...
This page holds a few example code-snippets for use in pyana analysis. The analysis is written in python and uses MatPlotLib.PyPlot for plotting of data. Compare with myana user examples to see how (some of) the same things can be done using the myana analysis framework. The most reliable place for up-to-date information about all the event getters in pyana, see: https://confluence.slac.stanford.edu/display/PCDS/Pyana+Reference+Manual#PyanaReferenceManual-Classpyana.event.EventImage Removed
For all the examples, you may assume that the pyana module contains a class with at least 'beginjob', 'event' and 'endjob' functions that starts something like this:
...
Code Block |
---|
| none |
---|
| none |
---|
title | getFeeGasDet |
---|
|
fee_energy_array = evt.getFeeGasDet()
gdENRC11 = fee_energy_array[0]
gdENRC12 = fee_energy_array[1]
gdENRC21 = fee_energy_array[2]
gdENRC22 = fee_energy_array[3]
energy = (gdENRC21 -+ gdENRC22) / 2.0
# or use the first two that has a different gain:
energy = (gdENRC11 -+ gdENRC12) / 2.0
|
BeamLine Data: Phase Cavity
...
Code Block |
---|
| none |
---|
| none |
---|
title | getPhaseCavity |
---|
|
pc = evt.getPhaseCavity()
try:
pcFitTime1 = pc.fFitTime1
pcFitTime2 = pc.fFitTime2
pcCharge1 = pc.fCharge1
pcCharge2 = pc.fCharge2
print "PhaseCavity: ", pcFitTime1, pcFitTime2, pcCharge1, pcCharge2
except :
print "No Phase Cavity object found"
|
...
Event code
Code Block |
---|
| none |
---|
| none |
---|
title | EncoderDataEnvData |
---|
|
def event(self, evt, env):
evrdata try:
= evt.getEvrData("NoDetector-0|Evr-0")
for i encoderin =range evt(evrdata.get(xtc.TypeId.Type.Id_EncoderData, self.enc_source )numFifoEvents()):
encoder_value = encoder.value()
except:
print "No encoder found in this event"
return
|
You could combine it with phase cavity time, and compute a time delay from it, for example (I don't know the origin of these parameters!):
...
# Encoder Parameters to convert to picoseconds
delay_a = -80.0e-6;
delay_b = 0.52168;
delay_c = 299792458;
delay_0 = 0;
delay_time = (delay_a * encoder_value + delay_b)*1.e-3 / delay_c)
delay_time = 2 * delay_time / 1.0e-12 + delay_0 + pcFitTime1
Time data
The time of the event can be obtained within the event function:
...
def event ( self, evt, env ) :
event_time = evt.getTime().seconds() + 1.0e-9*evt.getTime().nanoseconds() )
print "Event code: ", evrdata.fifoEvent(i).EventCode
|
In the example above, the address of the EvrData object is given as "NoDetector-0|Evr-0". The address may be different in other cases, so make sure you have the correct address. If you don't know what it is, you can use 'pyxtcreader -vv <xtcfile> | less' to browse your xtcfile and look for it. Look for a lines with 'contains=EvrConfig_V' or 'contains=EvrData_V'. The address will be found on the same line in 'src=DetInfo(<address>)'
Encoder data (delay scanner)
Code Block |
---|
|
def event(self,evt,env):
try:
encoder = evt.get(xtc.TypeId.Type.Id_EncoderData, self.enc_source )
encoder_value = encoder.value()
except:
print "No encoder found in this event"
return
|
You could combine it with phase cavity time, and compute a time delay from it, for example (I don't know the origin of these parameters!):
Code Block |
---|
|
# Encoder Parameters to convert to picoseconds
delay_a = -80.0e-6;
delay_b = 0.52168;
delay_c = 299792458;
delay_0 = 0;
delay_time = (delay_a * encoder_value + delay_b)*1.e-3 / delay_c)
delay_time = 2 * delay_time / 1.0e-12 + delay_0 + pcFitTime1
|
Time data
The time of the event can be obtained within the event function:
Code Block |
---|
|
def event ( self, evt, env ) :
event_time = evt.getTime().seconds() + 1.0e-9*evt.getTime().nanoseconds() )
|
IPIMB diode data
This is data from sets of 4 diodes around the beam line (Intensity Position, Intensity Monitoring Boards)
that measures the beam intensity in four spots, from which we can also deduce the position of the beam.
Currently there are two data structures that holds data from the same type of devices. Depending on DAQ
configuration, they are either DetInfo type or BldInfo type. Here are examples for extracting both types
in the user module event function:
Code Block |
---|
|
def event(self, evt, env):
# raw data
ipmRaw = evt.get(xtc.TypeId.Type.Id_IpimbData, source )
try:
ch = [ipmRaw.channel0(),
ipmRaw.channel1(),
ipmRaw.channel2(),
ipmRaw.channel3() ]
|
IPIMB diode data
This is data from sets of 4 diodes around the beam line (Intensity Position, Intensity Monitoring Boards)
that measures the beam intensity in four spots, from which we can also deduce the position of the beam.
Currently there are two data structures that holds data from the same type of devices. Depending on DAQ
configuration, they are either DetInfo type or BldInfo type. Here are examples for extracting both types
in the user module event function:
Code Block |
---|
none | none | title | DetInfo |
---|
def event(self, evt, env):
# raw data
ipmRaw = evt.get(xtc.TypeId.Type.Id_IpimbData, source )
try:
ch = [ipmRaw.channel0(),
ipmRaw.channel1(),
ipmRaw.channel2(),
ipmRaw.channel3() ]
ch_volt = [ipmRaw.channel0Volts(),
ipmRaw.channel1Volts(),
ipmRaw.channel2Volts(),
ipmRaw.channel3Volts() ]
except:
pass
# feature-extracted data
ipmFex = evt.get(xtc.TypeId.Type.Id_IpmFex, source )
try:
# array of 4 numbers
fex_channel = ipmFex.channel
# scalar values
fex_sum = ipmFex.sum
fex_xpos = ipmFex.xpos
fex_ypos = ipmFex.ypos
except:
pass
|
Code Block |
---|
none | none |
title | BldInfo |
---|
def event(self, evt, env):
ipm = evt.getSharedIpimbValue("HFX-DG3-IMB-02")
# or equivalently:
# ipm = evt.get(xtc.TypeId.Type.Id_SharedIpimb, "HFX-DG3-IMB-02")
try:
### Raw data ###
#ch_volt arrays of 4 numbers:
= [ipmRaw.channel0Volts(),
ch = [ipm.ipimbData.channel0(),
ipmRaw.channel1Volts(),
ipm.ipimbData.channel1(),
ipm.ipimbData.channel2ipmRaw.channel2Volts(),
ipmipmRaw.ipimbData.channel3channel3Volts()]
except:
ch_volt = [ipm.ipimbData.channel0Volts(),
pass
# feature-extracted data
ipmFex ipm.ipimbData.channel1Volts(),
= evt.get(xtc.TypeId.Type.Id_IpmFex, source )
try:
# array of ipm.ipimbData.channel2Volts(),4 numbers
fex_channel ipm.ipimbData.channel3Volts()]= ipmFex.channel
### Feature-extracted# datascalar ###values
# array of 4 numbers:
fex_channels = ipm.ipmFexData.channel
# scalars:
fex fex_sum = ipmipmFex.ipmFexData.sum
fex_xpos = ipmipmFex.ipmFexData.xpos
fex_ypos = ipm.ipmFexDataipmFex.ypos
except:
pass
|
Acqiris waveform data
This method can be used for any detector/device that has Acqiris waveform data. Edit the self.address field to get the detector of your choice.
Initialize class members:
Code Block |
---|
|
def __init__ ( self event(self, evt, env):
ipm # initialize data= evt.getSharedIpimbValue("HFX-DG3-IMB-02")
# or equivalently:
self.address # ipm = evt.get(xtc.TypeId.Type.Id_SharedIpimb, "AmoITof-0|Acqiris-0"HFX-DG3-IMB-02")
try:
### Raw self.data =###
[]
# self.counterarrays = 0
|
If you're curious to see any of the Acqiris configuration, e.g. how many channels do we have, you can inspect the AcqConfig object:
Code Block |
---|
none | none | of 4 numbers:
def beginjob (ch self, evt, env ) :
= [ipm.ipimbData.channel0(),
cfg = env.getConfig( _pdsdata.xtc.TypeId.Type.Id_AcqConfig, self.address )
ipm.ipimbData.channel1(),
self.num = cfgipm.ipimbData.nbrChannels()
|
The read the event waveform data (an array) and append it to the self.data list:
Code Block |
---|
none | none | channel2(),
def event ( self, evt, env ) :
ipm.ipimbData.channel3()]
channel = 0
acqData = evt.getAcqValue( self.address, channel, env)
ch_volt = [ipm.ipimbData.channel0Volts(),
if acqData :
ipm.ipimbData.channel1Volts(),
self.counter+=1
wf = acqDataipm.ipimbData.waveformchannel2Volts(),
# returns a waveform array of numpy.ndarray type.
self.data.append(wf)
|
At the end of the job, take the average and plot it:
Code Block |
---|
none | none | ipm.ipimbData.channel3Volts()]
def endjob( self, env ) :
### Feature-extracted data ###
data# = np.array(self.data) # this is an array of shape (Nevents, nSamples)
array of 4 numbers:
fex_channels = ipm.ipmFexData.channel
# take
the mean of all events for each sampling# timescalars:
xsfex_sum = np.mean(data, axis=0)
ipm.ipmFexData.sum
plt.plot(xs)
fex_xpos = ipm.ipmFexData.xpos
plt.xlabel('Seconds')fex_ypos = ipm.ipmFexData.ypos
plt.ylabel('Volts')
except:
plt.show()
|
Which gives you a plot like this
Image Removed
Princeton camera image
When plotting with MatPlotLib, we don't need to set the limits of the histogram manually, thus we don't need to read the Princeton configuration for this. If we want to sum the image from several events, we must first define and initialize some variables:
Acqiris waveform data
This method can be used for any detector/device that has Acqiris waveform data. Edit the self.address field to get the detector of your choice.
Initialize class members:
Code Block |
---|
Code Block |
---|
|
|
def __init__ ( self ):
# initialize data
self.address = "SxrEndstationAmoITof-0|PrincetonAcqiris-0"
self.data = None
[]
self.counter = 0
|
If you're curious to see any of the Acqiris configuration, e.g. how many channels do we have, you can inspect the AcqConfig objectIn each event, we add the image array returned from the getPrincetonValue function:
Code Block |
---|
| none |
---|
| none |
---|
title | getPrincetonValue |
---|
|
def eventbeginjob ( self, evt, env ) :
framecfg = evtenv.getPrincetonValuegetConfig( _pdsdata.xtc.TypeId.Type.Id_AcqConfig, self.address, env)
self.num = cfg.nbrChannels()
|
The read the event waveform data (an array) and append it to the self.data list:
Code Block |
---|
|
if frame :
def event ( self, evt, env # accumulate the data) :
channel if self.data is None := 0
acqData self.data = np.float_(frame.data()= evt.getAcqValue( self.address, channel, env)
if acqData else :
self.data += frame.data()
|
At the end of the job, display/save the array:
Code Block |
---|
counter+=1
def endjob( self, env ) :
wf = acqData.waveform() plt.imshow( self.data/self.countpass, origin='lower')
# returns a waveform array of numpy.ndarray type.
plt.colorbar()
pltself.data.showappend(wf)
|
At the end of the job, take the average and plot it:
Code Block |
---|
|
def endjob( self, env # save the full image to a png file
plt.imsave(fname="pyana_princ_image.png",arr=self.data, origin='lower')
|
Note that imsave saves the image only, pixel by pixel. If you want a view of the figure itself, lower resolution, you can save it from the interactive window you get from plt.show().
Image Removed
PnCCD image
Code Block |
---|
none | none |
title | getPnCcdValue |
---|
def event(self,evt,env):
try:
frame = evt.getPnCcdValue( self.source, env )
) :
data = np.array(self.data) # this is an array of shape (Nevents, nSamples)
# take the mean of all events for each sampling time
xs = np.mean(data, axis=0)
plt.plot(xs)
image = frameplt.dataxlabel('Seconds')
except: plt.ylabel('Volts')
pass
|
Other image (FCCD*,Opal,PIM (TM6740), ... )
Which gives you a plot like this
Image Added
Princeton camera image
When plotting with MatPlotLib, we don't need to set the limits of the histogram manually, thus we don't need to read the Princeton configuration for this. If we want to sum the image from several events, we must first define and initialize some variablesThese all use the generic getFrameValue function:
Code Block |
---|
| none |
---|
| none |
---|
title | getFrameValue |
---|
|
def event(self,evt,env
def __init__ ( self ):
try:
# frame = evt.getFrameValue( self.source )initialize data
imageself.address = frame.data() "SxrEndstation-0|Princeton-0"
except:
self.data = pass
|
FCCD (Fast CCD) image
In each event, we add the image array returned from the getPrincetonValue function:The Fast CCD is read out as two 8-bit images, therefore you need this extra line to convert it to the right format.
Code Block |
---|
| none |
---|
| none |
---|
title | getFrameValuegetPrincetonValue |
---|
|
def event ( self, evt, env ) :
try:
frame = evt.getFrameValuegetPrincetonValue( self.sourceaddress, env)
if image = frame.data()frame :
except:
# accumulate the passdata
# convert to 16-bit integer
if imageself.dtypedata is = np.uint16
|
CsPad data
Here's an example of getting CsPad data from an event:
Code Block |
---|
none | none |
title | getCsPadQuads |
---|
def event(self,evt,env):
None :
quads = evt.getCsPadQuads(self.img_source, env)
if notself.data quads := np.float_(frame.data())
print '*** cspad information is missing ***'else :
return
self.data += frame.data()
|
At the end of the job, display/save the array:
Code Block |
---|
def endjob( self, env ) :
plt.imshow( self.data/self.countpass, origin='lower')
plt.colorbar()
plt.show()
# save the full image to a png file
plt.imsave(fname="pyana_princ_image.png",arr=self.data, origin='lower')
|
Note that imsave saves the image only, pixel by pixel. If you want a view of the figure itself, lower resolution, you can save it from the interactive window you get from plt.show().
Image Added
PnCCD image
Code Block |
---|
| none |
---|
| none |
---|
title | getPnCcdValue |
---|
|
def event(self,evt,env):
try:
frame = evt.getPnCcdValue( self.source, env )
image = frame.data()
except:
pass
|
Other image (FCCD*,Opal,PIM (TM6740), ... )
These all use the generic getFrameValue function:
Code Block |
---|
| none |
---|
| none |
---|
title | getFrameValue |
---|
|
def event(self,evt,env):
try:
frame = evt.getFrameValue( self.source )
image = frame.data()
except:
pass
|
FCCD (Fast CCD) image
The Fast CCD is read out as two 8-bit images, therefore you need this extra line to convert it to the right format.
Code Block |
---|
| none |
---|
| none |
---|
title | getFrameValue |
---|
|
def event(self,evt,env):
try:
frame = evt.getFrameValue( self.source )
image = frame.data()
except:
pass
# convert to 16-bit integer
image.dtype = np.uint16
|
CsPad data
Here's an example of getting CsPad data from an event:
Code Block |
---|
| none |
---|
| none |
---|
title | getCsPadQuads |
---|
|
def event(self,evt,env):
quads = evt.getCsPadQuads(self.img_source, env)
if not quads :
print '*** cspad information is missing ***'
return
# dump information about quadrants
print "Number of quadrants: %d" % len(quads)
for q in quads:
print " Quadrant %d" % q.quad()
print " virtual_channel: %s" % q.virtual_channel()
print " lane: %s" % q.lane()
print " tid: %s" % q.tid()
print " acq_count: %s" % q.acq_count()
print " op_code: %s" % q.op_code()
print " seq_count: %s" % q.seq_count()
print " ticks: %s" % q.ticks()
print " fiducials: %s" % q.fiducials()
print " frame_type: %s" % q.frame_type()
print " sb_temp: %s" % map(q.sb_temp, range(4))
# image data as 3-dimentional array
data = q.data()
|
So far so good. 'quads' is a list of CsPad Element objects, and not necessarily ordered in the expected way. So you'll need to use q.quad() to obtain the quad number.
q.data() gives you a 3D numpy array [row][col][sec]. Here sections will be ordered as expected, but be aware in case of missing sections, that you may need to check the
configuration object. You can get that from the env object, typically something you do in beginjob:
Code Block |
---|
|
def beginjob(self,evt,env):
config = env.getConfig(xtc.TypeId.Type.Id_CspadConfig, self.img_source)
if not config:
# dump information about quadrants
print "Number of quadrants: %d" % len(quads)
for q in quads:
print " Quadrant %d" % q.quad()
print "'*** cspad config object virtual_channel: %s" % q.virtual_channel()is missing ***'
printreturn " lane: %s" % q.lane()
print " tid: %s" % q.tid()Cspad configuration"
print " N print "quadrants acq_count: %s%d" % qconfig.acq_countnumQuads()
print " Quad printmask " op_code: %s%#x" % qconfig.op_codequadMask()
print " payloadSize seq_count: %s%d" % qconfig.seq_countpayloadSize()
print " badAsicMask0 ticks: %s%#x" % qconfig.ticksbadAsicMask0()
print " badAsicMask1 fiducials: %s%#x" % qconfig.fiducialsbadAsicMask1()
print " asicMask print " frame_type: %s%#x" % qconfig.frame_typeasicMask()
print " numAsicsRead sb_temp: %s%d" % map(q.sb_temp, range(4))config.numAsicsRead()
# get the indices of sections in use:
qn = range(0,config.numQuads()) # image data as 3-dimentional array
dataself.sections = q.data()
|
So far so good. 'quads' is a list of CsPad Element objects, and not necessarily ordered in the expected way. So you'll need to use q.quad() to obtain the quad number.
q.data() gives you a 3D numpy arrayrowcolsec. Here sections will be ordered as expected, but be aware in case of missing sections, that you may need to check the
configuration object. You can get that from the env object, typically something you do in beginjob:
...
def beginjob(self,evt,env):
config = env.getConfig(xtc.TypeId.Type.Id_CspadConfig, self.img_source)
if not config:
print '*** cspad config object is missing ***'
return
print "Cspad configuration"
print " N quadrants : %d" % config.numQuads()
print " Quad mask : %#x" % config.quadMask()
print " payloadSize : %d" % config.payloadSize()
print " badAsicMask0 : %#x" % config.badAsicMask0()
print " badAsicMask1 : %#x" % config.badAsicMask1()
print " asicMask : %#x" % config.asicMask()
print " numAsicsRead : %d" % config.numAsicsRead()
# get the indices of sections in use:
qn = range(0,config.numQuads())
self.sections = map(config.sections, qn )
...
map(config.sections, qn )
|
If you want to draw the whole CsPad image, there's currently no pyana function that does this. Pyana only supplies the pixels in a numpy array, and the
exact location of each pixel depends on the conditions at the time of data collection. A simplified way of making the image can be seen in cspad_simple.py(newer version (cspad.py) available if you check out the XtcExplorer package).
CSPad pixel coordinates.
The CSPad detector image can be drawn by positioning the sections from the data array into a large image array. This is done in cspad_simple.py above. The positions are extracted from optical meterology measurements and additional calibrations. Alternatively one can find the coordinate of each individual pixel from a pixel map, based on the same optical metrology measurements. This is described in details here
Epics Process Variables and ControlConfig
EPICS data is different from DAQ event data. It stores the conditions and settings of the instruments, but values typically change more slowly than your
average shot-by-shot data, and EPICS data is typically updated only when it changes, or every second, or similar. It is not stored in the 'evt' (event) object,
but in the 'env' (environment) object. You typically would read it only at the beginning of each job or if your doing a scan, you'd read it in every calibration cycle:
Code Block |
---|
| none |
---|
| none |
---|
title | env.epicsStore() |
---|
|
def begincalibcycle(self,evt,env):
## The returned value should be of the type epics.EpicsPvTime.
pv = env.epicsStore().value( pv_name )
if not pv:
logging.warning('EPICS PV %s does not exist', pv_name)
else:
value = pv.value
status = pv.status
alarm_severity = pv.severity
|
Code Block |
---|
| none |
---|
| none |
---|
title | ControlConfig |
---|
|
def begincalibcycle(self,evt,env):
ctrl_config = env.getConfig(xtc.TypeId.Type.Id_ControlConfig)
nControls = ctrl_config.npvControls()
for ic in range (0, nControls ):
cpv = ctrl_config.pvControl(ic)
name = cpv.name()
value = cpv.value()
|