GENIE Status Gabriel N. Perdue Simulations for Neutrinos 30 - - PowerPoint PPT Presentation

SLIDE 1

Gabriel N. Perdue Simulations for Neutrinos 30 January 2017

GENIE Status

SLIDE 2

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017

Overview

• Recent release: GENIE 2.12.2
• Planned releases:
• GENIE 3.0
• GENIE 4.0

2

SLIDE 3

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017

GENIE 2.12.2

• Non-physics changes - but important!
• Moved z-expansion entries in EGSyst enum to the end of the list

so as to preserve the order of (prior) subsequent values. This keeps stored re-weighting values from 2.10.x backwards compatible.

• Note, however, that samples generated with 2.12.0 will have problems

using stored re-weights with 2.12.2. If you are in this position, we can show you how to patch GENIE to keep it compatible with your weights

• n disk, but we encourage you to do your next large production sample

with the new enum ordering.

• Add SingleKaon library to the set loaded by the genie-config

script.

• Name for TGraph of strange cross sections in ROOT files fixed.

3

SLIDE 4

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017

GENIE 3.0

• The basic idea behind GENIE 3.0 is to choose a set of new

default models based on studying agreement with data and considering recent advances on the theory side. We also plan to include technical changes that make changing the physics model simpler (in addition to bug fixes, etc.).

• We want to choose a set of models that describe GENIE for the

parameter tuning stage to follow. The result of that work will be GENIE 4.0.

• We may continue to consider model choices as we begin tuning

for 4.0.

• We will make comprehensive comparisons to experimental data

available as part of the release.

• We are currently testing five basic variations.

4

SLIDE 5

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017

GENIE 3.0 - versions

• The first two configurations are very close to the current default, with one as close

as possible and the other only additionally including the "Empirical" MEC model.

• The next update removes the Llewellyn Smith QE model in favor of the Nieves et al

model, with RPA and Coulomb effects disabled. It additionally updates the HA FSI model to the most recent version.

• The next version represents our "best theory model"
• Local Fermi Gas nuclear model
• z-expansion for the axial form factor
• Nieves et al CCQE with RAP and Coulomb effects
• Nieves et al MEC
• Berger-Sehgal coherent pion
• Berger-Sehgal resonance model
• Most recent HA FSI model
• We will also have a version with all these changes but using the HN (cascade) FSI

model.

5

SLIDE 6

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017

GENIE 3.0

• Evaluating ~400 page plot

book.

• Lots of effort currently being

invested in understanding signal definitions, normalization, etc.

• Some datasets pull the

choices in one direction, other datasets pull in another.

• We will make the comparison

plots available with the release so experiments have a sense for the impact of changes.

6

µ

θ Cos

1 − 0.5 − 0.5 1 /GeV/n] 2 cm

• 38

[10 µ T ∂ / µ θ Cos ∂ / σ 2 ∂

0.5 1 1.5

miniboone_nuccqe_2010 trunk:G00_00a:miniboone_fhc trunk:G00_00b:miniboone_fhc trunk:G16_01a:miniboone_fhc trunk:G16_02b:miniboone_fhc [0.2; 0.3 ] GeV ∈ µ T µ

θ Cos

1 − 0.5 − 0.5 1 /GeV/n] 2 cm

• 38

[10 µ T ∂ / µ θ Cos ∂ / σ 2 ∂

0.5 1 1.5

miniboone_nuccqe_2010 trunk:G00_00a:miniboone_fhc trunk:G00_00b:miniboone_fhc trunk:G16_01a:miniboone_fhc trunk:G16_02b:miniboone_fhc [0.3; 0.4 ] GeV ∈ µ T µ

θ Cos

1 − 0.5 − 0.5 1 /GeV/n] 2 cm

• 38

[10 µ T ∂ / µ θ Cos ∂ / σ 2 ∂

0.5 1 1.5 2

miniboone_nuccqe_2010 trunk:G00_00a:miniboone_fhc trunk:G00_00b:miniboone_fhc trunk:G16_01a:miniboone_fhc trunk:G16_02b:miniboone_fhc [0.4; 0.5 ] GeV ∈ µ T µ

θ Cos

1 − 0.5 − 0.5 1 /GeV/n] 2 cm

• 38

[10 µ T ∂ / µ θ Cos ∂ / σ 2 ∂

1 2 3

miniboone_nuccqe_2010 trunk:G00_00a:miniboone_fhc trunk:G00_00b:miniboone_fhc trunk:G16_01a:miniboone_fhc trunk:G16_02b:miniboone_fhc [0.5; 0.6 ] GeV ∈ µ T

MiniBooNE CCQE

~Current default (black) ~Current default + MEC (red) Simple changes (green) Best theory, hN FSI (blue)

SLIDE 7

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017

GENIE 4.0 - GENIE + Professor

7

Slide by C. Andreopoulos

Energy Frontier

SLIDE 8

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017

GENIE 4.0

8

Fundamentally, the idea of Professor is to:

• Reduce the exponentially expensive process of brute-force tuning to a scaling closer to a

power law in the number of parameters.

• Allow for massive parallelisation and systematically improve the scan results by use of a

deterministic parameterisation of the generator’s response to changes in the steering parameters. [H.Schulz]

Slide by C. Andreopoulos

SLIDE 9

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017

GENIE 4.0

9

GENIE / Professor interface

Bin #1 description Bin #2 description Bin #3 description Bin #4 description Bin #5 description

...

Bin #6 description Bin #7 description Bin #8 description Bin #9 description Bin #10 description Bin #11 description GENIE comparison 1 GENIE comparison 2 GENIE comparison 3

Pass to Professor:

• A single 1-D array with all data
• A single 1-D array with all predictions (per given model configuration and parameter values)
• A single 2-D covariance matrix

Slide by C. Andreopoulos

SLIDE 10

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017

GENIE 4.0

10

Current status

Batch system readiness

• Main MC production site at Lyon now functional / scripts available for that site

○ Several production tasks carried out at Lyon ■ Processing for v3.0.0 characterization and for Professor/GENIE tuning exercise

• An additional substantial resource at RAL/PPD Tier2 also functional / available as a backup for

more CPU-intensive stages of the Professor/GENIE tuning exercise

Code development

• Professor able to produce UserPhysicsOptions.xml steering cards - Done
• Automated GENIE MC production scripts - Completed and tested
• Scripts for automated configuration of GENIE Comparisons - on going
• Code for linearizing data and MC - Almost completed for MinibooNE and T2K comparisons
• Finalizing output data format (data, predictions, covariances) for Professor - Starting this week

Next steps

• Complete migration of integrated cross-section (bubble chamber) and

hadronization comparisons in the new framework

○ Compatibility with the new system / scripts ○ Add \chisquare calculations for quantitative model characterizations, as done for MB/T2K ○ Linearise data and predictions to append to standardised GENIE output for Professor and include in tunes

Slide by C. Andreopoulos

• n-going

...lots of progress completed here Borrowed slides are bit old here...

SLIDE 11

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017

Schedule

• GENIE 3.0 "soon"
• finalizing comparisons
• deciding on technical changes
• GENIE 4.0 is on an aggressive schedule also.
• the goal is this calendar year
• lots of technical progress already at Liverpool
• but many thorny physics issues (dataset tensions, experiment-

theory tensions, opinions about model selection, etc.) to sort out...

11

SLIDE 12

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017 12

Back-up

SLIDE 13

Gabriel Perdue | GENIE news | Simulations for Neutrinos 30 January 2017

New models in GENIE 2.12

• Valencia MEC
• Local Fermi Gas (nuclear model)
• Nieves et al CCQE model
• Oset modifications to cascade (hN) FSI model
• Kaon FSI
• z-expansion form of the Axial Form Factor for QE
• QEL hyperon production
• Berger-Sehgal coherent pion model
• Updated Rein diffractive pion model
• Energy-dependent MA model (Kuzmin-Naumov)

13