GLoBES Patrick Huber Center for Neutrino Physics, Virginia Tech What is ν ? Galileo Galilei Institute for Theoretical Physics, June 2012 P. Huber – p. 1
What? General Long Baseline Experiment Simulator GLoBES is a software package designed for • Simulation • Analysis • Comparison of neutrino oscillation experiments P. Huber – p. 2
Where? It is developed and maintained by • PH • Joachim Kopp • Manfred Lindner • Walter Winter URL – http://www.mpi-hd.mpg.de/lin/globes/ email – globes@mpi-hd.mpg.de P. Huber – p. 3
Why? Wide band beam Resolution δ CP vs Sin 2 2 θ 13 150 δ CP (deg.) 150 100 ν µ Running Only True value of ∆ CP 100 50 90 % C.L. 50 0 3 cntrs: STAT only STAT+10 % SYST 0 STAT+20 % SYST � 50 -50 � 100 -100 � 150 FNAL-HS 1290 km sin 2 2 θ ij (12,23,13) = 0.86/1.0/0.04, δ CP =45 o ∆ m ij2 (21,32) = 7.3e-5/2.5e-3 eV 2 -150 1 MW, 0.5 MT, 5e7 sec 10 � 3 10 � 2 10 � 1 0 0.02 0.04 0.06 0.08 0.1 0.12 True value of sin 2 2 Θ 13 Sin 2 2 θ 13 P. Huber – p. 4
Why? APS study Sensitivity reach in sin 2 2 Θ 13 sin 2 2 Θ 13 3000 km � 2 � sgn � � m 31 3000 km CP viol. sin 2 2 Θ 13 7500 km � 2 � sgn � � m 31 7500 km No sensitivity � CP viol. sin 2 2 Θ 13 7500 km � 2 � sgn � � m 31 3000 km CP viol. 10 � 5 10 � 4 10 � 3 10 � 2 10 � 1 sin 2 2 Θ 13 P. Huber – p. 5
Why? Fermilab’s Proton driver report 10 � 5 MINOS CNGS D � CHOOZ Ν� factories T2K 10 � 4 NO � A Reactor � II Superbeam upgrades sin 2 2 Θ 13 discovery reach � 3 Σ � NO � A � FPD 2 nd GenPDExp NuFact 10 � 3 Superbeams � Reactor exps 10 � 2 Branching point Conv. beams 10 � 1 CHOOZ � Solar excluded 10 0 2005 2010 2015 2020 2025 2030 Year P. Huber – p. 6
Why? White paper on reactor neutrinos Reactor-I Reactor-II BG in far detector 2 2 θ 13 sensitivity limit bin-to-bin error 5% 1% 0.5% -2 -2 10 10 sin 0 . 1 % no BG, no bin-to-bin error 2 2 3 3 4 4 5 5 10 10 10 10 10 10 10 10 luminosity [t GW y] P. Huber – p. 7
Why? CERN strategy group ! 35 P. Huber et al. P. Huber – p. 8
Why? ISS 1 0.8 SPL T2HK Fraction of ∆ CP WBB 0.6 NF BB 0.4 0.2 GLoBES 2006 0 10 � 5 10 � 4 10 � 3 10 � 2 10 � 1 True value of sin 2 2 Θ 13 P. Huber – p. 9
Why? LBNE P. Huber – p. 10
Why? MINOS+ 2 10 LSND 90% CL LSND 99% CL BUGEY 90% CL* MiniBooNE 90% CL 10 MiniBooNE 99% CL MINOS+ & Bugey Combination 90% CL 2 m 1 ∆ -1 10 * GLoBES 2012 fit with new reactor fluxes, courtesy of P. Huber -2 10 -4 -3 -2 -1 10 10 10 10 1 2 θ sin 2 µ e P. Huber – p. 11
Reliability • Re-use of code, the more a code has been used in real world applications the less likely are severe bugs. • Extensive testing • Good documentation • Intuitive API with error checking P. Huber – p. 12
Reproducibility The information given in a publication or proposal is not sufficient to reproduce the sensitivity estimates. • General data storage and exchange format for the inputs ⇔ flexibility? • All implicit assumptions and approximations have to be documented, that includes the actual algorithms ⇔ accuracy of documentation? • Version control and archiving P. Huber – p. 13
Documentation Without good documentation, the best software is useless or will be after very short time (=memory decay constant of typical physicist). This is a general problem with legacy code! Document what you do – do what you document and make sure that the average user understands what is going on. Also documentation needs testing and debugging. P. Huber – p. 14
GLoBES history • development started 2004 – PH, M. Lindner, W. Winter • major effort went into documentation • first release August 2004 – version 2.0.0 • major bug fix release March 2005 version 2.0.11 • J. Kopp joined in July 2005 • January 2007 – version 3.0, addition of major features • >180 publications citing the GLoBES papers, creating a total of >3000 citations • fall 2012 – GLoBES 4.0 P. Huber – p. 15
Design considerations • GPL • C-library – very portable, easy to interface, numerically efficient • Unix style separation of functionality – freedom to design analysis and to use any graphics tools • Experiments are defined using AEDL – relatively complicated parser, transparent experiment definition • Pull approach for systematics – flexible and intuitive • Local minimization instead of grids – much faster P. Huber – p. 16
Features • Accurate treatment of systematical errors • Arbitrary matter profile & uncertainties • Arbitrary energy resolution function • Single and multiple experiment simulation • Simple χ 2 calculation • Inclusion of external input • Projection of χ 2 (minimization) • User-defined systematics, oscillation probability engine, priors • Full support for lists in AEDL • Interpolating functions in AEDL • . . . P. Huber – p. 17
GLoBES – overview GLoBES AEDL GLoBES User Interface Abstract Experiment C−library which loads AEDL− Definition Language AEDL−file(s) and file(s) provides functions to Defines Experiments simulate experiment(s) and modifies them Application software to compute high−level sensitivities, precision etc. P. Huber – p. 18
GLoBES – AEDL Energy− Cross Initial / final Resolution Section flavor, polarity function Energy Flux Channel dependent efficiencies Event rates P. Huber – p. 19
GLoBES – AEDL Signal . . . Channel 1 Background . . . Channel 2 Rule Signal + Backgrounds with systematics ∆χ 2 P. Huber – p. 20
GLoBES – AEDL . . . Rule 1 Rule 2 Rule 3 Experiment Σ ∆χ 2 P. Huber – p. 21
Summary GLoBES • is the only open source software of its kind • has withstood the test of time • is at the core of most strategy documents • completely in C • flexibility to deal with complex many detector setups and non-standard physics • v4.0 will greatly enhance the ability to deal with realistic systematics P. Huber – p. 22
Installation If you have Linux • Install GSL if you don’t have it already – ftp://ftp.gnu.org/gnu/gsl/ • Go to http://www.mpi- hd.mpg.de/personalhomes/globes/download/globes- 3.1.11.tar.gz – download GLoBES • ./configure make sudo make install P. Huber – p. 23
Installation If you have a Mac • Install GSL if you don’t have it already – ftp://ftp.gnu.org/gnu/gsl/ • Download http://www.mpi- hd.mpg.de/personalhomes/globes/download/globes- 3.1.11.tar.gz • unpack it and change into the directory created by this ./configure --disable-rpath --enable-no-binary=yes make make install P. Huber – p. 24
Installation – Mac issues On some of them, there are problems with the architecture (32 vs. 64 bit) since Mac OS tries (and fails) to allow mixing 32 and 64 bit code. The solution to this is to compiile both the GSL and GLoBES with the -m32 option enabled. Finally, people sometimes have trouble compiling the examples, which is due to some yet unresolved problem in our autoconf script. The solution is to modify the Makefile in the examples directory in such a way that globes-config --libs is not used. Instead, use whatever options globes-config --libs proposes, minus any --rpath options. P. Huber – p. 25
Recommend
More recommend