RoboNet microlensing event prioritisation pages - help and advice
Some hints and tips on the use of the prioritisation pages. Please send any other questions / comments / bug reports to csnodgra@eso.org.
General description of the system
The background codes are now in two parts:
- The data subscriber. The data subscriber populates the directories http://robonet.astro.livjm.ac.uk/~robonet/newcode/data/OGLE ../ROBONET ../MOA etc with the latest data files and fits, and produces the file events.txt which lists all events. This file has one row per event, and events with multiple identifiers are listed on one line, with spaces separating them. OGLE numbers first, where they exist, just because I wrote the code that way. At the end of the end of the data subscriber run it calls the next code, in case it isn't already running. This is
- The main fitting routine. The first thing it does is collate from each of the directories OGLE, ROBONET, MOA etc *all* of the known parameters for an event, and stores them in http://robonet.astro.livjm.ac.uk/~robonet/newcode/data/XB07XXX.param , and it then calculates the priority for updating the fit on each event. The priority is given by the product of the observing priority, the time since the last update, and the number of new data points since the last update. The priority for the dChi^2 plots is also calculated, (with the times and amount of data since the last dChi^2 map was done), with a normalisation that sets the dChi^2 priorities at 1/20 of the fast fit, if all else is equal. This means that new events (where the time since the last update is the current JD) get a very high priority, and then any event that is high observing priority or has a lot of new data will also be high on the list. The code loops through all events in this first part, and produces a list of tasks according to priority. It then goes through the list of tasks, and calls routines to do the fitting as required. It rechecks the priorities every 10 tasks to make sure that it reacts to new data. This then loops until it runs out of interesting things to do (ie there is no new data on any event and all plots are up to date).
Together, these two codes ensure that there is a .param file for each event which has the latest possible information from all sources and the most up to date fit from PLENS. These files are read by the web page generation scripts to give observing priorities.
The public side of the system consists of two cgi scripts:
- http://robonet.astro.livjm.ac.uk/~robonet/newcode/cgi-bin/event_pages.cgi. This produces lists of events sorted by X, and the individual pages for each event with the plots. This code now does both of these, depending on whether you give it an argument 'event=XB07XXX' or not (all arguments are passed to it by button pressing, but the event_pages.cgi?event=XXXX url may be useful for external links or non human readers). The main list of events is fully customisable. you can display and sort by *any* of the many parameters that are stored in the .param files. At the top of the page is are form inputs, where you can select as many or as few columns as you like from the first list, and select which one to sort by from the second list. Also, this page will store a cookie through your browser, so it will remember which columns you prefer to have when you come back to the page. The priorities and current predicted magnitude are calculated at run time for each event, based on the model you choose (again, option buttons at the top of the page): the PLENS fit, the Signalmen fit, or the fits from the OGLE or MOA web pages.
When you click on one of the event buttons, the script will generate a page for that event. It has at the top navigation buttons to previous/next events, and a link back to the full list, and then it has a table which will show the same parameters as you selected on the main page, but for the single event. The page then has links to the finder charts, full parameter file, OGLE/MOA pages, and various other plots. Below that are the light-curve and dChi^2 plots, as on the previous version, but next to each plot is a button that says 're-do light-curve' or 're-do dChi^2 map'. So you can refit and regenerate the plots for each event on demand, if you want to see the result of new data *right now* before autodownload gets to that event. Of course, do remember that these do take time to produce, and that the background codes are probably doing something they regard as more important if they haven't got to this event yet. Use this feature sparingly!
- http://robonet.astro.livjm.ac.uk/~robonet/newcode/cgi-bin/choose-telescope.cgi. This is the web-PLOP code proper, that produces a prioritised list for each telescope. When you call choose-telescope.cgi without arguments it will give you a blank form. You can then select your telescope as before. The changes here are there are extra input fields: The 'data' parameter can now be set to 'SIGNALMEN' or 'MOA' to get prioritised lists using those fits, in addition to the PLENS and OGLE ones. Or, you can select 'ALL' and the prioritisation code will try each type of fit for each event until it gets to one that works (it will try fits in the order that they are given on the page - currently PLENS, OGLE, MOA, SIGNALMEN).
The 'time' parameter can now be set to 'VISIBLE' which will calculate the priorities for when the bulge is visible (above 30 degrees elevation, Sun below -18 ) from the specified site.
The 'output' parameter can be either HTML or ASCII. HTML returns a human readble page, ASCII gives simply a list of targets in plain text, with columns "EVENT RA DEC N_EXP EXP_TIME PREDICTED_MAG"
Below that is a table for you to select the output columns you would like to see in the html page, as with the event_pages.cgi code.
Hints and tips for using the web pages
- To select columns for output: you should be able to make multiple selections from the list by holding ctrl (or cmd on a mac) and clicking on the ones you want. The page will remember the last selection you used (if you have cookies enabled). Also, you can select some preset set ups by pressing the preset button(s). The question mark links next to these buttons show you what the presets include.
- Choosing which fit type to prioritise by: ALL means try each method in order until a fit works. i.e. if there isn't a PLENS fit, try the OGLE data, then the MOA, then Signalmen. In practice, this will nearly always be the PLENS fit, but there are some events where PLENS has failed to find a solution. Include the column 'CALC_P_USING' to see which type of fit was used for each event.
- The status of events. The STATUS parameter will be set to rejected if PLENS fails on an event, to stop PLENS spending all its time retrying it. The status can be changed back manually once there is new data that will give PLENS a solution. The status does not affect the prioritisation.
- What does the * by the event name mean? It means that the background codes are currently re-fitting that event. The parameters are likely to be updated (and you shouldn't try to refit it manually!).
- My telescope is missing parameters in choose-telescope.cgi. Let me know, changing parameters or adding telescopes is straight forward.
C.Snodgrass, ESO, 27/06/07