def getPrettyName ( self, anaid ):
""" get pretty name of ana id """
if False: ## set to true and we have the old analysis Ids
return anaid
if not hasattr ( self, "database" ):
from smodels.experiment.databaseObj import Database
dbname = "./original.pcl"
dbname = "/home/walten/git/smodels-database"
dbname = "/scratch-cbe/users/wolfgan.waltenberger/rundir/db31.pcl"
self.database = Database ( dbname )
from smodels_utils.helper.prettyDescriptions import prettyTexAnalysisName
if ":" in anaid:
anaid = anaid[:anaid.find(":")]
ers = self.database.getExpResults ( analysisIDs = [ anaid ] )
for er in ers:
if hasattr ( er.globalInfo, "prettyName" ):
pn = er.globalInfo.prettyName
sqrts = er.globalInfo.sqrts.asNumber(TeV)
ret = prettyTexAnalysisName ( pn, sqrts, dropEtmiss = True,
collaboration = True, anaid = er.globalInfo.id )
# for the 2020 paper to be consistent
ret = ret.replace( "+ top tag", "stop" )
ret = ret.replace( "+ 4 (1 b-)jets", "multijet" )
# ret += " -> " + anaid
return ret
# print ( "found no pretty name", ers[0].globalInfo )
return anaid
def checkOneAnalysis():
import argparse
#import IPython
argparser = argparse.ArgumentParser(
description='print the correlations of one specific analysis')
argparser.add_argument(
'-d',
'--dbpath',
help='specify path to database [<rundir>/database.pcl]',
type=str,
default="<rundir>/database.pcl")
argparser.add_argument('-a',
'--analysis',
help='print for <analysis>',
type=str,
default="CMS-SUS-19-006")
args = argparser.parse_args()
from smodels.experiment.databaseObj import Database
print("[analysisCombiner] checking %s" % args.dbpath)
db = Database(args.dbpath)
results = db.getExpResults()
info = getInfoFromAnaId(args.analysis, results)
sqrts = info.sqrts
collaboration = getExperimentName(info)
prettyName = info.prettyName
if args.analysis in moreComments:
prettyName += " (%s)" % moreComments[args.analysis]
# IPython.embed()
print("correlations for %s: %s" % (args.analysis, prettyName))
combs, nocombs = set(), set()
pnames = {}
for er in results:
if er.globalInfo.sqrts != sqrts:
continue
if getExperimentName(er.globalInfo) != collaboration:
continue
Id = er.globalInfo.id
Id = Id.replace("-eff", "").replace("-agg", "")
if Id == "CMS-SUS-19-006-2":
Id = "CMS-SUS-19-006"
if Id == args.analysis:
continue
pname = er.globalInfo.prettyName
if Id in moreComments:
pname += " (%s)" % moreComments[Id]
pnames[Id] = pname
cc = canCombine(info, er.globalInfo, "aggressive")
# cc = canCombine ( pred, er.globalInfo )
if cc:
combs.add(Id)
else:
nocombs.add(Id)
print("can combine with: ")
for Id in combs:
pname = pnames[Id]
print(" `- %s: %s" % (Id, pname))
print("cannot combine with: ")
for Id in nocombs:
pname = pnames[Id]
print(" `- %s: %s" % (Id, pname))
def collect():
db = Database ( "./database.pcl" ) # , force_load = "txt" )
ers = db.getExpResults ( dataTypes = [ "upperLimit" ], onlyWithExpected=True )
allSs = []
for er in ers:
txnlist = er.datasets[0].txnameList
for txn in txnlist:
ct=0
origdata = eval(txn.txnameData.origdata)
for point in origdata:
m = point[0]
rul = point[1]
ul,eul=None,None
try:
ul = txn.getULFor(m, False )
eul = txn.getULFor(m, True )
except Exception:
pass
if type(ul) == type(None) or type(eul) == type(None):
continue
sigma = eul / 1.96
S = float ( ( ul - eul ) / sigma )
if (S < -1.8 or S > 3.5) and ct<3:
# if S > 10. and ct<3:
print ( )
print ( "S=%.2f for ul=%s, eul=%s sigma=%s" % ( S, ul, eul, sigma ) )
print ( " at ", er.globalInfo.id, txn.txName, m, "rul", rul )
ct += 1
allSs.append ( S )
# print ( "->", er.globalInfo.id, txn, S )
print ("all", min(allSs), np.mean(allSs), max(allSs) )
f=open("ulSs.pcl","wb")
pickle.dump(allSs,f)
f.close()
sys.exit()
def __init__(self,
walkerid,
dbpath="./default.pcl",
expected=False,
select="all",
do_combine=False):
"""
:param do_combine: if True, then also use combined results,
both via simplified likelihoods and pyhf.
"""
self.walkerid = walkerid
self.do_combine = do_combine
self.modifier = None
self.select = select
self.expected = expected
self.rthreshold = 1.3 ## threshold for rmax
if expected:
from expResModifier import ExpResModifier
self.modifier = ExpResModifier()
force_load = None
if dbpath.endswith(".pcl"):
force_load = "pcl"
ntries = 0
while not os.path.exists(dbpath):
## give it a few tries
ntries += 1
time.sleep(ntries * 5)
if ntries > 5:
break
self.database = Database(dbpath, force_load=force_load)
self.fetchResults()
self.combiner = Combiner(self.walkerid)
def testWritePickle(self):
""" tests writing pickle file """
binfile = "./.database.pcl"
if os.path.exists ( binfile ):
os.unlink ( binfile )
self.logger.info ( "test writing pickle file """ )
writer = Database ( "./tinydb/", force_load = "txt" )
writer.createBinaryFile ( binfile )
reader = Database ( binfile, force_load="pcl" )
os.unlink ( binfile )
self.assertEqual( writer, reader )
def getSRs():
from smodels.experiment.databaseObj import Database
db = Database ( "official" )
ers = db.getExpResults( dataTypes=[ "efficiencyMap" ] )
stats = []
for er in ers:
for ds in er.datasets:
D = { "obsN": ds.dataInfo.observedN, "expectedBG": ds.dataInfo.expectedBG,
"bgError": ds.dataInfo.bgError, "upperLimit": ds.dataInfo.upperLimit,
"expectedUpperLimit": ds.dataInfo.expectedUpperLimit }
stats.append ( D )
return stats
def __init__(self, inputfile, rundir, verbose="info"):
self.inputfile = inputfile
self.rundir = rundir
self.nr = 0
self.verbstring = verbose
if type(verbose) == int:
self.verbose = verbose
else:
verbose = verbose.lower()
verbs = {"err": 10, "warn": 20, "info": 30, "debug": 40}
self.verbose = 50
for k, v in verbs.items():
if k in verbose:
self.verbose = v
self.database = Database(self.inputfile)
def testFallBack(self):
#Test if smodels/experiment/defaultFinalStates.py is being used if databaseParticle.py is missing
dbpath = "./database_simple"
dbOld = Database(dbpath, discard_zeroes=False, force_load='txt')
model = dbOld.databaseParticles
self.assertEqual(model.label,
'DB Final States (default)') #Simple fallback test
def initialize( self ):
self.setStatus ( "initialized" )
Cache.n_stored = 20000 ## crank up the caching
self.db = Database ( self.dbpath )
self.expResults = self.db.expResultList
self.sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
self.server_address = ( self.servername, self.port )
self.pprint ( 'starting up on %s port %s' % self.server_address )
self.pprint ( 'I will be serving database %s at %s' % \
(self.db.databaseVersion, self.dbpath ) )
try:
self.sock.bind( self.server_address )
except OSError as e:
self.pprint ( "exception %s. is host ''%s'' reachable?" % \
( e, self.server_address ) )
sys.exit(-1)
# Listen for incoming connections
self.sock.listen(1)
atexit.register ( shutdownAll )
while True:
# Wait for a connection
self.setStatus ( "waiting" )
self.log ( 'waiting for a connection' )
self.connection, self.client_address = self.sock.accept()
self.listen()
def loadDatabase(parser, db):
"""
Load database
:parameter parser: ConfigParser with path to database
:parameter db: binary database object. If None, then database is loaded,
according to databasePath. If True, then database is loaded,
and text mode is forced.
:returns: database object, database version
"""
try:
databasePath = parser.get("path", "databasePath")
if databasePath == "micromegas":
databasePath = installDirectory() + "/smodels-database/"
database = db
if database in [None, True]:
force_load = None
if database == True: force_load = "txt"
database = Database(databasePath, force_load=force_load)
databaseVersion = database.databaseVersion
except DatabaseNotFoundException:
logger.error("Database not found in %s" %
os.path.realpath(databasePath))
sys.exit()
return database, databaseVersion
def testCompare(self):
warnings.simplefilter("ignore", ResourceWarning)
from simplyGluino_default import smodelsOutputDefault
filename = "./testFiles/slha/simplyGluino.slha"
port = random.choice(range(31700, 42000))
# port = 31744
dbfile = "database/db30.pcl"
dbfile = "unittest"
startserver = f"../smodels/tools/smodelsTools.py proxydb -p {port} -i {dbfile} -o ./proxy.pcl -r -v error"
cmd = startserver.split(" ")
# print ( "starting server %s" % startserver )
myenv = os.environ.copy()
pp = ""
if "PYTHONPATH" in myenv:
pp = ":" + myenv["PYTHONPATH"]
myenv["PYTHONPATH"] = "../" + pp
subprocess.Popen(cmd, env=myenv)
time.sleep(3)
db = Database("./proxy.pcl")
outputfile = runMain(filename,
suppressStdout=True,
overridedatabase=db)
smodelsOutput = importModule(outputfile)
client = DatabaseClient(port=port, verbose="warn")
client.send_shutdown()
ignoreFields = [
'input file', 'smodels version', 'ncpus', 'Element',
'database version', 'Total missed xsec', 'Missed xsec long-lived',
'Missed xsec displaced', 'Missed xsec MET',
'Total outside grid xsec',
'Total xsec for missing topologies (fb)',
'Total xsec for missing topologies with displaced decays (fb)',
'Total xsec for missing topologies with prompt decays (fb)',
'Total xsec for topologies outside the grid (fb)'
]
smodelsOutputDefault['ExptRes'] = sorted(
smodelsOutputDefault['ExptRes'],
key=lambda res: res['r'],
reverse=True)
equals = equalObjs(smodelsOutput,
smodelsOutputDefault,
allowedDiff=0.08,
ignore=ignoreFields,
fname=outputfile)
if not equals:
e = "simplyGluino.py != simplyGluino_default.py"
logger.error(e)
# raise AssertionError( e )
self.assertTrue(equals)
self.removeOutputs(outputfile)
def getDatabaseVersion(protomodel, dbpath="default.pcl"):
dbver = "???"
if hasattr(protomodel, "dbversion"):
dbver = protomodel.dbversion
if not "???" in dbver:
return dbver
if os.path.exists(dbpath):
## try to get db version from db file
from smodels.experiment.databaseObj import Database
db = Database(dbpath)
dbver = db.databaseVersion
return dbver
def getSummary():
from smodels.experiment.databaseObj import Database
# dbpath = "official"
dbpath = "<rundir>/database.pcl"
print("[analysisCombiner] checking %s" % dbpath)
db = Database(dbpath)
results = db.getExpResults()
strategy = "aggressive"
ana1 = "CMS-SUS-16-042"
ana2 = "CMS-SUS-16-033"
canC = canCombine(ana1, ana2, strategy, results)
print("[analysisCombiner] can combine %s with %s: %s" %
(ana1, ana2, str(canC)))
ctr, combinable = 0, 0
for x, e in enumerate(results):
for y, f in enumerate(results):
if y <= x:
continue
ctr += 1
isUn = canCombine(e.globalInfo, f.globalInfo, strategy)
combinable += isUn
print("[analysisCombiner] can combine %d/%d pairs of results" %
(combinable, ctr))
def __init__(self, database, force_txt = False ):
"""
:param database: Path to the database or Database object
"""
self._selectedExpResults = []
load = None
if force_txt == True:
load = "txt"
if isinstance(database,str):
self.database = Database(database, load )
elif isinstance(database,Database):
self.database = database
else:
logger.error("The input must be the database location or a Database object.")
raise SModelSError()
self.loadAllResults()
def loadDatabase(parser, db):
"""
Load database
:parameter parser: ConfigParser with path to database
:parameter db: binary database object. If None, then database is loaded,
according to databasePath. If True, then database is loaded,
and text mode is forced.
:returns: database object, database version
"""
try:
dp = parser.get("path", "databasePath")
logger.error("``[path] databasePath'' in ini file is deprecated; " \
"use ``[database] path'' instead.(See e.g. smodels/etc/parameters_default.ini)")
parser.set("database", "path", dp)
except (NoSectionError, NoOptionError) as e:
## path.databasePath not set. This is good.
pass
try:
database = db
# logger.error("database=db: %s" % database)
if database in [None, True]:
databasePath = parser.get("database", "path")
checkForSemicolon(databasePath, "database", "path")
discard_zeroes = True
try:
discard_zeroes = parser.getboolean("database", "discardZeroes")
except (NoSectionError, NoOptionError) as e:
logger.debug(
"database:discardZeroes is not given in config file. Defaulting to 'True'."
)
force_load = None
if database == True: force_load = "txt"
if os.path.isfile(databasePath):
force_load = "pcl"
database = Database(databasePath, force_load=force_load, \
discard_zeroes = discard_zeroes)
databaseVersion = database.databaseVersion
except DatabaseNotFoundException:
logger.error("Database not found in ``%s''" %
os.path.realpath(databasePath))
sys.exit()
return database, databaseVersion
def __init__(self, database, force_txt = False ):
"""
:ivar _selectedExpResults: list of experimental results loaded in the browser.
Can be used to hold a subset of results in the database.
By default all results are loaded.
:param database: Path to the database or Database object
"""
self._selectedExpResults = []
load = None
if force_txt == True:
load = "txt"
if isinstance(database,str):
if database.endswith(".pcl"):
load = "pcl"
self.database = Database(database, load )
elif isinstance(database,Database):
self.database = database
else:
logger.error("The input must be the database location or a Database object.")
raise SModelSError()
self.loadAllResults()
#!/usr/bin/env python3
"""
.. module:: databaseLoader
:synopsis: When running the complete test suite, we need to
load the database only once
.. moduleauthor:: Wolfgang Waltenberger <*****@*****.**>
"""
import sys
sys.path.insert(0, "../")
from smodels.experiment.databaseObj import Database
# dbpath = "./database"
# dbpath = "../../smodels-database"
dbpath = "unittest"
database = Database(dbpath, discard_zeroes=False)
if __name__ == "__main__":
print(database)
class LlhdPlot:
""" A simple class to make debugging the plots easier """
def __init__ ( self, pid1, pid2, verbose, copy, max_anas,
interactive, drawtimestamp, compress, rundir,
upload ):
"""
:param pid1: pid for x axis, possibly a range of pids
:param pid2: pid for y axis
:param verbose: verbosity (debug, info, warn, or error)
:param copy: copy plot to ../../smodels.github.io/protomodels/latest
:param max_anas: maximum number of analyses on summary plot
:param interactive: prepare for an interactive session?
:param drawtimestamp: if true, put a timestamp on plot
:param compress: prepare for compression
:param upload: upload directory, default is "latest"
"""
self.rundir = rundir
self.upload = upload
self.setup( pid1, pid2 )
self.DEBUG, self.INFO = 40, 30
self.drawtimestamp = drawtimestamp
self.max_anas = max_anas ## maximum number of analyses
self.copy = copy
self.rthreshold = 1.7
self.interactive = interactive
self.hiscorefile = "./hiscore.hi"
if rundir != None:
self.hiscorefile = f"{rundir}/hiscore.hi"
self.setVerbosity ( verbose )
masspoints,mx,my,nevents,topo,timestamp = self.loadPickleFile( compress )
self.masspoints = masspoints
self.mx = mx
self.my = my
self.nevents = nevents
self.topo = topo
self.timestamp = timestamp
self.massdict = {}
self.rdict = {}
if masspoints == None:
return
for m in masspoints:
self.massdict[ (m[0],m[1]) ] = m[2]
if len(m)>3:
self.rdict[ (m[0],m[1]) ] = m[3]
def setVerbosity ( self, verbose ):
self.verbose = verbose
if type(verbose)==str:
verbose = verbose.lower()
if "deb" in verbose:
self.verbose = 40
return
if "inf" in verbose:
self.verbose = 30
return
if "warn" in verbose:
self.verbose = 20
return
if "err" in verbose:
self.verbose = 10
return
self.pprint ( "I dont understand verbosity ``%s''. Setting to debug." % verbose )
self.verbose = 40
def getHash ( self, m1=None, m2=None ):
""" get hash for point. if None, get hash for self.mx, self.my """
if m1 == None:
m1 = self.mx
if m2 == None:
m2 = self.my
return int(1e3*m1) + int(1e0*m2)
def getResultFor ( self, ana, masspoint ):
""" return result for ana/topo pair
:param ana: the analysis id. optionally a data type can be specificed, e.g.
as :em. Alternatively, a signal region can be specified.
:param masspoint: a point from self.masspoints
:returns: results for this analysis (possibly data type, possibly signal region)
and topology
"""
#self.pprint ( "asking for %s" % ana )
ret,sr = None, None
dType = "any"
if ":" in ana:
ana,dType = ana.split(":")
for k,v in masspoint.items():
tokens = k.split(":")
if dType == "ul" and tokens[1] != "None":
continue
if dType == "em" and tokens[1] == "None":
continue
if ana != tokens[0]:
continue
# self.pprint ( "asking for %s, %s %s" % ( tokens[0], tokens[1], dType ) )
if tokens[1] != None and dType not in [ "any", "ul", "None" ]:
# if signal regions are given, they need to match
if tokens[1] != dType:
continue
self.debug ( "found a match for", tokens[0], tokens[1], v )
if self.topo not in tokens[2]:
continue
if ret == None or v > ret:
ret = v
sr = tokens[1]
return ret,sr
def loadPickleFile ( self, returnAll=False ):
""" load dictionary from picklefile
:param returnAll: return all likelihoods info
"""
topo, timestamp = "?", "?"
allhds = None
with open ( self.picklefile, "rb" ) as f:
try:
allhds = pickle.load ( f )
mx = pickle.load ( f )
my = pickle.load ( f )
nevents = pickle.load ( f )
topo = pickle.load ( f )
timestamp = pickle.load ( f )
except EOFError as e:
print ( "[plotLlhds] EOF error %s, when reading %s" % \
( e, self.picklefile ) )
f.close()
if allhds == None:
print ( "couldnt read llhds in %s" % self.picklefile )
return None,None,None,None,None,None
if returnAll:
return allhds,mx,my,nevents,topo,timestamp
llhds=[]
mu = 1.
def getMu1 ( L ):
for k,v in L.items():
if abs(k-mu)<1e-9:
return v
print ( "couldnt find anything" )
return None
for llhd in allhds:
if self.pid1 in [ 1000001, 1000002, 1000003, 1000004 ]:
if llhd[0]<310.:
print ( "light squark mass wall, skipping mx %d < 310 GeV" % llhd[0] )
continue
if len(llhd)==4:
llhds.append ( (llhd[0],llhd[1],getMu1(llhd[2]),llhd[3]) )
else:
llhds.append ( (llhd[0],llhd[1],getMu1(llhd[2]),[0.,0.,0.]) )
return llhds,mx,my,nevents,topo,timestamp
def pprint ( self, *args ):
print ( "[plotLlhds] %s" % " ".join(map(str,args)) )
def debug ( self, *args ):
if self.verbose >= self.DEBUG:
print ( "[plotLlhds] %s" % " ".join(map(str,args)) )
def setup ( self, pid1, pid2 ):
""" setup rundir, picklefile path and hiscore file path """
self.hiscorefile = self.rundir + "/hiscore.hi"
if not os.path.exists ( self.hiscorefile ):
self.pprint ( "could not find hiscore file %s" % self.hiscorefile )
self.pid1 = pid1
self.pid2 = pid2
if type(self.pid1) in [ tuple, list ]:
pid1 = self.pid1[0]
self.picklefile = "%s/llhd%d%d.pcl" % ( self.rundir, pid1, self.pid2 )
if not os.path.exists ( self.picklefile ):
llhdp = self.picklefile
self.picklefile = "%s/mp%d%d.pcl" % ( self.rundir, pid1, self.pid2 )
if not os.path.exists ( self.picklefile ):
self.pprint ( "could not find pickle files %s and %s" % \
( llhdp, self.picklefile ) )
def describe ( self ):
""" describe the situation """
print ( "%d masspoints obtained from %s, hiscore stored in %s" % \
( len ( self.masspoints), self.picklefile, self.hiscorefile ) )
print ( "Data members: plot.masspoints, plot.massdict, plot.timestamp, plot.mx, plot.my" )
print ( " plot.pid1, plot.pid2, plot.topo" )
print ( "Function members: plot.findClosestPoint()" )
def getLClosestTo ( self, L, mx=None, my=None ):
""" get the L closest to your point """
if mx == None:
mx=self.mx
if my == None:
my=self.my
def distance_ ( k, mx, my ):
_x = int(math.floor(k/1000.))
_y = int(math.floor(k % 1000 ) )
ret= (mx - _x)**2 + (my - _y)**2
return ret
dmmin, vmin = float("inf"), 23.
for k,v in L.items():
dm = distance_ ( k, mx, my )
if dm < dmmin and not np.isnan(v):
dmmin = dmmin
vmin = v
return vmin
def getPrettyName ( self, anaid ):
""" get pretty name of ana id """
if False: ## set to true and we have the old analysis Ids
return anaid
if not hasattr ( self, "database" ):
from smodels.experiment.databaseObj import Database
dbname = "./original.pcl"
dbname = "/home/walten/git/smodels-database"
dbname = "/scratch-cbe/users/wolfgan.waltenberger/rundir/db31.pcl"
self.database = Database ( dbname )
from smodels_utils.helper.prettyDescriptions import prettyTexAnalysisName
if ":" in anaid:
anaid = anaid[:anaid.find(":")]
ers = self.database.getExpResults ( analysisIDs = [ anaid ] )
for er in ers:
if hasattr ( er.globalInfo, "prettyName" ):
pn = er.globalInfo.prettyName
sqrts = er.globalInfo.sqrts.asNumber(TeV)
ret = prettyTexAnalysisName ( pn, sqrts, dropEtmiss = True,
collaboration = True, anaid = er.globalInfo.id )
# for the 2020 paper to be consistent
ret = ret.replace( "+ top tag", "stop" )
ret = ret.replace( "+ 4 (1 b-)jets", "multijet" )
# ret += " -> " + anaid
return ret
# print ( "found no pretty name", ers[0].globalInfo )
return anaid
def plot ( self, ulSeparately=True, pid1=None ):
""" a summary plot, overlaying all contributing analyses
:param ulSeparately: if true, then plot UL results on their own
"""
if pid1 == None and type(self.pid1) in [ list, tuple ]:
for p in self.pid1:
self.plot ( ulSeparately, p )
return
if type(pid1) in [ tuple, list ]:
for p in pid1:
self.plot ( ulSeparately, p )
return
if pid1 == None:
pid1 = self.pid1
self.pprint ( "plotting summary for %s, %s" % ( pid1, self.topo ) )
resultsForPIDs = {}
from plotting.plotHiscore import getPIDsOfTPred, obtain
protomodel = obtain ( 0, self.hiscorefile )
for tpred in protomodel.bestCombo:
resultsForPIDs = getPIDsOfTPred ( tpred, resultsForPIDs, integrateSRs=False )
stats = self.getAnaStats( integrateSRs=False )
if stats == None:
self.pprint ( "found no ana stats?" )
return
anas = list(stats.keys())
if pid1 in resultsForPIDs:
self.debug ( "results for PIDs %s" % ", ".join ( resultsForPIDs[pid1] ) )
anas = list ( resultsForPIDs[pid1] )
anas.sort()
self.pprint ( "summary plot: %s" % ", ".join ( anas ) )
# print ( stats.keys() )
colors = [ "red", "green", "blue", "orange", "cyan", "magenta", "grey", "brown",
"pink", "indigo", "olive", "orchid", "darkseagreen", "teal" ]
xmin,xmax,ymin,ymax=9000,0,9000,0
for m in self.masspoints:
if m[0] < xmin:
xmin = m[0]
if m[0] > xmax:
xmax = m[0]
if m[1] < ymin:
ymin = m[1]
if m[1] > ymax:
ymax = m[1]
if abs(xmin-310.)<1e-5:
xmin=330. ## cut off the left margin
print ( "[plotLlhds] range x [%d,%d] y [%d,%d]" % ( xmin, xmax, ymin, ymax ) )
handles = []
existingPoints = []
combL = {}
namer = SParticleNames ( susy = False )
for ctr,ana in enumerate ( anas ): ## loop over the analyses
if ctr >= self.max_anas:
self.pprint ( "too many (%d > %d) analyses." % (len(anas),self.max_anas) )
for ana in anas[ctr:]:
self.pprint ( " - skipping %s" % ana )
break
color = colors[ctr]
x,y=set(),set()
L, R = {}, {}
minXY=( 0.,0., float("inf") )
s=""
r,sr = self.getResultFor ( ana, self.masspoints[0][2] )
if r:
s="(%.2f)" % (-np.log(r))
print ( "[plotLlhds] result for", ana,"is", s )
cresults = 0
for cm,masspoint in enumerate(self.masspoints[1:]):
#if cm % 10 != 0:
# continue
if cm % 1000 == 0:
print ( ".", end="", flush=True )
m1,m2,llhds,robs=masspoint[0],masspoint[1],masspoint[2],masspoint[3]
rmax=float("nan")
if len(robs)>0:
rmax=robs[0]
if m2 > m1:
print ( "m2,m1 mass inversion?",m1,m2 )
x.add ( m1 )
y.add ( m2 )
zt = float("nan")
result,sr = self.getResultFor ( ana, llhds )
if result:
zt = - np.log( result )
cresults += 1
if zt < minXY[2] and rmax<=self.rthreshold:
minXY=(m1,m2,zt)
h = self.getHash(m1,m2)
L[h]=zt
if not h in combL:
combL[h]=0.
if np.isnan(zt):
combL[h] = combL[h] + 100.
else:
combL[h] = combL[h] + zt
R[h]=rmax
print ()
# print ( "\n[plotLlhds] min(xy) for %s is at m=(%d/%d): %.2f(%.2g)" % ( ana, minXY[0], minXY[1], minXY[2], np.exp(-minXY[2] ) ) )
if cresults == 0:
print ( "[plotLlhds] warning: found no results for %s. skip" % \
str(masspoint) )
continue
# return
x.add ( xmax*1.03 )
x.add ( xmin*.93 )
y.add ( ymax+50. )
y.add ( 0. )
x,y=list(x),list(y)
x.sort(); y.sort()
X, Y = np.meshgrid ( x, y )
Z = float("nan")*X
RMAX = float("nan")*X
for irow,row in enumerate(Z):
for icol,col in enumerate(row):
h = 0
if len(x)>= icol and len(y) >= irow:
h = self.getHash(list(x)[icol],list(y)[irow])
if h in L:
Z[irow,icol]=L[h]
if h in R:
RMAX[irow,icol]=R[h]
if self.interactive:
self.RMAX = RMAX
# self.ZCOMB = ZCOMB
self.Z = Z
self.L = L
self.R = R
self.X = X
self.Y = Y
hldZ100 = computeHPD ( Z, None, 1., False, rthreshold=self.rthreshold )
cont100 = plt.contour ( X, Y, hldZ100, levels=[0.25], colors = [ color ], linestyles = [ "dotted" ], zorder=10 )
#hldZ95 = computeHPD ( Z, .95, False )
#cont95 = plt.contour ( X, Y, hldZ95, levels=[0.5], colors = [ color ], linestyles = [ "dashed" ] )
#plt.clabel ( cont95, fmt="95%.0s" )
hldZ50 = computeHPD ( Z, RMAX, .68, False, rthreshold=self.rthreshold )
cont50c = plt.contour ( X, Y, hldZ50, levels=[1.0], colors = [ color ], zorder=10 )
cont50 = plt.contourf ( X, Y, hldZ50, levels=[1.,10.], colors = [ color, color ], alpha=getAlpha( color ), zorder=10 )
plt.clabel ( cont50c, fmt="68%.0s" )
if hasattr ( cont50, "axes" ):
ax = cont50.axes
else:
ax = cont50.ax
while isCloseToExisting ( minXY, existingPoints ):
minXY = ( minXY[0]+8., minXY[1]+8., minXY[2] )
a = ax.scatter( [ minXY[0] ], [ minXY[1] ], marker="*", s=180, color="black", zorder=20 )
anan = ana.replace(":None",":UL") # + " (%.2f)" % (minXY[2])
label = self.getPrettyName ( ana )
a = ax.scatter( [ minXY[0] ], [ minXY[1] ], marker="*", s=110, color=color,
label=label, alpha=1., zorder=20 )
existingPoints.append ( minXY )
handles.append ( a )
ZCOMB = float("nan")*X
for irow,row in enumerate(Z):
for icol,col in enumerate(row):
h = 0
if len(x)> icol and len(y) > irow:
h = self.getHash(list(x)[icol],list(y)[irow])
if h in combL and not np.isnan(combL[h]):
ZCOMB[irow,icol]=combL[h]
if combL[h]==0.:
ZCOMB[irow,icol]=float("nan")
self.ZCOMB = ZCOMB
contRMAX = plt.contour ( X, Y, RMAX, levels=[self.rthreshold], colors = [ "gray" ], zorder=10 )
contRMAXf = plt.contourf ( X, Y, RMAX, levels=[self.rthreshold,float("inf")], colors = [ "gray" ], hatches = ['////'], alpha=getAlpha( "gray" ), zorder=10 )
hldZcomb68 = computeHPD ( ZCOMB, RMAX, .68, False, rthreshold=self.rthreshold )
contZCOMB = plt.contour ( X, Y, hldZcomb68, levels=[.25], colors = [ "black" ], zorder=10 )
# ax.scatter( [ minXY[0] ], [ minXY[1] ], marker="s", s=110, color="gray", label="excluded", alpha=.3, zorder=20 )
print()
self.pprint ( "timestamp:", self.timestamp, self.topo, max(x) )
dx,dy = max(x)-min(x),max(y)-min(y)
if self.drawtimestamp:
plt.text( max(x)-.37*dx,min(y)-.11*dy,self.timestamp, c="gray" )
### the altitude of the alpha quantile is l(nuhat) - .5 chi^2_(1-alpha);ndf
### so for alpha=0.05%, ndf=1 the dl is .5 * 3.841 = 1.9207
### for ndf=2 the dl is ln(alpha) = .5 * 5.99146 = 2.995732
### folien slide 317
if hasattr ( cont50, "axes" ):
ax = cont50.axes
else:
ax = cont50.ax
# Xs,Ys=X,Y
Xs,Ys = filterSmaller ( X, Y )
h = self.getHash()
# print ( "hash is", h )
#s=" (??)"
#if h in L:
# s=" (%.2f)" % L[h]
#s=" (%.2f)" % self.getLClosestTo ( L )
s=""
ax.scatter( [ self.mx ], [ self.my ], marker="*", s=200, color="white", zorder=20 )
c = ax.scatter( [ self.mx ], [ self.my ], marker="*", s=160, color="black",
label="proto-model%s" % s, zorder=20 )
handles.append ( c )
if sr == None:
sr = "UL"
# plt.title ( "HPD regions, %s [%s]" % ( namer.texName(pid1, addSign=False, addDollars=True), self.topo ), fontsize=14 )
plt.xlabel ( "m(%s) [GeV]" % namer.texName(pid1,addSign=False, addDollars=True), fontsize=14 )
plt.ylabel ( "m(%s) [GeV]" % namer.texName(self.pid2, addSign=False, addDollars=True), fontsize=14 )
circ1 = mpatches.Patch( facecolor="gray",alpha=getAlpha("gray"),hatch=r'////',label='excluded by critic', edgecolor="black" )
handles.append ( circ1 )
plt.legend( handles=handles, loc="upper left", fontsize=12 )
figname = "%s/llhd%d.png" % ( self.rundir, pid1 )
self.pprint ( "saving to %s" % figname )
plt.savefig ( figname )
if self.interactive:
self.axes = ax
self.plt = plt
plt.close()
if self.copy:
self.copyFile ( figname )
return
def copyFile ( self, filename ):
""" copy filename to smodels.github.io/protomodels/<upload>/ """
dest = os.path.expanduser ( "~/git/smodels.github.io" )
cmd = "cp %s %s/protomodels/%s/" % ( filename, dest, self.upload )
o = subprocess.getoutput ( cmd )
self.pprint ( "%s: %s" % ( cmd, o ) )
def getAnaStats ( self, integrateSRs=True, integrateTopos=True,
integrateDataType=True ):
""" given the likelihood dictionaries D, get
stats of which analysis occurs how often
:param integrateTopos: sum over all topologies
:param integrateSRs: sum over all signal regions
:param integrateDataType: ignore data type
"""
anas = {}
if self.masspoints == None:
return None
for masspoint in self.masspoints:
m1,m2,llhds=masspoint[0],masspoint[1],masspoint[2]
if len(masspoint)>3:
robs = masspoint[3]
for k,v in llhds.items():
tokens = k.split(":")
if not integrateTopos and self.topo not in tokens[2]:
continue
dType = ":em"
if tokens[1] in [ "None", None ]:
dType = ":ul"
name = tokens[0]
if not integrateDataType:
name = name + dType
if not integrateTopos:
name = tokens[0]+tokens[1]
if not name in anas.keys():
anas[name]=0
anas[name]=anas[name]+1
return anas
def listAnalyses( self ):
"""
:param verbose: verbosity: debug, info, warn, or error
"""
stats = self.getAnaStats( integrateDataType=False )
print ( "%6d masspoints with %s" % ( len(self.masspoints), self.topo ) )
for k,v in stats.items():
print ( "%6d: %s" % ( v, k ) )
def compress ( self ):
""" produce a pcl file with only a fraction of the points.
good for testing and development """
backupfile = self.picklefile.replace(".pcl",".bu.pcl")
subprocess.getoutput ( "cp %s %s" % ( self.picklefile, backupfile ))
newfile = self.picklefile.replace(".pcl",".comp.pcl")
mx,my=set(),set()
for m in self.masspoints:
mx.add ( m[0] )
my.add ( m[1] )
mx=list(mx)
my=list(my)
with open ( newfile, "wb" ) as f:
mps = []
for i,m in enumerate(self.masspoints):
if mx.index (m[0] ) % 2 == 0 and \
my.index (m[1] ) % 2 == 0:
# if i % 5 == 0:
mps.append ( m )
pickle.dump ( mps, f )
pickle.dump ( self.mx, f )
pickle.dump ( self.my, f )
pickle.dump ( self.nevents, f )
pickle.dump ( self.topo, f )
pickle.dump ( self.timestamp, f )
f.close()
def findClosestPoint ( self, m1=None, m2=None, nll=False ):
""" find the mass point closest to m1, m2. If not specified,
return the hiscore point.
:param nll: if True, report nlls, else report likelihoods.
"""
if m1 == None:
m1 = self.mx
if m2 == None:
m2 = self.my
dm,point = float("inf"),None
def distance ( m ):
return (m[0]-m1)**2 + (m[1]-m2)**2
for m in self.masspoints:
tmp = distance(m)
if tmp < dm:
dm = tmp
point = m
if not nll:
return point
# asked for NLLs
D = {}
for k,v in point[2].items():
D[k]=-np.log(v)
return ( point[0], point[1], D )
def interact ( self ):
import IPython
varis = "plot.describe()"
print ( "%s[plot] interactive session. Try: %s%s" % \
( colorama.Fore.GREEN, varis, colorama.Fore.RESET ) )
IPython.embed( using=False )
#Set up the path to SModelS installation folder if running on a different folder
import sys, os
sys.path.append(os.path.join(os.getenv("HOME"), "smodels/"))
# In[2]:
#Import those parts of smodels that are needed for this exercise
from smodels.tools.physicsUnits import GeV
from smodels.experiment.databaseObj import Database
# In[3]:
## Load the database:
databasePath = os.path.join(os.getenv("HOME"), "smodels-database/")
db = Database(databasePath)
# ## Look up upper limit for an Upper Limit-type result:
# In[4]:
#Select desired result:
resultID = ["CMS-PAS-SUS-13-016"]
txname = ["T1tttt"]
expResult = db.getExpResults(analysisIDs=resultID,
txnames=txname,
dataTypes='upperLimit')[0]
print 'selected result:', expResult
# In[5]:
#!/usr/bin/env python3
""" just test that the fake signal gets injected in the right part of the UL maps """
import math
from smodels.experiment.databaseObj import Database
from smodels.tools.physicsUnits import GeV
# ./ptools/fetchFromClip.py -R rundir.frozen1 --database
db = Database("default.pcl")
print("db", db.databaseVersion)
er = db.getExpResults(["CMS-SUS-19-006"])[0]
ds = er.datasets[0]
print(ds.txnameList)
txn = ds.txnameList[6]
print(txn)
def distance(mass):
# return math.sqrt ( (mass[0] - 735.)**2 + (mass[1]-162.6)**2 )
return math.sqrt((mass[0] - 1166.)**2 + (mass[1] - 162.6)**2)
masses = []
for mLSP in range(100, 240, 50):
for msquark in range(850, 1300, 50):
masses.append([msquark, mLSP])
# masses.append ( [[msquark*GeV,mLSP*GeV],[msquark*GeV,mLSP*GeV]] )
for mass in masses:
mvec = [[mass[0] * GeV, mass[1] * GeV], [mass[0] * GeV, mass[1] * GeV]]
oUL = txn.getULFor(mvec, expected=False)
eUL = txn.getULFor(mvec, expected=True)
class ProxyDBCreater:
def __init__(self, inputfile, rundir, verbose="info"):
self.inputfile = inputfile
self.rundir = rundir
self.nr = 0
self.verbstring = verbose
if type(verbose) == int:
self.verbose = verbose
else:
verbose = verbose.lower()
verbs = {"err": 10, "warn": 20, "info": 30, "debug": 40}
self.verbose = 50
for k, v in verbs.items():
if k in verbose:
self.verbose = v
self.database = Database(self.inputfile)
def create(self, servername, serverport):
if servername == None:
servername = socket.gethostname()
self.pprint("determined servername as '%s'" % servername)
if serverport == None:
serverport = 31770
self.servername = servername
self.serverport = serverport
self.database.client = DatabaseClient(servername,
serverport,
verbose=self.verbstring,
rundir=self.rundir,
clientid=self.nr)
for e, expRes in enumerate(self.database.expResultList):
for d, dataset in enumerate(expRes.datasets):
for t, txn in enumerate(dataset.txnameList):
self.database.expResultList[e].datasets[d].txnameList[
t].dbClient = copy.copy(self.database.client)
del self.database.expResultList[e].datasets[d].txnameList[
t].txnameData.tri
if txn.txnameDataExp != None:
del self.database.expResultList[e].datasets[
d].txnameList[t].txnameDataExp.tri
def pprint(self, *args):
if self.verbose > 25:
print ( "[proxyDBCreater-%s] %s" % \
( time.strftime("%H:%M:%S"), " ".join(map(str,args)) ) )
def store(self, outputfile):
""" store the outputfile """
self.outputfile = outputfile
self.pprint("writing to %s" % outputfile)
if os.path.exists(outputfile):
os.unlink(outputfile)
## first create it as temporary file, then move
tempf = outputfile + ".tmp" # tempfile.mktemp ( suffix=".pcl" )
self.database.createBinaryFile(tempf)
#cmd = f"mv {tempf} {outputfile}"
#subprocess.getoutput ( cmd )
os.rename(tempf, outputfile) ## would only work on same device
def symlink(self):
""" set a symlink from self.outputfile to default.pcl """
dirname = os.path.dirname(self.outputfile)
symfile = f"{dirname}/default.pcl"
self.pprint ( "setting a symlink from %s to %s" % \
( self.outputfile, symfile ) )
if os.path.exists(symfile):
os.unlink(symfile)
cmd = f"ln -s {self.outputfile} {symfile}"
subprocess.getoutput(cmd)
def run(self, really):
""" now run the server
:param really: if False, then only write out command
"""
dirname = os.path.dirname(__file__)
inputfile = self.inputfile
#if not "/" in inputfile:
# inputfile = os.getcwd() + "/" + inputfile
servercmd = "%s/databaseServer.py -R %s -p %d -d %s -v %s" % \
( dirname, self.rundir, self.serverport, inputfile, self.verbstring )
if really:
self.pprint ( "starting a server on %s: %s" % \
( self.servername, servercmd ) )
import subprocess
a = subprocess.getoutput(servercmd)
self.pprint("output %s" % a)
else:
print("not started a server. you can start one yourself:")
self.pprint(servercmd)
if args.prior:
normalizePrior()
sys.exit()
if args.upper_limits and args.efficiencyMaps:
print("[combiner] -u and -e are mutually exclusive")
sys.exit()
from smodels.experiment.databaseObj import Database
from smodels.theory import decomposer
from smodels.particlesLoader import BSMList
from smodels.share.models.SMparticles import SMList
from smodels.theory.model import Model
from smodels.tools.physicsUnits import fb
model = Model(BSMparticles=BSMList, SMparticles=SMList)
model.updateParticles(inputFile=args.slhafile)
print("[combiner] loading database", args.database)
db = Database(args.database)
print("[combiner] done loading database")
anaIds = ["CMS-SUS-16-033"]
anaIds = ["all"]
dts = ["all"]
if args.upper_limits:
dts = ["upperLimit"]
if args.efficiencyMaps:
dts = ["efficiencyMap"]
listOfExpRes = db.getExpResults(analysisIDs=anaIds,
dataTypes=dts,
onlyWithExpected=True)
smses = decomposer.decompose(model, .01 * fb)
#print ( "[combiner] decomposed into %d topos" % len(smses) )
from smodels.theory.theoryPrediction import theoryPredictionsFor
combiner = Combiner()
#!/usr/bin/env python3
"""
.. module:: databaseLoader
:synopsis: When running the complete test suite, we need to
load the database only once
.. moduleauthor:: Wolfgang Waltenberger <*****@*****.**>
"""
import sys
sys.path.insert(0, "../")
from smodels.experiment.databaseObj import Database
from smodels.installation import version
ver = "".join(map(str, version(True)[:3]))
#dbname="./database/db%d0.pcl" % int ( sys.version[0] )
dbname = "http://smodels.hephy.at/database/unittest%s" % ver
database = Database(dbname, discard_zeroes=False)
if __name__ == "__main__":
print(database)
def testLoadLatest(self):
dblatest=Database ("latest")
latestver = dblatest.databaseVersion.replace(".","")
from databaseLoader import database
thisver = database.databaseVersion.replace("unittest","").replace(".","")
self.assertTrue ( latestver[:2]==thisver[:2] )
from smodels.installation import installDirectory
from smodels.tools.physicsUnits import fb, GeV
from smodels.theory.theoryPrediction import theoryPredictionsFor
from smodels.experiment.databaseObj import Database
# In[3]:
#Define the SLHA input file name
filename = "%s/inputFiles/slha/gluino_squarks.slha" % installDirectory()
# In[4]:
#Load the database, do the decomposition and compute theory predictions:
#(Look at the theory predictions HowTo to learn how to compute theory predictions)
databasepath = os.path.join(os.getenv("HOME"), "smodels-database/")
database = Database(databasepath)
expResults = database.getExpResults()
topList = slhaDecomposer.decompose(filename,
sigcut=0.03 * fb,
doCompress=True,
doInvisible=True,
minmassgap=5 * GeV)
allThPredictions = [theoryPredictionsFor(exp, topList) for exp in expResults]
# In[5]:
#Print the value of each theory prediction for each experimental
#result and the corresponding upper limit (see the obtain experimental upper limits HowTo to learn how
#to compute the upper limits).
#Also print the expected upper limit, if available
for thPreds in allThPredictions:
class Predictor:
def __init__(self,
walkerid,
dbpath="./default.pcl",
expected=False,
select="all",
do_combine=False):
"""
:param do_combine: if True, then also use combined results,
both via simplified likelihoods and pyhf.
"""
self.walkerid = walkerid
self.do_combine = do_combine
self.modifier = None
self.select = select
self.expected = expected
self.rthreshold = 1.3 ## threshold for rmax
if expected:
from expResModifier import ExpResModifier
self.modifier = ExpResModifier()
force_load = None
if dbpath.endswith(".pcl"):
force_load = "pcl"
ntries = 0
while not os.path.exists(dbpath):
## give it a few tries
ntries += 1
time.sleep(ntries * 5)
if ntries > 5:
break
self.database = Database(dbpath, force_load=force_load)
self.fetchResults()
self.combiner = Combiner(self.walkerid)
def filterForAnaIdsTopos(self, anaIds, topo):
""" filter the list of expRes, keep only anaIds """
keepExpRes = []
nbefore = len(self.listOfExpRes)
for er in self.listOfExpRes:
eid = er.globalInfo.id
if not eid in anaIds:
continue
txnames = [x.txName for x in er.getTxNames()]
if not topo in txnames: ## can safely skip
continue
newDS = []
for dataset in er.datasets:
newTxNames = []
for txName in dataset.txnameList:
if txName.txName != topo:
continue
newTxNames.append(txName)
if len(newTxNames) > 0:
dataset.txnameList = newTxNames
newDS.append(dataset)
if len(newDS) > 0:
er.datasets = newDS
keepExpRes.append(er)
self.pprint ( "filtered for %s, keeping %d/%d expRes" % \
( topo, len(keepExpRes), nbefore) )
self.listOfExpRes = keepExpRes
def filterForTopos(self, topo):
""" filter the list of expRes, keep only the ones for topo """
keepExpRes = []
nbefore = len(self.listOfExpRes)
for er in self.listOfExpRes:
txnames = [x.txName for x in er.getTxNames()]
if not topo in txnames: ## can safely skip
continue
newDS = []
for dataset in er.datasets:
newTxNames = []
for txName in dataset.txnameList:
if txName.txName != topo:
continue
newTxNames.append(txName)
if len(newTxNames) > 0:
dataset.txnameList = newTxNames
newDS.append(dataset)
if len(newDS) > 0:
er.datasets = newDS
keepExpRes.append(er)
self.pprint ( "filtered for %s, keeping %d/%d expRes" % \
( topo, len(keepExpRes), nbefore) )
self.listOfExpRes = keepExpRes
def fetchResults(self):
""" fetch the list of results, perform all selecting
and modding """
dataTypes = ["all"]
if self.select == "em":
dataTypes = ["efficiencyMap"]
if self.select == "ul":
dataTypes = ["upperLimit"]
txnames = ["all"]
if self.select.startswith("txnames:"):
s = self.select.replace("txnames:", "")
txnames = s.split(",")
self.pprint("I have been asked to select txnames for %s" % s)
listOfExpRes = self.database.getExpResults(dataTypes=dataTypes,
txnames=txnames,
useNonValidated=True)
if self.modifier:
listOfExpRes = self.modifier.modify(listOfExpRes)
self.listOfExpRes = listOfExpRes
if False:
f = open("expresults.txt", "wt")
for expRes in self.listOfExpRes:
f.write("%s %s\n" % (expRes.id(), expRes.datasets[0]))
f.close()
def pprint(self, *args):
""" logging """
print("[predictor] %s" % (" ".join(map(str, args))))
self.log(*args)
def log(self, *args):
""" logging to file """
with open("walker%d.log" % self.walkerid, "a") as f:
f.write("[predictor-%s] %s\n" %
(time.strftime("%H:%M:%S"), " ".join(map(str, args))))
def predict(self,
protomodel,
sigmacut=0.02 * fb,
strategy="aggressive",
keep_predictions=False):
""" Compute the predictions and statistical variables, for a
protomodel.
:param sigmacut: weight cut on the predict xsecs for theoryPredictions
:param strategy: combination strategy, currently only aggressive is used
:param keep_predictions: if True, then keep all predictions (in self,
not in protomodel!!)
:returns: False, if no combinations could be found, else True
"""
if hasattr(self, "predictions"):
del self.predictions ## make sure we dont accidentally use old preds
# Create SLHA file (for running SModelS)
slhafile = protomodel.createSLHAFile()
# First run SModelS using all results and considering only the best signal region.
# thats the run for the critic
bestpreds = self.runSModelS(slhafile,
sigmacut,
allpreds=False,
llhdonly=False)
if keep_predictions:
self.bestpreds = bestpreds
# Extract the relevant prediction information and store in the protomodel:
self.updateModelPredictions(protomodel, bestpreds)
# self.log ( "model is excluded? %s" % str(protomodel.excluded) )
# Compute the maximum allowed (global) mu value given the r-values
# stored in protomodel
protomodel.mumax = self.getMaxAllowedMu(protomodel)
# now use all prediction with likelihood values to compute the Z of the model
predictions = self.runSModelS(slhafile,
sigmacut,
allpreds=True,
llhdonly=True)
if keep_predictions:
self.predictions = predictions
# Compute significance and store in the model:
self.computeSignificance(protomodel, predictions, strategy)
if protomodel.Z is None:
self.log(
"done with prediction. Could not find combinations (Z=%s)" %
(protomodel.Z))
protomodel.delCurrentSLHA()
return False
else:
self.log("done with prediction. best Z=%.2f (muhat=%.2f)" %
(protomodel.Z, protomodel.muhat))
protomodel.cleanBestCombo()
#Recompute predictions with higher accuracy for high score models:
if protomodel.Z > 2.7 and protomodel.nevents < 55000:
protomodel.nevents = 100000
protomodel.computeXSecs()
self.predict(protomodel, sigmacut=sigmacut, strategy=strategy)
protomodel.delCurrentSLHA()
return True
def runSModelS(self, inputFile, sigmacut, allpreds, llhdonly):
""" run smodels proper.
:param inputFile: the input slha file
:param sigmacut: the cut on the topology weights, typically 0.02*fb
:param allpreds: if true, return all predictions of analyses, else
only best signal region
:param llhdonly: if true, return only results with likelihoods
"""
if not os.path.exists(inputFile):
self.pprint("error, cannot find inputFile %s" % inputFile)
return []
model = Model(BSMList, SMList)
model.updateParticles(inputFile=inputFile)
mingap = 10 * GeV
# self.log ( "Now decomposing" )
topos = decomposer.decompose(model, sigmacut, minmassgap=mingap)
self.log("decomposed model into %d topologies." % len(topos))
if allpreds:
bestDataSet = False
combinedRes = False
else:
bestDataSet = True
combinedRes = self.do_combine
preds = []
# self.log ( "start getting preds" )
from smodels.tools import runtime
runtime._experimental = True
for expRes in self.listOfExpRes:
predictions = theoryPredictionsFor(expRes,
topos,
useBestDataset=bestDataSet,
combinedResults=combinedRes)
if predictions == None:
predictions = []
if allpreds:
combpreds = theoryPredictionsFor(
expRes,
topos,
useBestDataset=False,
combinedResults=self.do_combine)
if combpreds != None:
for c in combpreds:
predictions.append(c)
for prediction in predictions:
prediction.computeStatistics()
if (not llhdonly) or (prediction.likelihood != None):
preds.append(prediction)
sap = "best preds"
if allpreds:
sap = "all preds"
sllhd = ""
if llhdonly:
sllhd = ", llhds only"
self.log ( "returning %d predictions, %s%s" % \
(len(preds),sap, sllhd ) )
return preds
def printPredictions(self):
""" if self.predictions exists, pretty print them """
if hasattr(self, "predictions"):
print("[predictor] all predictions (for combiner):")
for p in self.predictions:
print ( " - %s %s, %s %s" % \
( p.analysisId(), p.dataType(), p.dataset.dataInfo.dataId, p.txnames ) )
if hasattr(self, "bestpreds"):
print("[predictor] best SR predictions (for critic):")
for p in self.bestpreds:
print ( " - %s %s, %s %s" % \
( p.analysisId(), p.dataType(), p.dataset.dataInfo.dataId, p.txnames ) )
def updateModelPredictions(self, protomodel, predictions):
""" Extract information from list of theory predictions and store in the protomodel.
:param predictions: all theory predictions
:returns: list of tuples with observed r values, r expected and
theory prediction info (sorted with highest r-value first)
"""
rvalues = [] #If there are no predictions set rmax and r2 to 0
tpList = []
for theorypred in predictions:
r = theorypred.getRValue(expected=False)
if r == None:
self.pprint("I received %s as r. What do I do with this?" % r)
r = 23.
rexp = theorypred.getRValue(expected=True)
# tpList.append( (r, rexp, self.combiner.removeDataFromTheoryPred ( theorypred ) ) )
tpList.append((r, rexp, theorypred))
rvalues.append(r)
while len(rvalues) < 2:
rvalues.append(0.)
rvalues.sort(reverse=True)
srs = "%s" % ", ".join(["%.2f" % x for x in rvalues[:3]])
self.log("top r values before rescaling are: %s" % srs)
protomodel.rvalues = rvalues #Do not include initial zero values
# protomodel.excluded = protomodel.rvalues[0] > self.rthreshold #The 0.99 deals with the case rmax = threshold
protomodel.tpList = tpList[:]
def getMaxAllowedMu(self, protomodel):
""" Compute the maximum (global) signal strength normalization
given the predictions.
"""
mumax = float("inf")
if protomodel.rvalues[0] > 0.:
#Set mumax slightly below threshold, so the model is never excluded
mumax = 0.999 * self.rthreshold / protomodel.rvalues[0]
return mumax
def computeSignificance(self, protomodel, predictions, strategy):
""" compute the K and Z values, and attach them to the protomodel """
self.log("now find highest significance for %d predictions" %
len(predictions))
## find highest observed significance
#(set mumax just slightly below its value, so muhat is always below)
mumax = protomodel.mumax
combiner = self.combiner
bestCombo, Z, llhd, muhat = combiner.findHighestSignificance(
predictions, strategy, expected=False, mumax=mumax)
prior = combiner.computePrior(protomodel)
if hasattr(protomodel, "keep_meta") and protomodel.keep_meta:
protomodel.bestCombo = bestCombo
else:
protomodel.bestCombo = combiner.removeDataFromBestCombo(bestCombo)
protomodel.Z = Z
if Z is not None: # Z is None when no combination was found
protomodel.K = combiner.computeK(Z, prior)
else:
protomodel.K = None
protomodel.llhd = llhd
protomodel.muhat = muhat
protomodel.letters = combiner.getLetterCode(protomodel.bestCombo)
protomodel.description = combiner.getComboDescription(
protomodel.bestCombo)
def draw( strategy, databasepath, trianglePlot, miscol,
diagcol, experiment, S, drawtimestamp, outputfile, nofastlim ):
"""
:param trianglePlot: if True, then only plot the upper triangle of this
symmetrical matrix
:param miscol: color to use when likelihood is missing
:param diagcol: color to use for diagonal
:param experiment: draw only for specific experiment ("CMS", "ATLAS", "all" )
:param S: draw only for specific sqrts ( "8", "13", "all" )
:param drawtimestamp: if true, put a timestamp on plot
:param outputfile: file name of output file (matrix.png)
:param nofastlim: if True, discard fastlim results
"""
ROOT.gStyle.SetOptStat(0000)
ROOT.gROOT.SetBatch()
cols = [ ROOT.kRed+1, ROOT.kWhite, ROOT.kGreen+1, miscol, diagcol ]
ROOT.gStyle.SetPalette(len(cols), (ctypes.c_int * len(cols))(*cols) )
ROOT.gStyle.SetNumberContours(len(cols))
ROOT.gStyle.SetPadLeftMargin(.25)
sqrtses = [ 8, 13 ]
if S not in [ "all" ]:
sqrtses = [ int(S) ]
colors.on = True
setLogLevel ( "debug" )
# dir = "/home/walten/git/smodels-database/"
dir = databasepath
d=Database( dir, discard_zeroes = True )
print(d)
analysisIds = [ "all" ]
exps = [ "CMS", "ATLAS" ]
if experiment in [ "CMS", "ATLAS" ]:
analysisIds = [ experiment+"*" ]
exps = [ experiment ]
results = d.getExpResults( analysisIDs = analysisIds )
if nofastlim:
results = noFastlim ( results )
results = sortOutDupes ( results )
if S in [ "8", "13" ]:
results = sortBySqrts ( results, int(S) )
#results.sort()
nres = len ( results )
ROOT.c1=ROOT.TCanvas("c1","c1",1770,1540)
ROOT.c1.SetLeftMargin(0.18)
ROOT.c1.SetBottomMargin(0.21)
ROOT.c1.SetTopMargin(0.06)
ROOT.c1.SetRightMargin(0.01)
if nres > 60:
ROOT.c1.SetLeftMargin(0.12) ## seemed to work for 96 results
ROOT.c1.SetBottomMargin(0.15)
ROOT.c1.SetTopMargin(0.09)
ROOT.c1.SetRightMargin(0.015)
h=ROOT.TH2F ( "Correlations", "",
nres, 0., nres, nres, 0., nres )
xaxis = h.GetXaxis()
yaxis = h.GetYaxis()
sze = 0.13 / math.sqrt ( nres )
xaxis.SetLabelSize( 1.3*sze )
yaxis.SetLabelSize( 1.3*sze )
bins= { "CMS": { 8: [999,0], 13:[999,0] },
"ATLAS": { 8: [999,0], 13: [999,0] } }
n = len(results )
for x,e in enumerate(results):
label = e.globalInfo.id
hasLikelihood = hasLLHD ( e )
ana = analysisCombiner.getExperimentName ( e.globalInfo )
#if not hasLikelihood:
# print ( "no likelihood: %s" % label )
sqrts = int(e.globalInfo.sqrts.asNumber(TeV))
color = ROOT.kCyan+2
ymax=0
if ana == "ATLAS":
color = ROOT.kBlue+1
if sqrts > 10.:
color += 2
if x < bins[ana][sqrts][0]:
bins[ana][sqrts][0]=x
if x > bins[ana][sqrts][1]:
bins[ana][sqrts][1]=x
ymax=x
color = ROOT.kGray+2
if len(exps)==1 and len(sqrtses)==1:
label = label.replace("CMS-","").replace("ATLAS-","").replace("-agg","")
label = "#color[%d]{%s}" % (color, label )
xaxis.SetBinLabel(n-x, label )
yaxis.SetBinLabel(x+1, label )
for y,f in enumerate(results):
if trianglePlot and y>x:
continue
isUn = analysisCombiner.canCombine ( e.globalInfo, f.globalInfo, strategy )
# isUn = e.isUncorrelatedWith ( f )
if isUn:
h.SetBinContent ( n-x, y+1, 1. )
else:
h.SetBinContent ( n-x, y+1, -1. )
if not hasLikelihood or not hasLLHD ( f ): ## has no llhd? cannot be combined
h.SetBinContent ( n-x, y+1, 2. )
if y==x:
h.SetBinContent ( n-x, y+1, 3. )
h.Draw("col")
ROOT.bins, ROOT.xbins, ROOT.lines = {}, {}, []
if len(exps)==1 and len(sqrtses)==1:
ROOT.t1 = ROOT.TLatex()
ROOT.t1.SetNDC()
ROOT.t1.DrawLatex ( .45, .95, "%s, %d TeV" % ( exps[0], sqrtses[0] ) )
for ana in exps:
for sqrts in sqrtses:
name= "%s%d" % ( ana, sqrts )
ROOT.bins[name] = ROOT.TLatex()
ROOT.bins[name].SetTextColorAlpha(ROOT.kBlack,.7)
ROOT.bins[name].SetTextSize(.025)
ROOT.bins[name].SetTextAngle(90.)
ROOT.xbins[name] = ROOT.TLatex()
ROOT.xbins[name].SetTextColorAlpha(ROOT.kBlack,.7)
ROOT.xbins[name].SetTextSize(.025)
xcoord = .5 * ( bins[ana][sqrts][0] + bins[ana][sqrts][1] )
ycoord = n- .5 * ( bins[ana][sqrts][0] + bins[ana][sqrts][1] ) -3
if len(sqrtses)>1 or len(exps)>1:
ROOT.bins[name].DrawLatex(-4,xcoord-3,"#splitline{%s}{%d TeV}" % ( ana, sqrts ) )
ROOT.xbins[name].DrawLatex(ycoord,-5,"#splitline{%s}{%d TeV}" % ( ana, sqrts ) )
yt = bins[ana][sqrts][1] +1
extrudes = 3 # how far does the line extrude into tick labels?
xmax = n
if trianglePlot:
xmax = n-yt
line = ROOT.TLine ( -extrudes, yt, xmax, yt )
line.SetLineWidth(2)
line.Draw()
ymax = n
if trianglePlot:
ymax = yt
xline = ROOT.TLine ( n-yt, ymax, n-yt, -extrudes )
xline.SetLineWidth(2)
xline.Draw()
ROOT.lines.append ( line )
ROOT.lines.append ( xline )
line = ROOT.TLine ( -extrudes, 0, xmax, 0 )
line.SetLineWidth(2)
line.Draw()
xline = ROOT.TLine ( n, ymax, n, -extrudes )
xline.SetLineWidth(2)
xline.Draw()
ROOT.lines.append ( line )
ROOT.lines.append ( xline )
h.LabelsOption("v","X")
if trianglePlot:
for i in range(n+1):
wline = ROOT.TLine ( n, i, n-i, i )
wline.SetLineColor ( ROOT.kWhite )
wline.Draw ()
ROOT.lines.append ( wline )
vline = ROOT.TLine ( i, n-i, i, n )
vline.SetLineColor ( ROOT.kWhite )
vline.Draw ()
ROOT.lines.append ( vline )
ROOT.title = ROOT.TLatex()
ROOT.title.SetNDC()
ROOT.title.SetTextSize(.025 )
ROOT.title.DrawLatex(.28,.89, "#font[132]{Correlations between analyses, combination strategy: ,,%s''}" % strategy )
ROOT.boxes = []
if trianglePlot:
for i,b in enumerate ( [ "pair is uncorrelated", "pair is correlated", "likelihood is missing" ] ):
bx = 51
by = 68 - 3*i
box = ROOT.TBox(bx,by,bx+1,by+1)
c = cols[i]
if i > 0:
c = cols[i+1]
box.SetFillColor ( c )
box.Draw()
ROOT.boxes.append ( box )
l = ROOT.TLatex()
l.SetTextSize(.022)
#if i == 2:
# c = 16
l.SetTextColor ( c )
b="#font[132]{%s}" % b ## add font
l.DrawLatex ( bx+2, by, b )
ROOT.boxes.append ( l )
l = ROOT.TLatex()
l.SetNDC()
l.SetTextColor(ROOT.kGray+1)
l.SetTextSize(.015)
if drawtimestamp:
l.DrawLatex ( .01, .01, "plot produced %s from database v%s" % \
( time.strftime("%h %d %Y" ), d.databaseVersion ) )
ROOT.gPad.SetGrid()
if "@M" in outputfile:
modifiers = ""
if len(exps)==1:
modifiers += exps[0]
if len(sqrtses)==1:
modifiers += str(sqrtses[0])
outputfile = outputfile.replace("@M",modifiers)
print ( "Plotting to %s" % outputfile )
ROOT.c1.Print( outputfile )
# In[1]:
#Set up the path to SModelS installation folder if running on a different folder
import sys, os
sys.path.append(os.path.join(os.getenv("HOME"), "smodels/"))
# In[2]:
from smodels.experiment.databaseObj import Database
from smodels.tools.physicsUnits import GeV
# In[3]:
## Load the database:
dbPath = os.path.join(os.getenv("HOME"), "smodels-database/")
database = Database(dbPath)
# ## How to select results from one publication (or conference note)
# In[6]:
#Select only the CMS SUS-12-028 conference note
expID = ["CMS-SUS-12-028"]
# In[7]:
#Loads the selected analyses
#(The INFO tells you that superseded analyses are not loaded, see below)
results = database.getExpResults(analysisIDs=expID)
# In[9]:
runtime.modelFile = 'smodels.share.models.mssm'
#runtime.modelFile = 'mssmQNumbers.slha'
from smodels.theory import decomposer
from smodels.tools.physicsUnits import fb, GeV, TeV
from smodels.theory.theoryPrediction import theoryPredictionsFor
from smodels.experiment.databaseObj import Database
from smodels.tools import coverage
from smodels.tools.smodelsLogging import setLogLevel
from smodels.particlesLoader import BSMList
from smodels.share.models.SMparticles import SMList
from smodels.theory.model import Model
setLogLevel("info")
# Set the path to the database
database = Database("official")
def main():
"""
Main program. Displays basic use case.
"""
model = Model(BSMparticles=BSMList, SMparticles=SMList)
# Path to input file (either a SLHA or LHE file)
# lhefile = 'inputFiles/lhe/gluino_squarks.lhe'
slhafile = 'inputFiles/slha/lightEWinos.slha'
# model.updateParticles(inputFile=lhefile)
model.updateParticles(inputFile=slhafile)
# Set main options for decomposition
sigmacut = 0.01 * fb
def RunSModelS(self,SLHAFilePath,SummaryFilePath):
# Set the path to the database
database = Database("/home/oo1m20/softwares/smodels-1.2.2/smodels-database")
self.SummaryFilePath = os.path.abspath(SummaryFilePath)
#Define your model (list of rEven and rOdd particles)
particlesLoader.load( 'smodels.share.models.secumssm' ) #Make sure all the model particles are up-to-date
# Path to input file (either a SLHA or LHE file)
self.SLHAFilePath = SLHAFilePath
slhafile = self.SLHAFilePath
#lhefile = 'inputFiles/lhe/gluino_squarks.lhe'
# Set main options for decomposition
sigmacut = 0.01 * fb
mingap = 5. * GeV
# Decompose model (use slhaDecomposer for SLHA input or lheDecomposer for LHE input)
slhaInput = True
if slhaInput:
toplist = slhaDecomposer.decompose(slhafile, sigmacut, doCompress=True, doInvisible=True, minmassgap=mingap)
else:
toplist = lheDecomposer.decompose(lhefile, doCompress=True,doInvisible=True, minmassgap=mingap)
# Access basic information from decomposition, using the topology list and topology objects:
f= open(self.SummaryFilePath,"a+")
print( "\n Decomposition Results: ", file=f )
print( "\t Total number of topologies: %i " %len(toplist), file=f )
nel = sum([len(top.elementList) for top in toplist])
print( "\t Total number of elements = %i " %nel , file=f)
#Print information about the m-th topology (if it exists):
m = 2
if len(toplist) > m:
top = toplist[m]
print( "\t\t %i-th topology = " %m,top,"with total cross section =",top.getTotalWeight(), file=f )
#Print information about the n-th element in the m-th topology:
n = 0
el = top.elementList[n]
print( "\t\t %i-th element from %i-th topology = " %(n,m),el, end="", file=f )
print( "\n\t\t\twith final states =",el.getFinalStates(),"\n\t\t\twith cross section =",el.weight,"\n\t\t\tand masses = ",el.getMasses(), file=f )
# Load the experimental results to be used.
# In this case, all results are employed.
listOfExpRes = database.getExpResults()
# Print basic information about the results loaded.
# Count the number of loaded UL and EM experimental results:
nUL, nEM = 0, 0
for exp in listOfExpRes:
expType = exp.getValuesFor('dataType')[0]
if expType == 'upperLimit':
nUL += 1
elif expType == 'efficiencyMap':
nEM += 1
print( "\n Loaded Database with %i UL results and %i EM results " %(nUL,nEM), file=f )
# Compute the theory predictions for each experimental result and print them:
print("\n Theory Predictions and Constraints:", file=f)
rmax = 0.
bestResult = None
for expResult in listOfExpRes:
predictions = theoryPredictionsFor(expResult, toplist, combinedResults=False, marginalize=False)
if not predictions: continue # Skip if there are no constraints from this result
print('\n %s ' %expResult.globalInfo.id, file=f)
for theoryPrediction in predictions:
dataset = theoryPrediction.dataset
datasetID = dataset.dataInfo.dataId
mass = theoryPrediction.mass
txnames = [str(txname) for txname in theoryPrediction.txnames]
PIDs = theoryPrediction.PIDs
print( "------------------------", file=f )
print( "Dataset = ", datasetID, file=f ) #Analysis name
print( "TxNames = ", txnames, file=f )
print( "Prediction Mass = ",mass, file=f ) #Value for average cluster mass (average mass of the elements in cluster)
print( "Prediction PIDs = ",PIDs, file=f ) #Value for average cluster mass (average mass of the elements in cluster)
print( "Theory Prediction = ",theoryPrediction.xsection, file=f ) #Signal cross section
print( "Condition Violation = ",theoryPrediction.conditions, file=f ) #Condition violation values
# Get the corresponding upper limit:
print( "UL for theory prediction = ",theoryPrediction.upperLimit, file=f )
# Compute the r-value
r = theoryPrediction.getRValue()
print( "r = ",r , file=f)
#Compute likelihhod and chi^2 for EM-type results:
if dataset.dataInfo.dataType == 'efficiencyMap':
theoryPrediction.computeStatistics()
print( 'Chi2, likelihood=', theoryPrediction.chi2, theoryPrediction.likelihood, file=f )
if r > rmax:
rmax = r
bestResult = expResult.globalInfo.id
# Print the most constraining experimental result
print( "\nThe largest r-value (theory/upper limit ratio) is ",rmax, file=f )
if rmax > 1.:
print( "(The input model is likely excluded by %s)" %bestResult, file=f )
else:
print( "(The input model is not excluded by the simplified model results)", file=f )
f.close()
