add testcase response model

[hs3bot]

response model

Submitted by Carsten Burgardq cburgard@cern.ch

This test implements a simple Gaussian detector response,

$$R(\mathrm{reco}\mid\mathrm{truth}) = \mathcal{N}(\mathrm{reco}; \mathrm{truth}, \sigma),$$

and verifies that the same model component can be reused in different analyses without modification. The response distribution itself is agnostic to the semantic roles of its variables—it does not distinguish between observables, parameters of interest, or nuisance parameters. Those roles are assigned only by the surrounding analysis configuration.

The test instantiates two analyses from the same response model:

• Simulation: truth is the parameter of interest, reco is the observable.
• Calibration: reco is the parameter of interest, truth is the observable.

The underlying Gaussian distribution is identical in both cases; only the analysis metadata changes. This verifies that the detector response remains independent of the role assigned to each variable by the analysis.

Field	Value
Testcase name	response model
Testcase ID	response_model
Schema version	0.2.9
Description	This test implements a simple Gaussian detector response,

$$R(\mathrm{reco}\mid\mathrm{truth}) = \mathcal{N}(\mathrm{reco}; \mathrm{truth}, \sigma),$$

and verifies that the same model component can be reused in different analyses without modification. The response distribution itself is agnostic to the semantic roles of its variables—it does not distinguish between observables, parameters of interest, or nuisance parameters. Those roles are assigned only by the surrounding analysis configuration.

The test instantiates two analyses from the same response model:

• Simulation: truth is the parameter of interest, reco is the observable.
• Calibration: reco is the parameter of interest, truth is the observable.

The underlying Gaussian distribution is identical in both cases; only the analysis metadata changes. This verifies that the detector response remains independent of the role assigned to each variable by the analysis. |
| Source | https://docs.google.com/presentation/d/1MsB_mgA46xE61NEKPJZXI8E7_E64bjBhxKmuU1yOnac/edit?slide=id.g3e80de5473d_0_233#slide=id.g3e80de5473d_0_233 |
| Modified from source | False |
| Notes | |
| Reference backend | stalbrec/hs3-testsuite-roofit-backend:e3a793 |

NLL scan simRecoObs

Field	Value
Target	pdf=response, data=dataRecoObs_response
Parameters	['truth']
Points	[[103.0], [113.0], [123.0], [133.0], [143.0]]

NLL scan simTruthObs

Field	Value
Target	pdf=response, data=dataTruthObs_response
Parameters	['reco']
Points	[[98.0], [108.0], [118.0], [128.0], [138.0]]