Configuring Analysis Output¶
Bob Michaels
email: rom@jlab.org
FEATURES¶
Class THaOutput can be used to control the output of the new Root/C++ analyzer . The output is defined by an output definition file, e.g. what is specified when you do 'analyzer>SetOdefFile("output.def");' and has the following elements. Note: To get vectorized formulas and histograms you must be using version 1.1 and higher of the analyzer. To get EPICS and Scaler data you need version 1.2.
 Global variables written to the tree. Can include arrays  fixed or variable size. To see a list of variables, type 'gHaVars>Print()' at the analyzer prompt.
 Blocks of variables, e.g. L.* is everything in Left Spectrometer.
 Formulas involving variables. These can be vectors.
 Cuts. They appear in tree as 0 (false) or 1 (true). Also can be vectors.
 Histograms. These can involve variables and/or formul40.65a, as well as optional cuts. The histograms can be vectors depending on the rules for dimensionality.
 EPICS data (slow controls variables recorded in datastream.)
 Scalers (i.e. data from devices called scalers that count things).
SYNTAX of "output.def" FILE¶
To add global variables, formulas, and histograms, the
syntax of each line of "output.def" is of the form
keyword restofline
where the keywords are as follows :
(keywords are caseinsensitive, the rest of the line is not)
VARIABLE  indicates that a global variable is to
be added to the tree. The next string is
the name of the variable. You can see a list
of variables by typing gHaVars>Print() at
the analyzer's root shell prompt.
A variable can also be an array, e.g. R.s1.lt
Then the values in the tree will be R.s1.lt[0],
R.s1.lt[1], etc, up to the array size which
is Ndata.R.s1.lt
BLOCK  An entire block of variables are written to the
output. E.g. "L.*" writes all Left HRS variables.
FORMULA  indicates a THaFormula to add to the output.
The next word will be the "name" of the formula result
in the tree. The 3rd string is the formula to evaluate.
Note, it cannot have spaces in it.
Formulas can be vectors if the arrays are fixed size.
CUT  Like a formula, but the data are 0 (false) or 1 (true)
TH1F, TH2F, TH1D, TH2D  defines histograms (1D or 2D).
The next word is the name of the object.
The following string (must be in single quotes) is 40.65
the title of the histogram. Next, the variables
or formulas names that are put into the histogram
are listed together with bin info: nbin, xlo, xhi and if 2D
the same bin info for y. Optional cuts can be specified
at the end of the line. See examples below.
One can add EPICS data or SCALER data to the output with a pair of
("begin", "end") statements in the output.def file. This adds the data
to both the main tree as well as to new trees. The new trees are "E"
for epics data, and "left" or "right" for the left and right HRS scalers.
begin epics40.65
# put list of variables here
end epics
And for the "left" spectrometer scalers (similarly for "right"):
begin scalers left
# put list of variables here
end scalers
LEARN BY EXAMPLE¶
#  Example output.def  # Variables appear in the tree. variable L.s1.lt[4] variable L.s1.rt40.65 # The 'block' variables: All data in Right HRS go to tree. block R.* # Formulas can be scalers or vectors. # Lt4a is a scaler. formula Lt4a 5.*L.s1.lt[4] # Lt4b is a vector. Access by Lt4b[0], Lt4b[1],... formula Lt4b 5.*L.s1.lt # Tsm is a vector formula Tsm 4*L.s1.lt+2*L.s1.rt+3*L.s1.lt[4] # Tsm[0] = 4*L.s1.lt[0]+2*L.s1.rt[0]+3*L.s1.lt[4] # Tsm[1] = 4*L.s1.lt[1]+2*L.s1.rt[1]+3*L.s1.lt[4] # etc... # Cuts can be defined globally and used in histograms. # Cut C1 is a scaler. Data is 0 or 1. cut C1 L.s1.lt[4]>1350 # Cut C2 is a vector. Access as C2[0], C2[1],... cut C2 L.s1.lt>1160 # Histograms can involve formulas, variables, and cuts. # th1f, th1d, th2f, th2d supported. # Formulas and cuts may be global, or defined in the line. # Histograms are vector or scalers whose dimensionality is # determined by the dimensionality of its components. # 1. dim(X)>=0 && dim(Y)==0 && dim(Cut)==0 > Data in 1 histo, # and if dim(X)>1 all components go into 1 histo. # 2. dim(X)>1 && (dim(Y)>1  dim(Cut)>1) > vector of histos # appear in output, the dim's must match and the indices track. # The "eye" variable [I] becomes an index that tracks the # index of the accompanying variable. L.s1.lt[I] is synonymous # with L.s1.lt. # Prefixes SUM:, OR:, and AND: supported. # SUM: the sum of variables or sum of a vector formula. # AND:, OR: logical "and","or" of cut conditions. TH1F rv1n 'Larm vdc hits on V1' L.vdc.v1.nhit 10 0 10 th1f rv1w 'Larm vdc wiremap on V1' L.vdc.v1.wire 200 0 400 TH1F rv1z 'Larm vdc TDC on V1' L.vdc.v1.rawtime 200 0 2500 TH1d hLt4 'Lt4a formula' Lt4a 100 0 9000 th1d hLta 'Lt4b formula' Lt4b 100 0 9000 TH1F Lt4c0 'Lt4 from var, no cut' L.s1.lt[4] 100 1 2000 TH1F Lt4c1 'Lt4 from var, cut' L.s1.lt[4] 100 1 2000 L.s1.lt[4]>1300. TH1F Lt4c2 'Lt4 from var, cut C1' L.s1.lt[4] 100 1 2000 C1 TH1F Lt4c3 'Lt4 from formula, cut C1' Lt4a 100 1 9000 C1 TH1F Lts1 'L.s1.rt (all elements)' L.s1.rt 100 1 2000 # Since Lt4b and C2 are vectors, get histograms Ltv0, Ltv1, ... Ltv5 TH1F Ltv 'Vector formula, cut C2' Lt4b 100 1 20000 C2 # The "eye" variable as R. Feuerbach requested TH2F a1eye 'Eye test 1' [I] L.s1.lt 14 0 7 100 0 2000 TH2F a2eye 'Eye test 2' [i] L.s1.lt[I] 14 0 7 100 0 2000 # same as above TH2F a3eye 'Eye test 3' [I] Lt4b 14 0 7 100 0 9000 TH2F a4eye 'Eye test 4' [I] Lt4b 14 0 7 100 0 9000 L.s1.lt>1300 TH2F a5eye 'Eye test 5' [I] Lt4b 14 0 7 100 0 9000 C2 # Prefixes (SUM: OR: AND:) TH1F Lsum1 'Test of sum 1' SUM:L.s1.lt 100 0 15000 TH2f Lsum2 'Test of sum 2' SUM:L.s1.rt SUM:L.s1.rt 100 0 15000 100 0 2000 TH1F Lor1 'Test of or 1' L.s1.lt[4] 100 0 2000 OR:L.s1.lt>1000 TH1F Lor2 'Test of or 2' L.s1.lt 100 0 2000 OR:L.s1.lt>1000 TH1F Land1 'Test of and 1' L.s1.lt 100 0 2000 AND:L.s1.lt<1200 #  # For reference, here is a collection of things that do NOT work. # Cannot define formula in terms of variable sized arrays. # Exception: 'formula Lv1w L.vdc.v1.wire' results in Lv1w[0], Lv1w[1]... # formula Lw2 L.vdc.v1.wire[2] # error # formula Lw3 3*L.vdc.v1.wire # error # A cut defined by a formula does NOT work. Perhaps in future. # cut C4 Lta>1160 # Cannot use prefix like SUM: inside a formula. # TH1F Lnot 'Invalid test1' 3*SUM:L.s1.lt 100 0 20000 # If Lta is a formula, then cut = Lta>1160 is illegal. # Must instead write the formula explicitly. # TH1F Lte 'Lta vector formula, cut 3' Lta 100 1 2000 Lta>1160 #  # EPICS data to appear in regular event tree ("T") and # a new EPICS tree ("E"). The E tree is only filled # on EPICS events. The E tree also has a timestamp. # Access via T>Draw("hac_bcm_average") # and E>Draw("hac_bcm_average") # and E>Draw("hac_bcm_average:timestamp") # Notice that slia:DataValid appears in the # tree as slia_DataValid (i.e. ":" is replaced # by "_"). This replacement makes TTree::Draw # work properly. Note also, the value of # slia_DataValid is 71 if slia:DataValid # was "Yes" (case senstive!), 422 if "No" and # 0 otherwise. begin epics hac_bcm_average IPM1H04A.XPOS IPM1H04A.YPOS IPM1H04B.XPOS IPM1H04B.YPOS slia:DataValid Yes=71 No=422 end epics # Rules about EPICS variable assignment: # You can now do things like this in # the output definition file begin epics FB_A:use_RF 'RF On'=14 'RF Off'=52 slia:DataValid Yes=1 No=0 end epics # The rule is that everything between single # quotes is EPICS data (string type) which, # if found in the datastream, gets assigned # the numerical value after the equal sign. # So, if the string "RF On" is found, you get # 14 in the tree. The strings with white space # must contain single quotes. The other strings # could contain single quotes, but it is not # necessary (e.g. slia:DataValid 'Yes'=1). # Examples of *GOOD* syntax: begin epics hac_bcm_average IPM1H04A.XPOS FB_A:use_RF 'RF On'=14 'RF Off'=52 slia:DataValid Yes=1 No=0 hla:DataValid Fred=2 'No'=22 Yes==452 end epics # Examples of *BAD* syntax: begin epics Joe Momma # erroneous space FB_A:use_RF 'RF On=14 # missing a quote slia:DataValid Yes = 1 # erroneous space hla:DataValid Fred= 2 # erroneous space end epics #  # Scalers to appear in the regular event tree ("T") and # new scaler trees, one for each bank. # Banks are "right", "left" (and perhaps 3rd arm). # Access to data: # T>Draw("right_bcm_u1"), right>Draw("bcm_u1") for Rarm # T>Draw("left_bcm_u1"), left>Draw("bcm_u1") for Larm # "default" and "default_helicity" cause a list # of normalization scalers to appear. Examples: # T>Draw("right_trigger_1") from default list # T>Draw("right_P_trigger_1") helicity plus ("P") # T>Draw("right_M_trigger_1") helicity minus ("M") # also right>Draw("trigger_1"), etc. appear. # Implictly the values are RATES (Hz) unless one # specifies the "counts" flag. # "myclk 8 7" is a rate from slot 8 chan 7 and # appears as T>Draw("myclk"). # "myclkc 8 7 counts" are raw counts, slot 8, chan 7. begin scalers right default # this expands to "trigger_1", "bcm_u1", etc... default_helicity # expands to "P_trigger_1", "M_trigger_1" ... mypulser 7 3 myclk 8 7 myclkc 8 7 counts end scalers begin scalers left default default_helicity mypulser 4 3 myclk 4 7 myclkc 4 7 counts end scalers
EXPECTED RESULTS¶
Below is a table showing examples of lines in the output definition file (output.def) and the corresponding expected output.
Behavior of Output  
output.def line  Example How to Access  Comments 

VARIABLE L.s1.lt[4] 
T>Draw("L.s1.lt[4]") 

VARIABLE L.s1.rt 
Tree has L.s1.lt[i], i = 0,1,...Ndata.L.s1.lt 
Array 
BLOCK R.* 
T>Draw("R.s1.lt[0]") , etc. 
All variables in Right (R) HRS 
FORMULA Lt4a 5*L.s1.lt[4] 
T>Draw("Lt4a") 
Scaler Formula 
FORMULA Lt4b 5*L.s1.lt 
Tree has Lt4b[i], i = 0,1,...Ndata.Lt4b 
Vector Formula 
CUT C1 L.s1.lt[4]>1350 
T>Draw("C1") 
Scaler Cut 
CUT C2 L.s1.lt>1350 
Tree has C2[i], i = 0,1,...Ndata.C2 
Vector Cut 
TH1F rs 'Scint' R.s1.lt 100 0 2000 
rs>Draw() 
Histogram (all elements of R.s1.lt[] in 1 histo) 
TH1F rs4 'Scint' R.s1.lt[4] 100 0 2000 R.s1.lt[4]>500 
rs4>Draw() 
Cut R.s1.lt[4]>500 is optional 
TH1F rv 'Scint' R.s1.lt 200 0 2000 R.s1.lt>500 
rv0>Draw(), rv1>Draw, ... rv5>Draw() 
Vector Histogram (size 6) 
Updated by Ole Hansen over 6 years ago · 1 revisions