processing script to draws sources from this catalogue based on external constraints. For example:
Choose the first pointing centre for observation.
Make a .fld file, select 500-800 high priority targets from the master catalogue. The UNIX grep and awk commands are ideal here, or a simple Perl script may be the best way to achieve this goal.
If there are a small number of high priority targets, pad out the .fld file with lower priority objects. Care should be taken if a large number of low priority targets are introduced. The user should examine the configuration at step 5 below, to ensure a sensible fibre distribution is being used.
Insert ~30-50 guide stars and ~50-100 blank sky positions.
Configure and inspect the configuration
Observe the field.
Examine the spectra (determine redshifts, measure radial velocities, classify objects etc.)
Update the master catalogue to reflect these observations. A safer approach is often to create a second catalogue of new target priorities. The processing script would then draw targets for subsequent configurations from the master file, but the priorities of classified targets are adjusted based on the information you have entered into the new target priorities catalogue. For example: Set the priority of objects with satisfactory spectra to the lowest value (Pri=1); Remove objects of the wrong spectral type; Flag objects with promising spectra, but which will need higher signal-to-noise. Note, you clearly cannot make the new target priorities until you have some data, and since one needs to have a number of .fld files ready in advance of each nights observing, this process is most efficient if any given region can be broken down into a number of independent (non-overlapping) pointing centres each night.
Note, that there is no underlying reason why the field centre need be identical to that previously used.
Repeat steps 5-9 until the required target completeness is achieved.
Sarah Brough (email@example.com)