Poisson constraint in HistFactory

[eled]

Hi,

I have a problem in building a model with HistFactory. I would like to set a Poisson constraint on one of the normalisation constant of a sample.
What I typically do is for example:

   RooStats::HistFactory::Sample bb( "bb", hists[h_name]->GetName(), filename);
    bb.ActivateStatError();
    bb.SetNormalizeByTheory(false);
    bb.AddNormFactor("nbb_MC_SR", n_MC_SR, 0, 10000, true );
    bb.AddNormFactor("nbb_MC_CR", 1./n_MC_CR, 0, 10000, true );
    bb.AddNormFactor("bbyield_CR", fit_value.value , 0, 10000, true );
    meas.AddConstantParam("nbb_MC_SR");
    meas.AddConstantParam("nbb_MC_CR");
    meas.AddConstantParam("bbyield_CR");
    bb.AddOverallSys("nbb_MC_SR", 1-n_MC_SR_err, 1+n_MC_SR_err);
    bb.AddOverallSys("nbb_MC_CR", 1-n_MC_CR_err, 1+n_MC_CR_err);
    bb.AddOverallSys("bbyield_CR", 1-fit_value.error/fit_value.value, 1+fit_value.error/fit_value.value);

In order to set the Poisson constrain I simply add the following line:
meas.AddGammaSyst(“nbb_MC_SR”, n_MC_SR_err);

When I perform the fit, there are a lot of errors related to the Poisson constraints and it badly converges.
Am I doing something wrong in setting the Poisson constraints?

Thanks in advance,
Elena

[StephanH]

Hi Elena,

what are the errors?

Also, what size is the uncertainty you assign?

[eled]

Hi Stephan,

I am assigning as value of the normalisation factor 1 and as uncertainty 1.

The error I get is the following:

[#0] ERROR:Integration -- RooNumIntFactory::createIntegrator: No integration method has been defined for an open ended 2-dimensional integral
[#0] ERROR:Integration -- RooRealIntegral::beta_nW_MC_SRConstraint_Int[beta_nW_MC_SR,nom_beta_nW_MC_SR,theta_nW_MC_SR]: failed to create valid integrator.
[#0] ERROR:Integration -- RooRealIntegral::beta_nW_MC_SRConstraint_Int[beta_nW_MC_SR,nom_beta_nW_MC_SR,theta_nW_MC_SR]:evaluate: cannot initialize numerical integrator
[#0] ERROR:Integration -- RooNumIntFactory::createIntegrator: No integration method has been defined for an open ended 2-dimensional integral
[#0] ERROR:Integration -- RooRealIntegral::beta_nbb_MC_SRConstraint_Int[beta_nbb_MC_SR,nom_beta_nbb_MC_SR,theta_nbb_MC_SR]: failed to create valid integrator.
[#0] ERROR:Integration -- RooRealIntegral::beta_nbb_MC_SRConstraint_Int[beta_nbb_MC_SR,nom_beta_nbb_MC_SR,theta_nbb_MC_SR]:evaluate: cannot initialize numerical integrator
[#0] WARNING:Minization -- RooMinimizerFcn: Minimized function has error status.
Returning maximum FCN so far (-1e+30) to force MIGRAD to back out of this region. Error log follows
Parameter values: alpha_Wyield_CR=-0.0181367, alpha_Zyield=0.172842, alpha_bbyield_CR=-0.00542618, alpha_eff_ratio=-1.94981e-06, alpha_nW_MC_CR=-2.43615e-05, alpha_nbb_MC_CR=-4.05417e-05, alpha_norm_yield=-2.34374e-05, alpha_qcdyield=-0.137655, beta_nW_MC_SR=0.746232, beta_nbb_MC_SR=0.771905, br=9.53674e-07, gamma_stat_SR_bin_0=0.997355, gamma_stat_SR_bin_1=0.999373, gamma_stat_SR_bin_10=0.999581, gamma_stat_SR_bin_11=0.999125, gamma_stat_SR_bin_2=0.999577, gamma_stat_SR_bin_3=0.999597, gamma_stat_SR_bin_4=0.999503, gamma_stat_SR_bin_5=0.998827, gamma_stat_SR_bin_6=0.999339, gamma_stat_SR_bin_7=1.02814, gamma_stat_SR_bin_8=0.999121, gamma_stat_SR_bin_9=1.04237
RooGamma::beta_nW_MC_SRConstraint[ x=beta_nW_MC_SR mean=k_nW_MC_SR beta=theta_nW_MC_SR mu=0 ]
     p.d.f normalization integral is zero or negative @ x=beta_nW_MC_SR=0.746232 +/- 0.667738, mean=k_nW_MC_SR=2, beta=theta_nW_MC_SR=1, mu=0=0
     getLogVal() top-level p.d.f evaluates to zero @ x=beta_nW_MC_SR=0.746232 +/- 0.667738, mean=k_nW_MC_SR=2, beta=theta_nW_MC_SR=1, mu=0=0
RooGamma::beta_nbb_MC_SRConstraint[ x=beta_nbb_MC_SR mean=k_nbb_MC_SR beta=theta_nbb_MC_SR mu=0 ]
     p.d.f normalization integral is zero or negative @ x=beta_nbb_MC_SR=0.771905 +/- 0.548173, mean=k_nbb_MC_SR=2, beta=theta_nbb_MC_SR=1, mu=0=0
     getLogVal() top-level p.d.f evaluates to zero @ x=beta_nbb_MC_SR=0.771905 +/- 0.548173, mean=k_nbb_MC_SR=2, beta=theta_nbb_MC_SR=1, mu=0=0

Thanks,
Elena

[StephanH]

I don’t see what’s causing this. Could you attach a short macro for testing?