Near Detector Optimization Task Force Steve Brice, Daniel Cherdack, Kendall Mahn
Draft Charge to the Task Force ❑ The near detector optimization task force is charged to: ▪ Develop GEANT4 simulations of the reference design near detector and possible alternatives ▪ Perform a full end-to-end simulation connecting the measurements in the near detector to the far detector systematics using, for example, the VALOR framework ▪ Evaluate the potential benefits of augmenting the reference design with • a LAr-TPC • the use of a High Pressure Gaseous TPC ▪ Produce a first report on their findings to the DUNE Technical Board by September 2016 and a final report by March 2017. Near Detector Optimization Task Force 2
Simulation and Analysis Path Near Detector Optimization Task Force 3 3
VALOR: ND Constraints - Costas Andreopoulos Near Detector Optimization Task Force 4
Points of Contact Flux: Laura Fields Infrastructure: Robert Hatcher Cross-Section Models and Systematics: Lorena Escudero FGT simulation: Tyler Alion & Chris Marshall LAr simulation: Sarah Lockwitz & James Sinclair GAr simulation: Justo Martín-Albo VALOR: Steve Dennis & Lorena Escudero & Costas Andreopoulos FD Simulation: Tingjun Yang & Tyler Alion FD Fit: Daniel Cherdack Figures of Merit: Brian Rebel • The points of contact are fully populated, but there is need for more effort within each of the 3 ND simulation efforts • Particularly improving the recon modeling and sample selection and generally validating what’s been done Near Detector Optimization Task Force 5
Phase 1 - focus on machinery Sept 2015 - Jan 2016 • Milestone 1: First complete run through of the machinery (before Arlington meeting) • Jan 2016 Phase 2 - incrementally add the necessary physics and improve simulations Jan 2016 - Sept 2016 • Milestone 2: 2nd run through (before SURF meeting) • April 2016 • Milestone 3: 3rd run through to generate material for initial report (before FNAL meeting) • August 2016 • Milestone 4: Initial Report • September 2016 Phase 3 - final improvements to the physics and simulations Sept 2016 - Mar 2017 • Milestone 5: Final run through to generate material for final report (before CERN meeting) • December 2016 • Milestone 6: Final Report • March 2017 Near Detector Optimization Task Force 6
Top Concerns and Current Focus We are working to alleviate several areas of concern before the 4th Run Through... ● Is the cross-section systematic model appropriately complex and descriptive of our actual level of uncertainty? ○ Commissioned a review by T2K experts - following up on recommendations with changes to the model ● Is the reconstruction cheating appropriately complex? ○ and is it at a comparable level of complexity for the 3 ND options? ○ The leaders of each ND simulation effort have agreed on a common level of complexity and are executing ● Uncertainties appear to be overconstrained ○ Is it a bug? ○ Carrying out battery of checks including using the same VALOR tool to replicate T2K constraints and compare to T2K official numbers ○ Do the systematic errors allow for too much constraint by not having enough wiggle room? ○ Is the overconstraint come from having reconstruction and event selection that is unrealistically good? ● Are the constrained (anti)correlations appropriate? ○ Seems like too little anti-correlation between flux and cross-section. Is this true? ○ How do we check? - coupled to the checks being performed in previous bullet Near Detector Optimization Task Force 7
A Priori Cross-Section Uncertainties - VALOR (Lorena Escudero) Near Detector Optimization Task Force 8
ND Task Force Approach to Reconstruction “Cheating but not lying” ● We have a generic problem across the 3 ND options and the FD - “How do you provide the best mimic of the reconstruction and PID algorithms we will have 10 years from now?” 1) Use the best algorithms we have now 2) Use our experience with past detectors and algorithms to appropriately smear truth quantities ● For all 4 reconstruction efforts (3ND + 1FD) we are evaluating each required observable individually, and: ○ Use 1) whenever practical, although this may be limited, esp for the 3 ND options ○ Use smearing that is well informed by studies of the full GEANT4 simulations and consistent with 2) ○ Rely upon 2) fully when extracting relevant information from the full GEANT4 simulations presents difficulties beyond the scope of the TF Near Detector Optimization Task Force 9
The Output Uncertainties - Steve Dennis Very Preliminary! Near Detector Optimization Task Force 10
The Output Correlation Matrix - Steve Dennis Pre-fit Post-fit FGT correlation matrix correlation matrix Very Preliminary! Near Detector Optimization Task Force 11
The Output Correlation Matrix - Steve Dennis Pre-fit Post-fit Gaseous Ar correlation matrix TPC correlation matrix Very Preliminary! Near Detector Optimization Task Force 12
The Output Correlation Matrix - Steve Dennis Pre-fit Post-fit LAr TPC correlation matrix correlation matrix Very Preliminary! Near Detector Optimization Task Force 13
Final Fits - Steve Dennis & Lorena Escudero & Dan Cherdack � CP sensitivity for normal (left) and inverted (right) mass ordering Very Preliminary! Near Detector Optimization Task Force 14
Figures of Merit - Brian Rebel How the Detectors Enable Oscillation Physics ● Sensitivity to δCP using each of the ND options How the Detectors Perform in the Beam ● Number of interactions per POT ● Pile-up in detector due to beam intensity ● Fraction of energy shared between neutrino interactions in the same beam spill ● Fraction of energy shared between cosmic rays and neutrino interactions How the Detectors Enable Physics Generally ● Vertex position resolution of the detector (can we tell which nuclear target was struck) ● Energy resolution for ○ EM showers ○ Hadronic showers ○ Minimum ionizing particles ○ Total neutrino interaction ● Acceptance of final state particles as a function of energy and direction ● Fraction of neutrino interactions on each species of nuclear target ● Fraction of energy contained in the detector as a function of the vertex distance from detector edge ● Purity for distinguishing different interaction types as a function of energy ● Energy thresholds for observing different particle species (p, n, π) Near Detector Optimization Task Force 15
Summary ● The 3rd Run Through of the processing chain is complete ○ Documented at ■ https://docs.google.com/document/d/1OarI91-vwgTIMS2Lm8ydLMxnISd_P0IthjkVoiqefCo/edit# ○ VALOR technote coming soon with ID 1712 ● The Initial Report of the Task Force is complete and can be found at http://docs.dunescience.org:8080/cgi-bin/ShowDocument?docid=1792 or https://www.overleaf.com/5943230rkdrtw#/20548076/ ● The task force has made great strides with each step in the chain having strong leadership and significant effort ● 4th Run Through scheduled just before the next Collaboration Meeting ● Final Report due in March 2017 Near Detector Optimization Task Force 16
Backups Near Detector Optimization Task Force 17
Axioms • The Task Force will evaluate ND options based just on science • Budgetary and other concerns can wait • Task force charged with making science based recommendations and any decisions by the collaboration will likely include other factors. • Near Detector performance is judged by its ability to improve the sensitivity of DUNE to CPV • Any Near Detector optimized this way will be very capable of the other analysis envisaged for the Near Detector • Sensitivity to other physics will be a secondary consideration; cannot degrade oscillation physics • The ND should allow for measurements on the same target nucleus as the FD (Ar) • T2K oscillation systematics increased by target nucleus differences • Should include a clear and proven path to extracting cross section measurements on the target nucleus Near Detector Optimization Task Force 18
How to Optimize the CP Violation Oscillation Analysis • Adopt and extend the approach of the experiment that is presently at the cutting edge of this work – T2K • Use the VALOR package for ND fits • Inputs • Event samples from simulations of the Near Detector options • Detailed systematic uncertainties (spectral changes, and priors) Outputs • • Fits of all possible nuisance parameters for a FD fit • A covariance matrix that encodes all prior and correlations • Oscillation parameter fits with FD event samples • Several current tools in use and development • A full VALOR ND+FD fit is also a good possibility Near Detector Optimization Task Force 19
Recommend
More recommend
