Motivation

Development of this application was stimulated by the discussion with Marcin Sikorski (meeting on 2012-08-30), doing xcs experiments.
Users need in real-time algorithm for calculation of image vs time auto-correlation function

g2(tau) = <I(t)*I(t+tau)> / (<I(t)> * <I(t+tau)>),

where I(t) is an image intensity at time t, and tau is a delay between two measurements.
Typical experimental condition can be described as follows:

• Run duration is about one hour at frequency up to 120 Hz that gives up to 10^5-10^6 images.
• Currently typical imaging devise is a Princeton camera with 1300x1340 pixels.
• Need to calculate g2(tau) for each pixel, averaged over all possible image times t with time difference tau between images.
• A set of tau should have about 30-100 points in log scale uniformly covering the run duration.
• Use for example xcsi0112-r0015: 500 images with 8 sec delay between images.
  Desired time for evaluation of the auto-correlation function should be comparable with run duration <1 hour. Currently this algorithm takes a few hours that can not be used for fast feedback in real time experiment.

In first approximation this problem was solved, as it is described in the Command Line Interface For Time Correlation Analysis. However, in order to be useful, the command line algorithm needs to be integrated in the global analysis system, which de bene esse called as Integrated Analysis Environment for Time Correlation Experiments and is discussed below.

Problems and tentative solutions

Based on 2012-10-03 meeting:

In order to be useful this application should do correct math, accounts for image mask, discards bad events (noisy and "bright" pixels), apply normalization etc., and have a convenient GUI. Below is a list of requirements, marked as , with suggested solutions, marked as if exists or as if needs to be implemented.

Pedestals

"dark" run number should be provided by user and the imaging camera pedestals should be evaluated and applied for all runs until the "dark" run number has not changed.
For pedestals evaluation: use available ImgAlgos::ImgAverage psana module for "dark" run, which produces file with averaged over events pedestals and the file with their rms values.
For pedestals subtraction: use ImgAlgos::ImgCalib psana module in the same job which evaluates correlators; the pedestals will be subtracted and corrected image will be retained in the event and used for correlator calculations.

Low level threshold

Image pixel intensity physically can't be negative. Low amplitude noise should be suppressed by the threshold. The threshold amplitude should be provided by user (along with substituting amplitude).
Is implemented in the ImgAlgos::ImgCalib psana module, right after pedestal subtractuion.

Image filtering

Usually users use different type of intensity monitor signals in order to retain/discard image for/from further processing. Discarded images should not contribute into the correlation analysis. The spectra of intensity monitors should be available for browsing. User should be able to select the intensity monitor(s) from the list and set low and high thresholds.
The filtering module may be implemented in psana. Based on selected intensity monitor(s) and thresholds it will decide to retain or discard event and accumulate spectral histograms. The histograms will be saved in file at the end of run.
Control GUI should be able to browse the intensity monitor histograms and set the thresholds.

Selection of intensity monitors

It would be nice to have an algorithm like in XTC explorer
Possible options:

• run application as a plug-in for XTC Explorer,
• pyana module performing similar to XTC Explorer algorithm,
• stand-alone C++ module reading XTC datagrams,
• hardwired list of intensity monitors.
• psana module, which needs to be developed

Dynamic mask

Imaging camera may have permanently hot pixels or some pixels may be saturated during the run. User need to set a thresholds on hot pixels and high intensity.
If the pixel amplitude crosses the high intensity threshold at least once during the run, then this pixel should be excluded from further analysis. The same is valid for hot pixels, which shows above threshold intensity in large fraction of events.
This is implemented in the psana module ImgAlgos::ImgMaskEvaluation, which works before event selection algorithm. Two files of image size may be produced 1) for saturated and 2) for hot pixels.

Static mask

The beam-stopper region and some areas with fringes should be masked. It could be useful to have a graphical editor for mask.
See section for GUI.

Graphical editor for selected regions

Sometimes it is useful to select good region of the image. It could be convenient to use a graphical editor, as for mask.
See section for GUI.

Center of the image

User should have an option to set a center of the rings for histograms.
See section for GUI.

Correct normalization of g2

Evaluation of g2 for image regions is not that simple as presented by the formula for a single pixel:

g2(tau) = <I(t)*I(t+tau)> / (<I(t)> * <I(t+tau)>),

In order to get physically meaningful results for g2, the correlators <I(t)> and <I(t+tau)> should be averaged in the fine rings around center with number of bins N2, which is order of 100, with dR down to 1-2 pixels.
Then the <I(t)*I(t+tau)> correlator should be averaged over bold rings intended for G2 evaluation. The number of these rings N1 should be order of 10.
The N2 and N1 should be defined by user.
It might be useful to define the histogram region by the sector in the user-defined angular range.
In order to have required normalization of correlators, it is not enough to save the g2 value only. So, the format of the resulting file has changed. Now for each value of tau the output file contains the <I(t)>, <I(t+tau)>, and <I(t)*I(t+tau)>, each for entire image written in binary for float format. Not all masks, selection regions, normalization etc. are available during correlators calculation, so correlators are evaluated for all pixels. Which pixels should be included in the G2 for each region can be decided at the final stage of processing. This approach allows to perform the most time consuming procedure - the correlators calculation once and do the analysis after that.

See section for GUI.

GUI

In order to get an easy interface to all sub-processes, it seems to be useful to have a GUI with configuration of everything through the GUI.
Well, presumably users will want different specific features in their analyses which can not be foreseen in implementation of GUI. It is pretty unlikely that everything in analysis can be done clicking on buttons in GUI. Then, it could be nice if user understand what he is doing step by step and have a monitoring at the end of each stage. We are doing science, not a standard pre-defined things... Most generic way to process data is to have a separate procedures with command line interface.
Anyway, the browser/presenter of data stored in the files after pre-processing could be provided for a set of common plots.
All features listed in previous sections, such as static and dynamic mask, restriction of the region(s) of interest, selection of the image center, the binning scheme etc., can be done in the browser at the final stage of the analysis.
Generic GUI as a shell for entire analysis can be implemented as an interface to the command line procedures:

• Each command line procedure may have a dedicated GUI for procedure configuration and monitoring.
• All procedures may be listed in the main GUI with a status sign.
• The main issue of this approach is a cross-configuration between separate procedures. This can be achieved if all procedures will have a common list of configuration parameters.

GUI Implementation

Updated on 2012-12-02

Three versions of GUI were implemented for this application. It was decided to use the GUIMainTB
layout, where:

• All input windows are integrated in one with tool bar
• Logger and File browser windows are isolated
• Plots are isolated

Code location

Everything resides in CorAna package, which is included in LCLS offline releases beginning from ana-0.7.16.

Modules

Updated on 2012-12-10
Notations:
- is done (at least it is assumed for now...)
- needs more work
- test or deprecated modules
All module names beginning with letters GUI implements different Graphical User Interfaces.

Status of the project

Updated on 2013-02-07

PSANA modules

For dark run pre-processing

Scaner
Runs in batch for the dark file to get preliminary information.
ImgAlgos.ImgTimeStampList

• counts number of events in the file
• makes file with time stamps
• evaluate time intervals between frames (for dark run)

Pedestals
Runs in batch for the dark file to get averaged pedestals.
ImgAlgos.PrincetonImageProducer gets image from event as an ndarray object
ImgAlgos.Tahometer evaluates performance of the batch job
ImgAlgos.ImgAverage produces file with averaged image for given range of events.

For data pre-processing

Scaner
Runs in batch for the data file to get preliminary information.
ImgAlgos.Tahometer evaluates performance of the batch job, counts number of events in the data file
ImgAlgos.ImgTimeStampList makes file with time stamps and time record counters for tau
ImgAlgos.IntensityMonitorsData makes file with intensity monitor records

Average
Runs in batch for the data file to get averaged image
ImgAlgos.Tahometer evaluates performance of the batch job
ImgAlgos.PrincetonImageProducer gets image from event as an ndarray object
ImgAlgos.ImgAverage produces file with averaged image for given range of events
ImgAlgos::ImgMaskEvaluation - module is available, but needs to be included in psana.cfg file for evaluation of:

• saturated mask
• noise mask
• combined mask

For data processing

Split
ImgAlgos.Tahometer
ImgAlgos.PrincetonImageProducer
ImgAlgos.ImgVsTimeSplitInFiles

Process
Data processing is implemented in stand alone (non-psana) modules
ImgAlgos.CorAna.cpp
ImgAlgos.CorAnaInputParameters.cpp
ImgAlgos.CorAnaData.cpp

Merge
CorAnaMergeFiles.cpp - saves binary file for float(32) with shape (Ntau,3,rows,cols), where 3 stands for <Ip>, <If>, and <Ip*If>
Can be acessed in python as

    sp.cor_arr = np.fromfile(sp.fname, dtype=np.float32)
    <image-size> = rows * cols
    nptau = <file-size>/<image-size>/3
    sp.cor_arr.shape = (nptau, 3, rows, cols)

GUI

The system of GUIs, consisting of dozens of CorAna.GUI... modules is implemented in the draft approximation. Roughly it reproduces all features of the old program.

Input parameters GUIs

CorAna.GUI... most of them are available. Will be added or extended if necessary.
Files - define input files and do pre-processing
Setup Info
Analysis Info
System
Intensity Monitor needs to be used at processing stage

Run GUI

Input - short summary of input info for data processing
Split - control and monitoring for the 1st stage of processing
Process - ... 2nd stage ...
Merge - ... 3d stage ...
Auto - ... for all 3 stages ...

View Results GUI

Contains a set of control fields for presentation of results

Graphics

PlotArray*.py - for Intensity Monitors
PlotImgSpe*.py - for images, partition maps, masks etc.
PlotTime*.py - for time stamp monitoring
PlotG2*.py - G2 plot peresentation
Mask editors...

Infrastructural modules

Infrastructural modules provide basic infrastructure of the project.
ConfigParameters.py, ConfigParametersCorAna.py, and GUIConfigParameters.py provides convenient approach for maintenance of all configuration parameters.
Infrastructural modules Logger.py and GUILogger.py provides a generic approach to logging system.
Module ConfigFileGenerator.py use current settings of configuration parameters and stub-file scripts from CorAna/data/scripts/ and generates the psana configuration files.
Module FileNameManager.py is a single place which provides a dynamic file names for current version of the configuration parameters.
GlobalUtils.py - global utilities for common operations
BatchJob.py - superclass for batch job submission
BatchJobPedestals.py - pre-processing for dark run files
BatchJobData.py - re-processing for data files
BatchJobCorAna.py - main data processing - calculation algorithm.

Progress

2012-11-05 – 2012-11-11 Week progress

A src/PsanaConfigFileGenerator.py
A src/FileNameManager.py
A src/BatchJobPedestals.py
A src/AppDataPath.py
A src/ConfigFileGenerator.py
A src/GUIDark.py
A src/ImgSpeButtons.py
A src/ImgSpeNavToolBar.py
A src/ImgSpeWidget.py
A src/ImgSpeWithGUI.py
A data/scripts/psana-pedestals.cfg
A data/scripts/psana-tahometer.cfg

2012-11-12 – 2012-11-18 Week progress

A src/GUIFileBrowser.py - file browser for text files of this project.
A src/GUIHelp.py - pop-up box with message as a widget.
A src/GUIWorkResDirs.py - separate widget for setting of work and result output directory and common prefix file name.

In src/GUILogger.py and src/Logger.py - now messages are accumulated in 2d tuple. Add filtration by the message type "debug", "info", "warning", etc.

src/GUIFiles.py is significantly re-organized; now it is GUI with tabs.
All significant fields are moved in GUIs like src/GUIDark.py (GUIData, GUIFlat).

src/GUIDark.py - is re-organized.

In the interactive plot:

• In src/ImgSpeWithGUI.py the src/ImgSpeNavToolBar.py - is depricated
• src/ImgSpeButtons.py - re-organized: add save, reset, and menu buttons, log check-box.
• In src/ImgSpeWidget.py - the log functionality needs to be implemented...

2012-11-19 – 2012-11-25 Week progress

Suggestions from Marcin:

• Time(sec) - should be the averaged time between frame time-stamps.
• Spectral plot: log-scale for X-axis only.
• Flat field and Blamish files will be provided from external apps. No need to evaluate them now. Needs to be selected by the path.
• Substitute the start and end events after the scan.

Progress:

• Restracturization: tahometer replaced by peds_scan.
• update the start and end fields after scan.
• remove psana log files before job submission in batch.
• the start and end fields are set from cofig.pars, if the scan log-file is availabe, they are updated from this file.
• reset fields to default if the input xtc file is changed.
• A src/GUIFlatField.py - file selection, plot, browse file
• A src/GUIBlamish.py
• A src/GUIMainTB.py - try different view for composite window
• A src/GUIMainSplit.py - try different view for composite window
• Add psana module: ImgAlgos::ImgTimeStampList - creates the time-stamp list and evaluates the mean time between frames and its RMS.
• in GUIDark.py use variables produced by the ImgAlgos::ImgTimeStampList.

2012-11-26 – 2012-12-02 Week progress

Suggestions from Marcin:

• Marcin prefers layout with separated GUILogger window, like in GUIMainTB.py
• Need in a tool reporting problems in time-stamp sequence.
• In GUIDark.py - untill the total number of events is unknown, use end "-1" and generate a warning if it is submitted in batch for pedestals.
• Add check-box showing if the Blamish and Flat-field files will be applied
• Marcin returned the paper-copy of GUIs with comments about titles and parameters.
• The list of variables for intensity monitor:

  • IPM2 [CH0,...,CH3]

  • IPMMON0 [...]

  • IPM4 [...]

  • IPM5 [...]

  • GASDET [...]

• Needs in separate GUI for intensity monitors for normalization and veto.

Progress:

• In GUIBlamish.py and src/GUIFlatField.py add check-boxes.
• Apply Marcin's comments to GUIs with System, Setup etc. parameters, GUICCDCorrectionSettings.py is removed..
• In src/GUIMainTB.py optimized the GUI sizes.
• Changed names from GUIBatch... to GUISystem...
• New psana module Psana Module Catalog, which gets data for equivalent HDF5 variables:

  /Configure:0000/Run:0000/CalibCycle:0000/Bld::BldDataFEEGasDetEnergy/FEEGasDetEnergy
  /Configure:0000/Run:0000/CalibCycle:0000/Bld::BldDataIpimbV1/XCS-IPM-02
  /Configure:0000/Run:0000/CalibCycle:0000/Bld::BldDataIpimbV1/XCS-IPM-mono
  /Configure:0000/Run:0000/CalibCycle:0000/Ipimb::DataV2/XcsBeamline.1:Ipimb.4
  /Configure:0000/Run:0000/CalibCycle:0000/Ipimb::DataV2/XcsBeamline.1:Ipimb.5
  

  in psana they can be accessed as:

    m_srcFEEGasDetE = configStr("feeSource", "BldInfo(FEEGasDetEnergy)");
    m_srcIPM2       = configStr("ipm2",      "BldInfo(XCS-IPM-02)");
    m_srcIPMMono    = configStr("ipmmono",   "BldInfo(XCS-IPM-mono)");
    m_srcIPM4       = configStr("ipm4",      "DetInfo(XcsBeamline.1:Ipimb.4)");
    m_srcIPM5       = configStr("ipm5",      "DetInfo(XcsBeamline.1:Ipimb.5)");
  

  and share pointers:

    shared_ptr<Psana::Bld::BldDataFEEGasDetEnergy> fee = evt.get(m_srcFEEGasDetE);
    shared_ptr<Psana::Ipimb::DataV2> data2 = evt.get(m_src); // for m_srcIPM4 and m_srcIPM5
    shared_ptr<Psana::Bld::BldDataIpimbV1> ipimb1 = evt.get(m_src_ipimb); // for m_srcIPM2 and m_srcIPMMono
  

• In GUISystemPars: add total number of events and dt+-rms to GUISystemPars.
• In GUIDark: add check-box and file-browser button with complete functionality.
• In GUIDark: add protection for pedestal job submission in batch with default number of events.
• In BatchJobPedestals: add protection against duplicated job submission in batch using timeout (now 100 sec) or until the previous job is DONE.
• In ImgSpeWidget: implemented the log-scale plot with full functionality and protections...

2012-12-03 – 2012-12-09 Week progress

No meeting this week due to experiments in xcs hatch.

Progress:

• In GUIFileBrowser: add one more parameters after the list of files, the file to start browser
• GUIData is implemented with full functionality including
  • check box to activate/protect buttons
  • "File" button and info field - selects the xtc file
  • "Scan" button - submits scanner job in batch which
    1. counts events,
    2. saves in file the time records ind evaluate an average interval between frames
    3. saves in file the intensity monitor data
  • "Average" button - submits job in batch for averaging image for the specified range of events
  • "Check status" button - checks the status of both batch jobs, check if the file with averaged image is available, update the info fields for number of events and averaged time between frames
  • "Check files" button - prints the list of work files for data and check their availability
  • "Browse" button - open the file browser with a list of work files for data
  • "img-Plot" button - plots the averaged image from file
  • "t-Plot" button - shows the plot for time records saved in file
  • "Remove" button - removes all work files from the work directory
• GUIIntensityMonitors: GUI for Intensity monitors
• PlotArray.py, PlotArrayButtons.py, PlotArrayWidget.py - plot for array presentation
• PlotTime.py, PlotTimeWidget.py - plot for time records presentation

2012-12-10 – 2012-12-16 Week progress

Q&A Marcin:

• Q: XcsBeamline.2 - is it used simultaneously with XcsBeamline.1 ?
  • A: XcsBeamline.1
• Q: What control parameters should be added for intensity monitors?
  • A: should be added:
    1. radio button for normalization or no-normalization
    2. filter data check box
    3. low-upper amplitude range
• Q: Layout of GUIMainTB.py with variable height, is it ok?
  • A: Looks fine.
• Q: Layout variation for check-box - disappearance of buttons, is it ok?
  • A: Change color of button-titles.
• Q: What image buttons are used as a mask editor etc.
  • A: GUIAnalysisSettings:
    • "MaskPolygon" button - for polygon editor on data image
    • "Show Mask & Partitions" button - shows the mask and partitions on data image

Suggestions from Marcin:

• Add button to all plots and gui to submit them in ELog.
• Change log presentation for image histogram - log for vertical scale only.

Progress

• GUIDark and GUIData - inappropriate change of style for "end" field is fixed
• GUIDark, GUIData, GUIBlamish, GUIFlatField - have changed behavior for non-active buttons:
  • for now it is set to default - frame of buttons disappeared for non-active
  • GUIFlatField has an example of how to set a dynamic style-sheet for non-default behavior styles
• PlotImgSpeWidget.py - change log scale from x to y
• GUIIntensityMonitors.py - add more fields:
  • radio buttons for normalization
  • check boxes for selection
  • edit fields for Imin and Imax values with changing behavior and style depending on radio and check box status. Check them for equity to default and Imin >= Imax
• Check move window coordinates for plots... Some of them opens in unexpected places.
• GUIELogPostingDialog, GUIELogPostingFields, PlotArrayButtons, PlotImgSpeButtons, and GlobalUtils: submission of message with attached plots in ELog
• Add icons to some of frequently used buttons.
• Add button for some of GUIs amd plots to submit them in ELog.

2012-12-17 – 2012-12-23 Week progress

Q&A Marcin:

• Discuss layout of fields in
  • GUIIntensityMonitors - Is it necessary to use Imax/Imin for normalization?
  • GUIELogPostingDialog, GUIELogPostingFields - is it consistent with what you want?

• Layout of buttons
  • GUIMainTB - it is possible to change the "buttons with icons and text" -> "buttons with icons" -> "icons". Which style is preferred?
  • GUIDark.py, GUIData.py, etc. - style for active/non-active buttons - just default or add more colors, like in GUIFlatField ?

Progress

• GUIDark and GUIData - add check box for parameters cp.use_dark_xtc_all and cp.use_data_xtc_all with appropriate behavior for check box but_path and edi_path fields. Now, the name of the file for batch job is taken from fnm.path_dark_xtc_cond() and fnm.path_data_xtc_cond(), respectively.
• FileNameManager - fixed a few typos in the file name methods: "dark"->"data"

2012-12-24 – 2013-01-06 Winter break weeks progress

Progress

• GUIRun, GUIRunInfo, GUIRunSplit - new style GUI for the splitting procedure.

2013-01-07 – 2013-01-13 Week progress

Progress

• GUIRunProc, BatchJobCorAna, GlobalUtils

2013-01-14 – 2013-01-20 Week progress

Q&A Marcin:
Add lock for ELog submission dialog box
Add lock for nparts field in GUIRunInput
GUIIntensityMonitors:

• change response on radio button
• short intensity monitor names

Progress

• GUIRunInfo change name to GUIRunInput
• GUIRunProc - is completed
• GUIRunMerge
• GUIRunAuto
• GUIELogPostingDialog, GUIELogPostingFields - add lock, restructorization for editable fields.
• GUIIntensityMonitors - (1) remove the selection fields style change for radio button, (2) use a list of short intensity monitor names.

2013-01-21 – 2013-02-03 Week progress

Q&A Marcin:
Reorganize GUISetupInfoLeft.py for CCD position input in modes: Beam Zero, Specular, and Data

Progress
Fixed image partitioning, processing, and merging in ImgAlgos/src/ImgVsTimeSplitInFiles.cpp, CorAnaData.cpp, and CorAnaMergeFiles.cpp
Add CCD orientation in GUICCDSettings

• CCD is always displayed as a matrix of (row,col) with origin in the top left corner.
• CCD orientation is assumed as entire world is rotated w.r.t. CCD

Work on

• GUIViewControl.py
• GUIViewResults.py
• ViewResults.py
• g2 Normalization - need to use parameters from GUISetupInfo etc.
  Map evaluation speed in ViewResults.py for 1300x1340 img
• x,y maps = 0.02sec
• r map = 0.05sec
• theta map = 0.12sec
• q map = 0.4sec

2013-02-04 – 2013-02-10 Week progress

Progress
Work in progress on

• GUIViewControl.py
• ViewResults.py
• PlotG2*.py

Currently available plots for:

• Raw data maps (tau): <Ip>, <If>, <Ip*If>, g2-raw
• Geometry maps: X, Y, R, Phi, Q for direct beam mode
• Partitioning maps: Phi, Q, Q-Phi maps for static and dynamic partitions
• Masks: Image limits, blemish, hot pixels, good regions, total
• Calculated maps (tau): total mask as applied at normalization and for maps
  • normalization factors 1/<Ip>, 1/<If> for static bins,
  • g2 as a map
  • g2 averaged over dynamic bins
  • g2 for dynamic bins vs tau as a 2D histogram
  • g2(tau) plot for dynamic bins as a set of graphic

Q to Marcin:

• How to calculate angle and Q in reflective geometry?
• Presentation of CCD image and detector orientation issues.

To-do list

Masks
Evaluation at data pre-processing the noisy and saturated pixel mask
Graphical editor for regions of interest
Graphical editor for blemish mask

Processing (at psana Split level?)
Intensity monitors: needs to be used at processing in psana
Account for thresholds

View Results
geometry for reflected beam
proper combination for total mask in ViewResults.py
fits

ELog
Add submission using responce ID, when Igor provide an interface
GlobalUtils: comment/uncomment submission in ELog