compareRootHistos Namespace Reference

Functions
def	rec
def	composition
def	comparePaths
def	compareHistos
Variables
list	histos = ['TH1D', 'TH2D', 'TProfile']
string	ser = 'SERIAL'
string	par = 'PARALL'
list	pFile = sys.argv[0]
list	sFile = sys.argv[1]
tuple	tfs = TFile( sFile, 'REC' )
tuple	tfp = TFile( pFile, 'REC' )
tuple	lser = rec(tfs)
tuple	dserial = dict( [(n, o) for n, o in lser] )
tuple	dparall = dict( [(n, o) for n, o in lpar] )
tuple	ts = ( ser, dserial )
tuple	state = comparePaths( ts, tp )

---

Function Documentation

def compareRootHistos::compareHistos	(		t1,
			t2,
			state
	)

Definition at line 121 of file compareRootHistos.py.

00121                                  :
00122   
00123   ( ((serialObjects,serialHistos),(parallObjects, parallHistos)), (uniqueSerPaths,uniqueSerHistos), (uniqueParPaths,uniqueParHistos), mh ) = state
00124  
00125   # deduce which one is parallel, which serial
00126   if   t1[0]  == ser : ds = t1[1] ; dp = t2[1]
00127   elif t2[0]  == ser : ds = t2[1] ; dp = t1[1]
00128   else : print 'Neither tuple is Serial Root file reference?' ; return
00129 
00130   # histocount, objectcount for parallel/serial
00131   hcp = 0 ; pHistos = []
00132   hcs = 0 ; sHistos = []
00133   
00134   omit = ['/stat/Brunel/MemoryTool/Virtual mem, all entries',
00135           '/stat/Brunel/MemoryTool/Virtual mem, downscaled']
00136   omit = []
00137   
00138   # find the histos in the serial file
00139   for k in ds.keys() : 
00140     if k not in omit : 
00141       if ds[k].__class__.__name__ in histos : hcs += 1 ; sHistos.append( k )
00142   # same for parallel
00143   for k in dp.keys() : 
00144     if k not in omit : 
00145       if dp[k].__class__.__name__ in histos : hcp += 1 ; pHistos.append( k )
00146       
00147 
00148   cEntries   = 0 ; xEntries   = 0 ; diffEntries   = []
00149   cIntegrals = 0 ; xIntegrals = 0 ; diffIntegrals = []
00150   passedKol  = 0 ; failedKol  = 0 ; diffKols      = [] ; zeroIntegrals = 0
00151   kTested    = 0
00152   notfound   = 0 ; integralMatch = 0 ; otherTest = 0 ; zeroIntegralMatch = 0
00153   for h in sHistos :
00154     if h in pHistos : 
00155       # matching histos to check
00156       cEntries += 1
00157       sh = ds[h] ; ph = dp[h]
00158       # first check entries
00159       if sh.GetEntries() != ph.GetEntries() : diffEntries.append(h) ; xEntries += 1 ; continue
00160       # check for (non-zero sum of bin error) && (non-zero integrals) for K-Test
00161       sBinError = 0.0 ; pBinError = 0.0
00162       for i in xrange(sh.GetNbinsX()) : sBinError += sh.GetBinError(i)
00163       for i in xrange(ph.GetNbinsX()) : pBinError += ph.GetBinError(i)
00164       sint = sh.Integral() ; pint = ph.Integral()
00165       if (bool(sint) and bool(pint)) and ( sBinError>0 and pBinError>0 ) : 
00166         kTested += 1
00167         kTest = sh.KolmogorovTest(ph)
00168         if int(kTest) : passedKol += 1 
00169         else          : failedKol += 1 ; diffKols.append(h) # ; print 'KTest result : ', kTest        
00170       else : 
00171         # try the integral test?
00172         otherTest += 1
00173         if all((sint, pint)) and (sint==pint) : 
00174           integralMatch += 1
00175         elif (sint==pint) :
00176           zeroIntegralMatch += 1
00177         else : 
00178             diffIntegrals.append( h )
00179             xIntegrals += 1
00180     else :
00181       notfound += 1 ; print 'not found? ', h
00182 
00183   # report on Failed Entry-Checks      
00184   print '\n\n'+'-'*80
00185   print 'Summary of histos with different Entries'
00186   print '-'*80
00187   if diffEntries :
00188     diffEntries.sort()  
00189     for e in diffEntries : print '\t\t\t%s:\t%i != %i'%( e, int(ds[e].GetEntries()), int(dp[e].GetEntries()) )  
00190   print '-'*80
00191   
00192   # report on Failed Kolmogorov Tests    
00193   print '\n\n'+'-'*60
00194   print 'Summary of histos which failed Kolmogorov Test'
00195   print '-'*60
00196   if diffKols :
00197     diffKols.sort()  
00198     for e in diffKols : 
00199       result = ds[e].KolmogorovTest(dp[e])
00200       print '%s\t\t%s :\tK-Test Result :\t %5.16f'%( type(ds[e]), e, result )  
00201   print '-'*60
00202   
00203   # report on Failed Integral Checks
00204   print '\n\n'+'-'*60
00205   print 'Summary of histos which failed Integral Check'
00206   print '-'*60
00207   if diffIntegrals :
00208     diffIntegrals.sort()
00209     for e in diffIntegrals : 
00210       diff = dp[e].Integral()-ds[e].Integral()
00211       pc   = (diff*100)/ds[e].Integral()
00212       print '%s\t\t%s:\t Diff = %5.6f\tPercent Diff to Serial : %5.6f '%( type(ds[e]), e, diff, pc  )
00213   print '-'*60 + '\n' 
00214   print '='*80 + '\n'
00215   
00216   print '\n' + '='*80
00217   print 'Comparison : Serial/Parallel ROOT Histo files'
00218   print '\n\t\tSerial\tParall'
00219   print '\tObjects : %i\t%i\t\t( p-s = %i )'%( serialObjects, parallObjects, parallObjects-serialObjects )  
00220   print '\tHistos  : %i\t%i\t\t( p-s = %i )'%( serialHistos,  parallHistos,  parallHistos-serialHistos  )
00221   print '\t          __________'
00222   print '\tTotal   : %i\t%i\n'%( serialHistos+serialObjects,  parallHistos+parallObjects  )  
00223   print 'Objects/Histos unique to Serial File : %i / %i'%( len(uniqueSerPaths)-uniqueSerHistos, uniqueSerHistos )
00224   print 'Objects/Histos unique to Parall File : %i / %i'%( len(uniqueParPaths)-uniqueParHistos, uniqueParHistos )
00225   print '\nMatching Histograms valid for Comparison : %i'%( mh )
00226   print '\nOmissions : '
00227   for entry in omit : print '\t%s'%( entry )   
00228   print '\nHistograms for Comparison (after Omissions) : %i'%( mh-len(omit) )  
00229   print '\n\tHISTOGRAM TESTS : '
00230   print '\t\tKOLMOGOROV TEST      : %i'%( kTested )
00231   print '\t\tINTEGRAL TEST        : %i'%( otherTest )
00232   print '\t\tENTRIES TEST         : %i'%( xEntries )
00233   print '\t\t                       ____'
00234   print '\t\tTested               : %i'%( cEntries )  
00235 
00236   print '\n\tDISCREPANCIES : '
00237   print '\t\tK-Test      : %i'%( failedKol  )
00238   print '\t\tIntegrals   : %i'%( xIntegrals )    
00239   print '\t\tEntries     : %i'%( xEntries   )  
00240   print '\n'+'='*80  
00241 
00242 # =================================================================================================
00243 
if __name__ == '__main__' : 

def compareRootHistos::comparePaths	(		t1,
			t2
	)

Definition at line 64 of file compareRootHistos.py.

                           : 
  if   t1[0]  == ser : ds = t1[1] ; dp = t2[1]
  elif t2[0]  == ser : ds = t2[1] ; dp = t1[1]
  else : print 'Neither tuple is Serial Root file reference?' ; return
  
  dsks = ds.keys() ; dpks = dp.keys()
  dsks.sort()      ; dpks.sort()
  
  sset = sets.Set( dsks )
  pset = sets.Set( dpks )
  os, hs = composition( (ser, ds) )
  op, hp = composition( (par, dp) )
  print '\n' + '='*80
  print 'Comparison of Paths : Serial vs Parallel ROOT files'
  print '-'*80
  print 'Number of paths in Serial file : %i (objects, histos) = ( %i, %i )'%( len(dsks), os, hs )
  print 'Number of paths in Parall file : %i (objects, histos) = ( %i, %i )'%( len(dpks), op, hp )
  matching = sset.intersection(pset)
  matchingHistos = 0
  for n in matching :
    if ds[n].__class__.__name__ in histos : matchingHistos += 1    
  print '\nMatching paths                 : %i'%( len(matching) )
  uSer = sset - pset
  # work out histos unique to parallel file
  uniqueSerialHistos = 0
  for n in uSer :
    if ds[n].__class__.__name__ in histos : uniqueSerialHistos += 1
  print 'Paths unique to Serial file : %i ( %i Histos )'%( len(uSer), uniqueSerialHistos )
  if uSer : 
    for n in uSer : print '\t%s : \t%s'%( ds[n], n )
  uPar = pset - sset
  uniqueParallHistos = 0  
  for n in uPar :
    if dp[n].__class__.__name__ in histos : uniqueParallHistos += 1  
  print 'Paths unique to Parall file : %i ( %i Histos )'%( len(uPar), uniqueParallHistos )
  if uPar : 
    for n in uPar : print '\t%s : \t%s'%( dp[n], n )
  print 'Matching Histos to test : %i'%( matchingHistos )
  print '='*80 + '\n'
  return ( ((os,hs),(op,hp)), (uSer, uniqueSerialHistos), (uPar, uniqueParallHistos), matchingHistos )
# =================================================================================================  

# =================================================================================================
# Method        : compareHistos( t1, t2 )
# 
# @param t1, t2 : a tuple of ( type, d ) where type is either 'SERIAL' or 'PARALL'
#                 and d is a dictionary of ROOT objects, with each key = ROOT path 
# 
# function      : compare the histograms in Serial/Parallel ROOT files.  First, go through each 
#                 dict to collect the histos (ignore TDirectory objects, etc).  Then the histos
#                 in the parallel file (experimental) are compared to their equivalents in the 
#                 serial file (definitely correct) using 3 methods.
#                 1) The entries are checked, they should be equal
#                 2) If entries are equal, check the Integral(); should be equal
#                 3) If integrals are equal, check the KolmogorovTest() ; should be 1
#                 Arguments t1 and t2 are checked and the parallel/serial auto-detected
# 

def compareRootHistos::composition

(

t

)

Definition at line 42 of file compareRootHistos.py.

00042                      :
00043   typ, d = t 
00044   hists = 0 ; objs = 0
00045   for k in d.keys() : 
00046     if d[k].__class__.__name__ in histos : hists += 1
00047     else                                 : objs  += 1
00048   return objs, hists
00049 # =================================================================================================
00050   
00051 # =================================================================================================
00052 # Method        : comparePaths( t1, t2 )
00053 # 
00054 # @param t1, t2 : a tuple of ( type, d ) where type is either 'SERIAL' or 'PARALL'
00055 #                 and d is a dictionary of ROOT objects, with each key = ROOT path 
00056 # 
00057 # function      : compare the paths between the two histo files.  If the files are identical, they
00058 #                 should have the same set of paths.  The Parallel file should definitely have the
00059 #                 same paths as the Serial.  Perhaps the Serial file will have some more paths due
00060 #                 to extra histos added as part of Application Sequencer finalisation
00061 #                 Arguments t1 and t2 are checked and the parallel/serial auto-detected
00062 #                 Uses sets module for intersections/unions, etc.
00063 # 
def comparePaths( t1, t2 ) : 

def compareRootHistos::rec	(		o,
			path = None,
			lst = None
	)

Definition at line 20 of file compareRootHistos.py.

00020                                   :
00021   if not path : path = '/stat' ; lst = []
00022   else        : path = path + '/' + o.GetName()
00023   lst.append( (path,o) )
00024   if 'GetListOfKeys' in dir(o) : 
00025     keys = o.GetListOfKeys()
00026     for k in keys :
00027       name =  k.GetName()
00028       rec( o.Get(name), path, lst ) 
00029   else :
00030     pass
00031   return lst
00032 # =================================================================================================
00033 
00034 # =================================================================================================
00035 # Method   : composition( t )
00036 # 
00037 # @param t : a tuple of ( type, d ) where type is either 'SERIAL' or 'PARALL'
00038 #            and d is a dictionary of ROOT objects, with each key = ROOT path 
00039 # 
00040 # function : deduce the composition, (objects/histos) counts
00041 #
def composition( t ) :

---

Variable Documentation

tuple compareRootHistos::dparall = dict( [(n, o) for n, o in lpar] )

Definition at line 261 of file compareRootHistos.py.

tuple compareRootHistos::dserial = dict( [(n, o) for n, o in lser] )

Definition at line 260 of file compareRootHistos.py.

list compareRootHistos::histos = ['TH1D', 'TH2D', 'TProfile']

Definition at line 5 of file compareRootHistos.py.

tuple compareRootHistos::lser = rec(tfs)

Definition at line 258 of file compareRootHistos.py.

string compareRootHistos::par = 'PARALL'

Definition at line 7 of file compareRootHistos.py.

list compareRootHistos::pFile = sys.argv[0]

Definition at line 247 of file compareRootHistos.py.

string compareRootHistos::ser = 'SERIAL'

Definition at line 6 of file compareRootHistos.py.

list compareRootHistos::sFile = sys.argv[1]

Definition at line 248 of file compareRootHistos.py.

tuple compareRootHistos::state = comparePaths( ts, tp )

Definition at line 269 of file compareRootHistos.py.

tuple compareRootHistos::tfp = TFile( pFile, 'REC' )

Definition at line 255 of file compareRootHistos.py.

tuple compareRootHistos::tfs = TFile( sFile, 'REC' )

Definition at line 254 of file compareRootHistos.py.

tuple compareRootHistos::ts = ( ser, dserial )

Definition at line 263 of file compareRootHistos.py.