Merge pull request #26 from gonzaponte/sophronia

Add Sophronia documentation

docs/source/IC.rst

@@ -137,6 +137,7 @@ MAIN PRODUCTION:
 
      Irene <irene>
      Penthesilea <penthesilea>
+     Sophronia <sophronia>
      Dorothea <dorothea>
      Esmeralda <esmeralda>
      Beersheba <beersheba>

docs/source/sophronia.rst

@@ -0,0 +1,251 @@
+Sophronia
+=========
+
+*From ancient Greek: sensible, prudent.*
+
+
+This city processes each S2 signal previously selected as pmaps in
+irene assuming a unique S1 within an event to produce a set of
+reconstructed energy depositions (**Hits**). Hits consist of three
+dimensional coordinates with an associated energy (PMT signal) and
+charge (SiPM signal). The city contains a peak/event filter, which can
+be configured to find events with a certain number of S1/S2 signals
+that satisfy certain properties. Currently, the city is designed to
+accept only 1 S1 signal. If the configuration allows for more than one
+S1, the first one will be used to estimate z. Furthermore, hits can be
+optionally corrected for geometrical and lifetime effects. Besides
+hits, the city also stores the global (x, y) position of each S2
+signal.  The tasks performed are:
+    - Classify peaks according to the filter.
+    - Filter out events that do not satisfy the selector conditions.
+    - Resample S2 signals.
+    - Compute a set of hits for each slice in the resampled S2 signal.
+    - If there are more than one hit per slice, share the energy
+      according to the charge recorded in the tracking plane.
+    - If correction maps are provided, apply corrections to hits.
+
+.. _Sophronia input:
+
+Input
+-----
+
+ * ``/Run/events``: Table containing the event number and timestamps
+ * ``/Run/runInfo``: Table containing the run number for each event
+ * ``/PMAPS/S1``: Table containing S1 pmt-summed peak information
+ * ``/PMAPS/S2``: Table containing the S2 pmt-summed peak information
+ * ``/PMAPS/S2Si``: Table containing the S2 sipm peak information
+
+.. _Sophronia output:
+
+Output
+------
+
+  * ``/DST/Events``: summary of the S1 and S2 information of the reconstructed events. See :doc:`dorothea` for more details
+  * ``/RECO/Events``: list of hit data for each event
+  * ``/Filters/s12_selector``: flag for whether an event passed the S1 and S2 selections
+  * ``/Filters/valid_hit``: flag for whether an event passed the empty pmap filter
+
+.. _Sophronia config:
+
+Config
+------
+
+Besides the :ref:`Common arguments to every city`, *Sophronia* has the following arguments:
+
+.. list-table::
+   :widths: 50 40 120
+   :header-rows: 1
+
+   * - **Parameter**
+     - **Type**
+     - **Description**
+
+   * - ``drift_v``
+     - ``float``
+     - Drift velocity. Used to calculate the z position from the drift time
+
+   * - ``s1_params``
+     - ``dict``
+     - Selection criteria for S1 signals (see `Peak filtering`_)
+
+   * - ``s2_params``
+     - ``dict``
+     - Selection criteria for S2 signals (see `Peak filtering`_)
+
+   * - ``global_reco_algo``
+     - ``XYReco``
+     - Reconstruction algorithm applied to entire peaks (kDST)
+
+   * - ``global_reco_params``
+     - ``dict``
+     - Configuration parameters for the global recontruction algorithm
+
+   * - ``rebin``
+     - ``int``
+     - Level of resampling or grouping of consecutive samples for reconstruction. Must be :math:`\geq` 1
+
+   * - ``rebin_method``
+     - ``RebinMethod``
+     - Resampling method (usually stride)
+
+   * - ``q_thr``
+     - ``float``
+     - Charge threshold to be applied to the SiPMs
+
+   * - ``sipm_charge_type``
+     - ``SiPMCharge``
+     - Type of SiPM charge to use (usually raw)
+
+   * - ``same_peak``
+     - ``bool``
+     - Controls whether invalid hits can only be merged with other hits from the same peak
+
+   * - ``corrections_file``
+     - ``str``
+     - Path to the file holding the correction maps
+
+   * - ``apply_temp``
+     - ``bool``
+     - Whether to apply temporal corrections (in general, False in MC, True in data)
+
+.. _Sophronia workflow:
+
+Workflow
+--------
+
+Sophronia performs the following data transformations:
+
+ * :ref:`Peak filtering <Peak filtering>`
+ * :ref:`Build event summary <Build event summary>`
+ * :ref:`Resampling and hit creation <Resampling and hit creation>`
+ * :ref:`Hit merging <Hit merging>`
+ * :ref:`Hit correction <Hit correction>`
+
+
+.. _Peak filtering:
+
+Peak filtering
+::::::::::::::
+
+:doc:`irene` finds peaks in the waveform with general characteristics,
+but does not impose strong requirements on them. *Sophronia*, however,
+can be more specific and selects S1 and S2 signals based on the
+following peak properties: width, height, and integral. These
+quantities are calculated based on a (low) threshold applied to the
+PMT waveform of the peak. For S2 signals, the number of SiPMs with
+some signal is also taken into account [1]_. The variables that
+control this filtering are provided in the parameters
+``{s1,s2}_params`` and are defined as follows:
+
+- ``{s1,s2}_w{min,max}``: minimum/maximum width of S1/S2 peaks
+- ``{s1,s2}_h{min,max}``: minimum/maximum height of S1/S2 peaks in a 1-:math:`\mu`\ s sample
+- ``{s1,s2}_e{min,max}``: minimum/maximum integral of S1/S2 peaks
+- ``{s1,s2}_ethr``: threshold applied to the PMT-summed waveform to compute the quantities above
+- ``s2_nsipm{min,max}``: minimum/maximum number of SiPMs with signal in an S2 peak
+- ``{s1,s2}_n{min,max}``: minimum/maximum number of S1/S2 peaks that satisfy *all* of the criteria above for given event
+
+.. [1] Bear in mind that :doc:`irene` requires a minimum amount of
+       charge per peak and per slice for the SiPM to be considered.
+
+
+.. _Build event summary:
+
+Build event summary
+:::::::::::::::::::
+
+This part of the processing is equivalent to :doc:`dorothea` using
+``global_reco_algo`` and ``global_reco_params``. See that city's
+documentation for more details.
+
+.. _Resampling and hit creation:
+
+Resampling and hit creation
+:::::::::::::::::::::::::::
+
+In numerous occasions, the energy deposition in a 1-:math:`\mu`\ s
+sample is not enough to produce a significant signal in the SiPMs,
+which results in poor reconstruction. It is therefore useful to be
+able to resample the waveforms and increase the sampling period (lower
+sampling rate). The variable that controls this resampling is
+``rebin`` and it specifies how many consecutive 1-:math:`\mu`\ s
+slices are added up together. A value of ``rebin = 1`` means no
+resampling is performed. Only integer values greater than or equal to
+1 are accepted.
+
+Unlike *Penthesilea*, *Sophronia* considers the response of a SiPM in
+a single (resampled) slice a *Hit*. Hits are defined as the
+aggregation of the absolute position of the SiPM, the time difference
+between the slice and the S1 peak time [2]_, the amplitude of the SiPM
+waveform (charge) and the energy corresponding to said charge. The
+corresponding energy is defined as
+
+.. math::
+
+   E_i = \frac{Q_i}{\sum_{k=0}^{N} Q_{k}} E_{slice}
+
+where :math:`Q_i` and :math:`E_i` are the charge and corresponding
+energy of SiPM :math:`i`, respectivly; :math:`N` is the number of
+SiPMs with signal in the slice and :math:`E_{slice}` is the energy of
+the slice, given by the amplitude of the PMT-summed waveform in the
+slice.
+
+.. [2] This information is used to obtain the z position of the hit by
+       dividing the time difference by the drift velocity
+       (``drift_v``).
+
+
+.. _Hit merging:
+
+Hit merging
+:::::::::::
+
+In some occasions, a time slice might have no SiPMs recorded. This
+means that all SiPM waveforms have been disregarded in :doc:`irene` or
+that no SiPM has a charge above ``q_thr`` in such slice. However,
+because the slice is considered part of a peak, there is some energy
+associated to it, which should not be discarded. In these cases,
+*Sophronia* generates a fake hit (a.k.a. *NN-hits*) with no charge and
+no position associated [3]_.
+
+After all hits are generated, a second step is performed, in which the
+fake hits are merged with existing valid hits. This operation searches
+for the hits that are closest (in z) to the fake one. Each fake hit's
+energy is shared among its neighbours and added to the neighbouring
+slice energy. The energy is not share homogeneously, but
+proportionally to the charge of each hit. The parameter ``same_peak``
+controls whether only slices from the same peak are taken into account
+or if other peaks might also be taken into consideration.
+
+.. [3] Actually, the algorithm associates an unphysical charge (NN =
+       -999999) and a position :math:`x = y = 0`.
+
+.. _Hit correction:
+
+Hit correction
+::::::::::::::
+
+If a correction map is provided (``corrections_file``), the
+geometrical and lifetime corrections are applied. These corrections
+are composed of three factors:
+
+- geometrical: accounts for different light collection efficiencies in
+  x,y
+- lifetime: accounts for the loss of electrons (and therefore signal)
+  due to electron attachment during the drift
+- temporal: changes in the previous two corrections over time. This is
+  applied only in data, as MC has no temporal dependence.
+
+The geometrical correction is a 2d-function: given the hit position in
+x,y, we obtain a factor that normalizes the response to the center of
+the chamber. It may also scale the response to different units
+(pes-to-keV, for instance).
+
+The lifetime correction is a 3d-function: given the hit position in
+x,y,z we obtain the correction factor that normalizes the response to
+that of z=0.
+
+The temporal correction is a 1d-function: given the time since the
+start of the run, we estimate variations in the geometrical and
+lifetime corrections that normalize the response to that of the
+beginning of the run. This correction is only applied if
+``apply_temp`` is set to ``True``.