Skip to content

Generate and fit toy MC

RhiannonSJ edited this page Feb 23, 2021 · 6 revisions

Generate and fit a single toy MC experiment

%valor-run_fit 
   --config $VALOR_EXPT/config/analysis/(analysis_variant).xml   
   --physics-hypothesis eff_sterile_osc 
   --param-values-option prefit-errors 
   --osc-prior-file priors/void_priors.xml 
   --load-prior-from-file - 
   --fitter minuit+with_simplex_fallback 
   --mesg-thresholds verbosity.xml 
   --without-stat-fluct   
   --tag sbn_toy
   --run 1 
   --ntoymc 1
   (+ command line arguments specifying oscillation parameters)

For the calculation of exclusion sensitivities (eg. for muon-neutrino disappearance), first we generate an unoscillated toy experiment and, subsequently, we loop over a grid of relevant parameter values (eg. sin^2(2\theta_{\mu\mu}) and \Delta m^2_{41}) and, for every point on that grid, we fix the oscillation parameters to the coordinates of the grid point and float the remaining parameters trying to obtain the best possible agreement with the spectra of the unoscillated toy experiment.

For example, to run the muon-neutrino disappearance fit for an unoscillated toy experiment, used for estimating whether the bin sin^2(2\theta_{\mu\mu} = value1 and \Delta m^2_{41} = value2 is in the exluded region at a given C.L., add the following options:

   --input-dm41sq         0.         // generated toy experiment is unoscillated 
   --fix-at-dm41sq        value2     // during the fit set to the coordinate of the current grid point
   --fix-dm41sq                      // do not float during the fit - fix to the grid point coordinate
   --input-sinsq2thMuMu   0.         // generated toy experiment is unoscillated 
   --fix-at-sinsq2thMuMu  value1     // during the fit set to the coordinate of the current grid point
   --fix-sinsq2thMuMu                // do not float during the fit - fix to the grid point coordinate
   --input-sinsq2thMuE    0.         // irrelevant for numu disappearance, set to 0
   --fix-at-sinsq2thMuE   0.         // irrelevant for numu disappearance, set to 0
   --fix-sinsq2thMuE                 // irrelevant for numu disappearance, fix
   --input-sinsq2thEE     0          // irrelevant for numu disappearance, set to 0
   --fix-at-sinsq2thEE    0.         // irrelevant for numu disappearance, set to 0
   --fix-sinsq2thEE                  // irrelevant for numu disappearance, fix

For the calculation of an allowed region, for an injected oscillation signal, the procedure is the same as above but the generated toy experiment spectra are oscillated.

For example, to run the muon-neutrino disappearance fit of a toy experiment with an injected oscillation signal with parameters sin^2(2\theta_{\mu\mu} = signal_value1 and \Delta m^2_{41} = signal_value2, used for estimating whether the bin sin^2(2\theta_{\mu\mu} = value1 and \Delta m^2_{41} = value2 is within the allowed region at a given C.L., add the following options:

   --input-dm41sq         signal_value2  // generated toy experiment has given oscillation effects 
   --fix-at-dm41sq        value2         // during the fit set to the coordinate of the current grid point
   --fix-dm41sq                          // do not float during the fit - fix to the grid point coordinate
   --input-sinsq2thMuMu   signal_value1  // generated toy experiment has given oscillation effects 
   --fix-at-sinsq2thMuMu  value1         // during the fit set to the coordinate of the current grid point
   --fix-sinsq2thMuMu                    // do not float during the fit - fix to the grid point coordinate
   --input-sinsq2thMuE    0.             // irrelevant for numu disappearance, set to 0
   --fix-at-sinsq2thMuE   0.             // irrelevant for numu disappearance, set to 0
   --fix-sinsq2thMuE                     // irrelevant for numu disappearance, fix
   --input-sinsq2thEE     0              // irrelevant for numu disappearance, set to 0
   --fix-at-sinsq2thEE    0.             // irrelevant for numu disappearance, set to 0
   --fix-sinsq2thEE                      // irrelevant for numu disappearance, fix

To generate and fit an ensemble of toy experiments with systematic parameters randomised within their limits, add the argument: --tweak-parameter-set <set_name> to the arguments of the fit and change the value of ntoymc, like so:

%valor-run_fit 
   --config $VALOR_EXPT/config/analysis/(analysis_variant).xml   
   --physics-hypothesis eff_sterile_osc 
   --param-values-option prefit-errors 
   --osc-prior-file priors/void_priors.xml 
   --load-prior-from-file - 
   --fitter minuit+with_simplex_fallback 
   --mesg-thresholds verbosity.xml 
   --without-stat-fluct   
   --tag sbn_toy
   --run 1 
   --ntoymc 1
   --tweak-parameter-set (parameter_set_identifier)
   (+ command line arguments specifying oscillation parameters)

Clone this wiki locally