Page MenuHomeHEPForge
Files F7877336

No OneTemporary
Actions

View Options

diff --git a/bin/prof2-eigentunes b/bin/prof2-eigentunes
new file mode 100755
--- /dev/null
+++ b/bin/prof2-eigentunes
@@ -0,0 +1,254 @@
+#! /usr/bin/env python
+
+"""\
+%prog -d <datadir> -m <resultsfile> [ipolfile(s)] [opts]
+
+Read in a minimisation result, datadir and ipolfile(s) to
+construct eigentunes.
+
+TODO:
+ * Replace manual target with something clever
+"""
+
+def mkEigenTunes(T_trans, point, fixed, GOFdef, target, plus=True):
+ """
+ COV ... real symmetric covariance matrix
+ point ... could be any point in the true parameter space but should be
+ the minimisation result i.e. the center of COV
+ """
+ import numpy as np
+
+ rv = np.matrix([v for k, v in point.iteritems() if not k in fixed.keys()]).transpose()
+
+ ret = []
+
+ # Construct all base vectors (in rotated system) with pos and neg directions
+ dim = len(point.values()) - len(fixed.values())
+ EVS = []
+ for i in xrange(dim):
+ ev = np.zeros(dim) # A zero vector in len(S) dimensions
+ # Set one of the coordinates to 1 or -1
+ ev[i] = 1 if plus else -1
+ EVS.append(np.array(T_trans*np.matrix(ev).transpose()))
+
+
+ # Get the eigentunes
+ for num, ev in enumerate(EVS):
+ thisEigentune = ETSolve(rv, ev, T_trans, GOFdef, target)
+ ret.append([(num+1) if plus else -(num+1), thisEigentune])
+
+ return ret
+
+def ETSolve(center, direction_t, TRAFO, GOFdef, target):
+ exec GOFdef in globals() # Note globals!
+
+ def getVal(a):
+ temp = center + a*direction_t
+ locval = profGoF(*temp) - target
+ return locval
+
+ def getP(a):
+ temp = center + a*direction_t
+ return temp
+
+ from scipy.optimize import fsolve
+ x = fsolve(getVal,0.1)
+ return np.array(getP(x)).ravel()
+
+import optparse, os, sys
+op = optparse.OptionParser(usage=__doc__)
+op.add_option("-d", "--datadir", dest="DATADIR", default=None, help="The data directory")
+op.add_option("-m", "--minres", dest="RESULT", default=None, help="Minimisation result to use for eigentunes calculation")
+op.add_option("-o", "--outdir", dest="OUTDIR", default="valley", help="Output directory")
+op.add_option("-r", "--runsdir", dest="RUNSDIR", default=None, help="The runs directory")
+op.add_option("-v", "--debug", dest="DEBUG", action="store_true", default=False, help="Turn on some debug messages")
+op.add_option("-q", "--quiet", dest="QUIET", action="store_true", default=False, help="turn off messages")
+op.add_option("--wfile", dest="WFILE", default=None, help="Path to a weight file, used to restrict ipol building to a subset of bins (default: %default)")
+op.add_option("--limit-errs", dest="USE_RUNSDIR", action="store_true", default=False, help="Re-read the runsdir to regularise error ipols")
+op.add_option("--limits", dest="LIMITS", default=None, help="Simple text file with parameter limits and fixed parameters")
+op.add_option("--filter", dest="FILTER", action="store_true", default=False, help="Filter out data bins that have 0 error")
+op.add_option("--target", dest="TARGET", type=float, default=2, help="Target factor for delta chi2")
+op.add_option("--kernel", dest="KERNELCOV", action="store_true", default=False, help="Estimate covariance matrix using emcee")
+op.add_option("--kernel-np", dest="KERNELCOVNP", type=int, default=1000, help="Number of emcee points to generate (default: %default)")
+opts, args = op.parse_args()
+
+## Get mandatory arguments, same as prof2-tune
+if len(args) < 1:
+ print "Argument missing... exiting\n\n"
+ op.print_usage()
+ sys.exit(1)
+REFDIR = opts.DATADIR
+if REFDIR is None:
+ print "Error, no data directory specified (-d/--datadir), exiting\n\n"
+ op.print_usage()
+ sys.exit(1)
+IFILES = args
+RUNSDIR = opts.RUNSDIR
+
+# Sanity
+if not os.path.exists(REFDIR):
+ print "Error, specified data directory '%s' does not exist, exiting\n\n"%REFDIR
+ op.print_usage()
+ sys.exit(1)
+if not os.path.exists(opts.OUTDIR):
+ os.makedirs(opts.OUTDIR)
+
+if opts.RESULT is None:
+ print "Error, no result file given"
+ sys.exit(1)
+
+## Load Professor and show the standard banner
+import professor2 as prof
+if not opts.QUIET:
+ print prof.logo
+
+# Read data files
+DHISTOS = prof.read_all_histos(REFDIR)
+
+from collections import OrderedDict
+MASTERBOX=OrderedDict()
+MASTERCENTER=OrderedDict()
+
+## Weight file parsing --- by default read from results file
+matchers = prof.read_pointmatchers(opts.WFILE) if opts.WFILE else prof.read_pointmatchersfromresults(opts.RESULT)
+
+## Try to read run histos and extract maximum errors --- NOTE this bit might be broken with patches NOTE
+MAXERRDICT = None
+if opts.USE_RUNSDIR:
+ try:
+ _, RUNHISTOS = prof.read_all_rundata(RUNSDIR, None) #< don't care about reading params files
+ MAXERRDICT = prof.find_maxerrs(RUNHISTOS)
+ except:
+ print "Could not read run data for error regularisation -- chi2 may be unstable"
+
+for IFILE in IFILES:
+ box, center, fitdata = prof.prepareBox(IFILE, DHISTOS, matchers, MAXERRDICT, opts.FILTER, opts.DEBUG)
+ MASTERBOX[box]=dict(fitdata)
+ MASTERCENTER[center]=MASTERBOX[box]
+
+## Take parameter names from the first box and assert that all other boxes (if present) have the same names
+PNAMES=MASTERBOX.values()[0]["PNAMES"]
+for v in MASTERBOX.values()[1:]:
+ assert PNAMES == v["PNAMES"]
+
+
+# Central tuning result's GOF, min/max paramvalues
+P_min, OTH = prof.readResult(opts.RESULT)
+GOF_min = [float(x.split()[-1]) for x in OTH if "GOF" in x][0]
+gof_target = opts.TARGET+GOF_min
+
+
+pmins, pmaxs = [], []
+for num, pname in enumerate(PNAMES):
+ testmin = [box[num][0] for box in MASTERBOX.keys()]
+ testmax = [box[num][1] for box in MASTERBOX.keys()]
+ pmins.append(min(testmin))
+ pmaxs.append(max(testmax))
+IBOX=zip(pmins,pmaxs)
+
+
+# Fixed Parameters
+fixed=prof.getFixedParams(OTH)
+
+
+
+funcdef = prof.mk_fitfunc("prof.simpleGoF", PNAMES, "profGoF", ["MASTERBOX", "MASTERCENTER", "opts.DEBUG"])
+exec funcdef in locals()
+if not opts.QUIET:
+ print "Info: GOF minimum from file as:",GOF_min
+ print "Info: This should be reasonably close to GOF evaluated at minimum:", profGoF(*P_min.values())
+assert(abs(GOF_min - profGoF(*P_min.values()))<1e-2)
+if not opts.QUIET:
+ print "Info: Target GOF for eigntunes is:", gof_target
+
+if opts.KERNELCOV:
+ if not opts.QUIET:
+ print "Running emcee sampler"
+ import numpy as np
+ try:
+ import emcee
+ except ImportError:
+ raise Exception("Cannot use emcee, try pip install emcee")
+ def loglike(PP):
+ if not prof.pInBOX(PP, IBOX):
+ return -np.inf
+ likelihood = -0.5*profGoF(*PP)
+ return likelihood
+ ndim, nwalkers = len(PNAMES), 100
+ p0 = [np.array([np.random.uniform(p[0],p[1]) for p in IBOX]) for i in range(nwalkers)]
+ sampler = emcee.EnsembleSampler(nwalkers, ndim, loglike, args=[])
+ pos, prob, state = sampler.run_mcmc(p0, opts.KERNELCOVNP)
+ import matplotlib.pyplot as pl
+ for i in range(ndim):
+ pl.clf()
+ pl.figure()
+ pl.hist(sampler.flatchain[:,i], 100, color="k", histtype="step")
+ pl.title(PNAMES[i])
+ pl.savefig(os.path.join(opts.OUTDIR, "emcee_%i_%s.pdf"%(i, PNAMES[i])))
+
+ if not opts.QUIET:
+ print "Mean acceptance fraction: {0:.3f}".format(np.mean(sampler.acceptance_fraction))
+
+ # Kernel density estimator
+ try:
+ from scipy import stats
+ except ImportError:
+ raise Exception("Cannot use scipy stats for kernel estimation, try pip install scipy")
+ kernel = stats.gaussian_kde(pos.T)
+ C_kernel=kernel.covariance
+ if not opts.QUIET:
+ print "Using kernel covariance"
+ T_fwd, S, T_back = prof.eigenDecomposition(C_kernel) # S is vector, not a diagonal matrix
+else:
+ # Parameter covariance from result
+ C_param = prof.getParamCov(OTH) # That's the minimiser covariance
+ T_fwd, S, T_back = prof.eigenDecomposition(C_param) # Eigen decomposition
+
+import numpy as np
+
+# TODO make the whole thing work with fixed parameters, too
+# Calculate the eigen tunes (points in parameter space)
+E_plus = mkEigenTunes(T_fwd, P_min, {}, funcdef, gof_target)
+E_minus = mkEigenTunes(T_fwd, P_min, {}, funcdef, gof_target, plus=False)
+ETs = dict(E_plus+E_minus)
+
+# Print/save eigentunes
+etable = prof.mkEigenTable(ETs, opts.RESULT)
+if not opts.QUIET:
+ print etable
+etableoutfname="%s/Eigentunes_%s.params"%(opts.OUTDIR, str(opts.TARGET))
+with open(etableoutfname, "w") as f:
+ f.write(etable)
+if not opts.QUIET:
+ print "Stored Eigentune parameters in file %s"%etableoutfname
+
+try:
+ import yoda
+except ImportError:
+ print "Cannot find yoda, not writing output histograms"
+ sys.exit(1)
+
+# Get the corresponding ipol histos
+EThists = {}
+for k, ET in ETs.iteritems():
+ thisEThists = prof.mkScatters(MASTERBOX, MASTERCENTER, ET)
+ sgn = "+" if k > 0 else "-"
+ yoda.writeYODA(thisEThists, "%s/Eigentunes_%.1f_%i%s.yoda" % (opts.OUTDIR,opts.TARGET, int(abs(k)), sgn))
+ EThists[k]=thisEThists
+
+# And for convenience corresponding envelopes
+H_min = prof.mkScatters(MASTERBOX, MASTERCENTER, P_min)
+envelopes = prof.mkEnvelopes(H_min, EThists)
+for k, v in envelopes.iteritems():
+ yoda.writeYODA(v, "%s/EigentunesComb_%.1f_%i.yoda" % (opts.OUTDIR,opts.TARGET, k))
+
+# This is the envelope of the eigentunes
+totvelopes = prof.mkTotvelopes(H_min, EThists)
+# This is the deltas added in quadrature
+quadvelopes = prof.mkAddvelopes(H_min, EThists)
+# This is the deltas added linearly
+linvelopes = prof.mkAddvelopes(H_min, EThists, addLinear=True)
+
+yoda.writeYODA(totvelopes, "%s/Totvelopes_%.1f.yoda"%(opts.OUTDIR,opts.TARGET))
+yoda.writeYODA(quadvelopes, "%s/Quadvelopes_%.1f.yoda"%(opts.OUTDIR,opts.TARGET))
+yoda.writeYODA(linvelopes, "%s/Linvelopes_%.1f.yoda"%(opts.OUTDIR,opts.TARGET))

File Metadata

Mime Type
text/x-diff
Expires
Tue, Nov 19, 3:16 PM (1 d, 14 h)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
3801097
Default Alt Text
(9 KB)

Event Timeline