Histogram Presenter

[ Up ]

The Histogram Presenter should offer as many features as possible, as it will be one of the main tools to understand the status of the experiment, and to guarantee the best quality of the recorded data. It should also provide a stable configuration, with well defined display pages, with list and location of histograms on the page. This configuration should be ‘frozen’ for the shift crew, but it should also be editable by authorized users. Furthermore, it should be possible to interactively define “test pages” with an intuitive user interface, to test and validate new configurations. These test pages could be saved and retrieved from personal storage.

The predefined page structure should differentiate between standard Shift Crew Pages, with a limited number of pages that should be looked at regularly, and Expert Pages, with as many pages as needed, grouped by detector or project. (Expert Pages could also be used to store “test pages”.) Shift Crew Pages should be protected from modification, and their layout defined and maintained by the operation group.

The Presenter will be based on  ROOT, and will provide many interactive features necessary to manipulate histograms. Part of the monitoring information will be PVSS trend plots. It is desirable that they can also be displayed in the same presenter. This means that a page could contain both histograms AND trend plots, with the same look and feel, if technically possible.

Two modes of operation will be available: an “online mode” and a “history mode” (histograms and time charts corresponding to a specified time interval). Online mode is used during data taking by the shift crew. In history mode, all files saved during the specified time range are retrieved and their content added up before display. History mode is typically used every morning by sub-detector experts to monitor the performance of the apparatus during the previous 24 hours, or to investigate problems that occurred in their absence.

It will also be possible to display reference plots of all histograms in the “current page”, with a single mouse stroke, overlaying them with different colour. The question of relative normalisation is not simple. Normalisation to the same highest bin, to the same integral, to the same number of monitored events should all be possible. Their selection depends on the physical meaning of a histogram, and may be even more complex. (A noisy channel will create a spike in the hit distribution; this channel should be ignored in the normalization if one wants to compare the shape while ignoring this noisy channel.)

Reference files should be structured in such a way that it is easy and fast to find the reference of any plot in the system. One should foresee several reference files, corresponding to different conditions (luminosity, trigger configuration) and a mechanism to select which one to use.