Hectospec Run Preparation

Catalogue issues

• First and foremost be sure that Guide Stars and objects are in the same astrometric reference frame. The fibers cover 1.5" and an RMS error < 0.3" is extremely advisable. Fibers collide at distances of about 20".

• The magnitudes of guide stars must be brighter than R=15.5 according to hectospec documentation. Too faint means loss of guiding or goofing up the instrument rotation. Another recommendation is that these stars should not range by more than 1 magnitude. Guide stars can be selected from the Guide Star Catalogue. Otherwise the SDSS can be used, which has the advantage of good magnitudes and colors at the expense of smaller coverage on the sky. In either case, make sure to use either catalogue to find offsets between science objects and guide stars.
  One recommendation by D. Eisenstein is that the guide stars should not be very red, since these could be M dwarfs with large proper motions.

• According to documentation of Hectospec, the safe detection limit for absorption line spectra in 60 minutes is about R = 21.5.
  For 22.0 <= R <= 22.5 the success rate depends on color. Successful detection in 1 hour integration is possible for sources whose "I-band central surface brightness ~ 10 % night sky brightness". I could not find in the MMT documentation what is the typical sky brightness in I band at the MMT.

• Select Flux calibration stars. For fields with SDSS photometry this can be done by querying the SDSS Query/CasJobs site. This was the query used to select Spectro-photometry calibration stars. These are typically F stars whose fluxes can be approximately fitted using the multi-band photometry. The stars were selected to be in the range 15 < g <18. F stars will have colors 0.1 <= g-r <= 0.3. For the EGS run, stars selected for fiber placement were constrained to 0.1 <= g-r <= 0.5, -0.1 <= r-i <= 0.2 and 0.9 <= u-g <= 1.4, containing the range from A5 to G5.

• When setting priority to sources, be aware that the rank goes from 1.0 (high) to N > 1 (low). xfitfibs will only place fibers on rank = 2 objects after it has finished placing fibers on rank = 1.0000-1.9999. However the manual is not clear how fibers are allocated within the 1.0000-1.9999 rank interval.

• When preparing the catalogue ensure that enough spectro-photometric calibration stars are placed in each field. Write each one out the number of times corresponding to the configurations that will be used. Example - if the field will be covered by 16 configurations, write out each spec-ph star 16 times with rank = 1.

• The xfitfibs manual states that sky positions (given by the user, not those configured by the programme) should be given lowest rank possible. However, since these will be the last fibers to be placed, it is very likely that only a few will be used (it depends on the total number of objects in higher ranks). If the user's positions for sky fibers must be used, they cannot be the lowest priority objects.On the other hand, xfitfibs seems to do a good job in placing the sky fibers.

Running xfitfibs

1. First of all find out the dates of observation and times when the field transits. Since hectospec is operated on queue scheduling, any time while the field is below 1.5 airmasses is valid. However, note that since the MMT has an alt-az mounting, fields that go close to the zenith cannot be observed while transiting because of the very quick change in instrument rotation.

2. Create the field catalogue which will be used to determine the field coordinates. These must be in the "starbase" format where columns are separated by tabs, not blanks. See the example perl scripts for the field center and the object catalog . Make sure the field (*.fld) and catalogue (*.cat) have the same name. Notice that the test field name has an '.orig' appended. It must be copied as '.fld' so xfitfibs recognizes it. The 'fld' file will be overwritten by xfitfibs. If the field and catalogue do not follow the convention above (2 line header etc), xfitfibs will not run . The times that were found in item 1 are placed in the fld file under variable '$start'.

3. Using a catalogue with the guide stars only, use the default parameters under the parameters tab of xfitfibs (the radio boxes with red colors). These specify the parameters used by xfitfibs to classify the guide stars, interacting directly with the Guide Star Catalogue and DSS (at the CfA presumably).

4. Press the "Fit Guides" tab. If the radio box just before the RA column is red, this means that the field position may be tweaked, using the grid specified in the tab (default 2 arc minutes at 0.25 arc minutes intervals) and the final result written to the "fld" file. The grid parameters can be changed ad libitum. Often it is easier to tweak by hand, toggling the guide stars and guide annuli, placing the mouse on the ring representing the hectospec field of view and moving it. Before doing this, unset the red button next to the field RA column so that changes will not affect the "fld" file. Move the field around examining when stars fall inside the guide annuli. The boxes with r0, r1, r2 show the rotation angle for beginning, middle and end of exposures and should be green. If any box is yellow or red, the field may need to be rotated so all three boxes are green. To do this toggle on the guide annuli and place the mouse on the red circle at the ends.

5. Once candidate stars are selected, they can be inspected with ds9. They must also be classified by pressing the "Classify Candidate Guide Stars" button. This will retrieve and analyse DSS images, deciding if the candidates are worthy enough to iluminate the guide probes. Stars that are in legal positions have blue colors in the field plot, while stars that will be monitored by the guide probes have green colors. The last box in the field table will show how many guide stars are available. If it is red that is bad news - a new position has to be searched for. The bottom line is that there must be at least two probes which can access stars.

6. To make life easier configure guide stars for one field at a time (i.e., a separate file for each configuration. xfitfibs updates these files.

7. The ensemble of configured files can then be "glued" together:
   head -2 egs_field_01.fld > egs_final.fld ; grep '14:' egs_field_*.fld >> egs_final.fld

8. Edit the rank file (in this case junk.rnk) writing the minimum and maximum number of objects of a given rank should appear in each configuration. In order to have 6 spec-ph stars, write 6 in both the minfield and maxfield columns.

9. Now that the ensemble of field centers are ready run xfitfibs again. If the configurations were done correctly for the individual fields, fitting guide stars and classifying them will be quick and easy.

10. In the parameters tab set the minimum and maximum number of sky fibers and how they will be chosen.These numbers refer to the sky positions "calculated" (somehow) by xfitfibs, not those indicated in the catalogue file by the user.

11. Press the "Fit Fibers" tab, set the number of rank levels ("depth") set the rank button (it becomes red) and press "begin fit". This will fit all configurations and place sky fibers at the end. Depending on the number of ranks this is a long process. Unfortunately the present incarnation of xfitfibs does not tell you which rank it is fitting. I have found out that if there are too many rank levels (I tried 10) it crashes with "segmentation volation". This was solved by limiting the number of ranks to 6 or 7. Of course this need further checking. Once all ranks are exhausted, xfitfibs will place the sky fibers at sky positions it calculates.

12. xfitfibs outputs the final configurations in a '*.cfg' file. This will contain the fiber number, RA, DEC, plate X, plate Y position, catalogue number and rank. If the number is -1, this is a sky position chosen by xfitfibs. A zero means that the fiber is parked (unused); otherwise it corresponds to the entry number in the input catalogue. The sky Ra and Dec seem bogus.

13. Once finished press the "Send" tab, pickout the project number. If it is not there, let the hectospec team know so they can update the code. After pressing send several tests will be carried out. If there are no problems the files are shipped to the CfA.

Viewing configured objects