NEUT model improvements and external data fits Tom Feusels for T2K - - PowerPoint PPT Presentation

▶
neut model improvements and external data fits
SMART_READER_LITE
LIVE PREVIEW

NEUT model improvements and external data fits Tom Feusels for T2K - - PowerPoint PPT Presentation

Oct 31, 2023 198 likes •484 views

NEUT model improvements and external data fits Tom Feusels for T2K Collaboration University of British Columbia NuFact 2015 Rio de Janeiro Aug 13, 2015 T. Feusels (UBC) External data fits 13/08/2015 1 / 27 Table of Contents Introduction

slide-1
SLIDE 1

NEUT model improvements and external data fits

Tom Feusels for T2K Collaboration

University of British Columbia

NuFact 2015 Rio de Janeiro Aug 13, 2015

  • T. Feusels (UBC)

External data fits 13/08/2015 1 / 27

slide-2
SLIDE 2

Table of Contents

1

Introduction

2

NEUT Spectral Function Nieves 2p2h RPA

3

External Data Fits CCQE fits CC1pi fits FSI fits

4

Summary and Outlook

  • T. Feusels (UBC)

External data fits 13/08/2015 2 / 27

slide-3
SLIDE 3

Table of Contents

1

Introduction

2

NEUT Spectral Function Nieves 2p2h RPA

3

External Data Fits CCQE fits CC1pi fits FSI fits

4

Summary and Outlook

  • T. Feusels (UBC)

External data fits 13/08/2015 3 / 27

slide-4
SLIDE 4

T2K Oscillation Analysis Overview

NEUT

  • CCQE
  • pion production
  • CC coherent
  • FSI cascade model
  • ...

External Data

  • MiniBooNE
  • MinerνA
  • ANL
  • BNL
  • π-C scattering
  • ...

model parameters priors & covariance matrix

ND280 Data

Neutrino

  • CC0π
  • CC1π
  • CCother

Anti-Neutrino

  • CC1track
  • CCNtrack

extrapolation to Super-K

Near Detector Fit

With covariance matrix:

  • Detector systematics
  • Flux systematics

Super-K Data

  • 1 Ring e-like
  • 1 Ring mu-like

Oscillation Analysis

+ Super-K detector systematics see Talk K. Mahn (Tue Aug 11) see Talk K. Duffy (Tue Aug 11)

  • T. Feusels (UBC)

External data fits 13/08/2015 4 / 27

slide-5
SLIDE 5

Table of Contents

1

Introduction

2

NEUT Spectral Function Nieves 2p2h RPA

3

External Data Fits CCQE fits CC1pi fits FSI fits

4

Summary and Outlook

  • T. Feusels (UBC)

External data fits 13/08/2015 5 / 27

slide-6
SLIDE 6

The NEUT Neutrino Generator

Default (NEUT < 5.3.2) Llewellyn-Smith model for 1p1h CCQE interactions. Rein-Sehgal model for resonant and coherent pion production. BBBA05 form factor with default MA of 1.21 GeV/c2 for CCQE and resonant production. GRV94/GRV98 pdfs with Bodek-Yang corrections for DIS. Relativistic Fermi Gas (RFG) from Smith-Moniz with Pauli-blocking as initial state nuclear model. Latest improvements (NEUT 5.3.2): Initial state nuclear model: Spectral function model by Benhar et al. implemented. 2p2h model of Nieves implemented. Random Phase Approximation (RPA). More realistic pion form factors from Graczyk and Sobczyk (Phys. Rev. D77, 053001 (2008)).

  • T. Feusels (UBC)

External data fits 13/08/2015 6 / 27

slide-7
SLIDE 7

Spectral Function

Spectral Function in nucleon momentum and removal energy by Benhar et al. (Phys. Rev. C62, 034304 (2000)). Standard Impulse Approximation used. Available for C, O and Fe. Pauli-blocking with hard cut-off. Effective Spectral Function by Bodek et al. recently added to NEUT but not yet candidate model for T2K.

  • T. Feusels (UBC)

External data fits 13/08/2015 7 / 27

slide-8
SLIDE 8

Nieves 2p2h model

Interaction with pair of short range correlated nucleons. Use pre-calculated tables based on Nieves et al. (Phys. Rev. C83, 045501 (2011)). Only lepton kinematics predicted, hadronic part through Sobczyk model. High energy extension based on Gran et al. (Phys. Rev. D88, 113007 (2013)).

  • T. Feusels (UBC)

External data fits 13/08/2015 8 / 27

slide-9
SLIDE 9

Random Phase Approximation (RPA)

ν ¯ ν Nuclear screening effect due long range nucleon-nucleon correlations. NEUT implementation depends on Q2 and Eν, based on Nieves et al. (Phys. Rev. C83, 045501 (2011)). Assumes Local Fermi Gas, but also valid for Relativistic Fermi Gas. No Spectral Function RPA correction.

  • T. Feusels (UBC)

External data fits 13/08/2015 9 / 27

slide-10
SLIDE 10

Table of Contents

1

Introduction

2

NEUT Spectral Function Nieves 2p2h RPA

3

External Data Fits CCQE fits CC1pi fits FSI fits

4

Summary and Outlook

  • T. Feusels (UBC)

External data fits 13/08/2015 10 / 27

slide-11
SLIDE 11

CCQE Fit procedure and datasets

Procedure:

1

Float model parameters (MA, 2p2h normalization, pF ) and MiniBooNE normalization in χ2 fit within each model.

2

Test agreement of complete dataset with model through standard Pearson χ2 Goodness of Fit test.

3

Use Parameter Goodness of Fit (PGoF) test for consistency between datasets within each model.

4

Rescale parameter errors to span differences between datasets according to PGoF test.

5

Apply PGoF procedure to an ND280 data fit and external data fits to ensure prior errors cover both the external fit results and ND280 data at 1σ. Datasets: MiniBooNE:

d2σ dTµd cos θµ for ν and ¯

ν (CCQE-corrected data from Phys. Rev. D81, 092005 (2010) and Phys. Rev. D88, 032001 (2013)) MINERνA:

dσ dQ2

QE for ν and ¯

ν with restricted phase space θµ ≤ 20◦, including cross-correlations (Phys. Rev. Lett. 111, 022502 (2013) and Phys. Rev. Lett. 111, 022501 (2013)) Models: RFG + RPA + MEC vs SF + MEC.

  • T. Feusels (UBC)

External data fits 13/08/2015 11 / 27

slide-12
SLIDE 12

Neutrino CCQE fits: MiniBooNE and MINERνA

(GeV)

µ

T /GeV)

2

cm

  • 39

10 × (

µ

θ dcos

µ

dT σ

2

d

= 37.1 (97.5)

2

χ SF+MEC = 37.9 (97.8)

2

χ RFG+RPA+MEC DATA

0.5 1 1.5 2 2 4 6 8 10 12 14 16 18 20

< 0.7

µ

θ 0.6 < cos

0.2 0.4 0.6 0.8 1 1 2 3 4 5 6 7

< 0.0

µ

θ

  • 1. < cos

0.5 1 1.5 2 2 4 6 8 10 12 14 16 18 20 22

< 0.8

µ

θ 0.7 < cos

0.2 0.4 0.6 0.8 1 1 2 3 4 5 6 7 8 9

< 0.3

µ

θ 0.0 < cos

0.5 1 1.5 2 5 10 15 20 25

< 0.9

µ

θ 0.8 < cos

0.2 0.4 0.6 0.8 1 2 4 6 8 10

< 0.6

µ

θ 0.3 < cos

0.5 1 1.5 2 5 10 15 20 25

< 1.0

µ

θ 0.9 < cos

)

2

(GeV

QE 2

Q

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

)

2

/GeV

2

(cm

2

/dQ σ d

2 4 6 8 10 12 14 16

  • 39

10 ×

= 26.8 (97.8)

2

χ RFG+RPA+MEC = 20.6 (97.5)

2

χ SF+MEC DATA

MiniBooNE MINERνA χ2 is from each dataset while total χ2 in brackets. Dashed line: without MiniBooNE normalization terms. Even though much more bins for MiniBooNE, fit not dominated by MiniBooNE data.

  • T. Feusels (UBC)

External data fits 13/08/2015 12 / 27

slide-13
SLIDE 13

Anti-Neutrino CCQE fits: MiniBooNE and MINERνA

(GeV)

µ

T /GeV)

2

cm

  • 39

10 × (

µ

θ dcos

µ

dT σ

2

d

= 27.5 (97.5)

2

χ SF+MEC = 25.2 (97.8)

2

χ RFG+RPA+MEC DATA

0.5 1 1.5 2 0.5 1 1.5 2 2.5 3 3.5 4

< 0.7

µ

θ 0.6 < cos

0.2 0.4 0.6 0.8 1 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22

< 0.0

µ

θ

  • 1. < cos

0.5 1 1.5 2 1 2 3 4 5 6

< 0.8

µ

θ 0.7 < cos

0.2 0.4 0.6 0.8 1 0.1 0.2 0.3 0.4 0.5 0.6 0.7

< 0.3

µ

θ 0.0 < cos

0.5 1 1.5 2 2 4 6 8 10

< 0.9

µ

θ 0.8 < cos

0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 1.2 1.4

< 0.6

µ

θ 0.3 < cos

0.5 1 1.5 2 2 4 6 8 10 12 14

< 1.0

µ

θ 0.9 < cos

)

2

(GeV

QE 2

Q

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

)

2

/GeV

2

(cm

2

/dQ σ d

2 4 6 8 10 12 14

  • 39

10 ×

= 14.5 (97.8)

2

χ RFG+RPA+MEC = 13.8 (97.5)

2

χ SF+MEC DATA

MiniBooNE MINERνA χ2 is from each dataset while total χ2 in brackets. Dashed line: without MiniBooNE normalization terms. Even though much more bins for MiniBooNE, fit not dominated by MiniBooNE data.

  • T. Feusels (UBC)

External data fits 13/08/2015 13 / 27

slide-14
SLIDE 14

Fit results

Fit type χ2/DOF MA (GeV) MEC (%) pF (MeV)

  • Rel. RPA

97.8/195 1.15 ± 0.03 27 ± 12 223 ± 5 Non-rel. RPA 117.9/195 1.07 ± 0.03 34 ± 12 225 ± 5 SF + MEC 97.5/196 1.33 ± 0.03 0 (at limit) 234 ± 4 Best fit χ2 good for both RFG + Rel. RPA + MEC and SF + MEC, but no MB correlations so Gaussian statistics invalid. Parameter Goodness of Fit: ¯ χ2 := χ2

tot − D

  • i=1

χ2

i , where χ2 i is the result of an individual fit

to Dataset i. Number of degrees of freedom: PPGoF :=

D

  • i=1

Pi − Ptot. Test compatibility between datasets in framework of model (Phys. Rev. D65, 014011 (2001)) If different datasets favour different parameter values, model cannot describe all the data consistently.

  • T. Feusels (UBC)

External data fits 13/08/2015 14 / 27

slide-15
SLIDE 15

PGoF tests: RFG + RPA + MEC

Datasets χ2/DOF Goodness of Fit(%) χ2/DOF PGoF (%) All 97.8/195 100.00 17.9/6 0.66 MINERνA (ν vs ¯ ν) 23.4/13 3.74 1.0/3 79.03 MiniBooNE (ν vs ¯ ν) 58.3/179 100.00 2.0/3 58.69 ν (MB vs MIN) 62.6/127 100.00 16.1/3 0.11 ¯ ν (MB vs MIN) 38.5/65 99.64 6.1/3 10.75 MINERνA vs MiniBooNE 97.8/195 100.00 15.9/3 0.12 ν vs ¯ ν 97.8/195 100.00

  • 3.3/3

100.0 Largest tension between MINERνA and MiniBooNE neutrino results. Note: PGoF test requires uncorrelated datasets, so result for MINERνA’s PGoF for ν vs ¯ ν is too good.

  • T. Feusels (UBC)

External data fits 13/08/2015 15 / 27

slide-16
SLIDE 16

PGoF tests: SF + MEC

Datasets χ2/DOF Goodness of Fit(%) χ2/DOF PGoF (%) All 97.5/196 100.00 41.1/4 0.0 MINERνA (ν vs ¯ ν) 12.6/13 47.75 1.0/2 59.87 MiniBooNE (ν vs ¯ ν) 50.2/180 100.00 6.5/2 3.85 ν (MB vs MIN) 54.8/128 100.00 25.1/3 0.0 ¯ ν (MB vs MIN) 34.1/65 99.64 8.5/2 1.40 MINERνA vs MiniBooNE 97.5/196 100.00 34.6/2 0.12 ν vs ¯ ν 97.5/196 100.00 8.5/2 1.39 Largest tension between MINERνA and MiniBooNE neutrino results. SF + MEC model finds much worse agreement between datasets. ⇒ Choose RFG+RPA+MEC as default T2K model.

  • T. Feusels (UBC)

External data fits 13/08/2015 16 / 27

slide-17
SLIDE 17

Parameter errors and covariance matrix for RFG+RPA+MEC

∆χ2 = 1 for 1σ errors. Scale parameter errors to cover all datasets with PGoF test, ie. scale errors to make sure ∆χ2 = χ2

PGoF /DOF.

Most tension between MiniBooNE and MINERνA PGoF value, hence scale by

  • 15.9/3 ≈ 2.3.

Additional inflation of errors when compared to ND280 data, based on PGoF between ND280 vs MiniBooNE vs MINERνA by a factor of

  • 38.0/6 ≈ 2.5 wrt initial MINUIT

errors.

Fit type χ2/DOF MA (GeV) MEC (%) pF (MeV) Unscaled 97.8/195 1.15 ± 0.03 27 ± 12 223 ± 5 PGoF scaling 1.15 ± 0.06 27 ± 27 223 ± 11 Rescaled 1.15 ± 0.07 27 ± 29 223 ± 12

  • T. Feusels (UBC)

External data fits 13/08/2015 17 / 27

slide-18
SLIDE 18

Procedure for pion production tuning

1

Update vector and axial form factors to those

  • btained by Graczyk and Sobczyk.

2

Tune model parameters (MRES

A

, C A

5 (0) and

non-resonant background scale) to available pion production data on deuterium (proxy for free nucleons) and hydrogen to avoid convoluted FSI effects.

3

Compare NEUT predictions with MiniBooNE and MINERνA data without fitting.

4

Choose errors to cover discrepancies between datasets.

CV: MRES

A

= 0.95 GeV/c2, C A

5 = 1.01, nonresonant (I = 1/2) BG scale = 1.3, χ2 =2.1

  • T. Feusels (UBC)

External data fits 13/08/2015 18 / 27

slide-19
SLIDE 19

Results of tuning in Q2 and pion kinematics

Reasonable description of MiniBooNE normalization and MINERνA shape. Tension between normalizations of MiniBooNE and MINERνA.

  • T. Feusels (UBC)

External data fits 13/08/2015 19 / 27

slide-20
SLIDE 20

FSI tuning

NEUT Intra-nuclear cascade model tuned absorption, charge exchange and quasi-elastic pion-nucleus cross sections to external pion-carbon data. Currently revisiting FSI tuning, eg. adding recent DUET pion scattering data (pion charge exchange and absorption cross sections).

  • T. Feusels (UBC)

External data fits 13/08/2015 20 / 27

slide-21
SLIDE 21

Table of Contents

1

Introduction

2

NEUT Spectral Function Nieves 2p2h RPA

3

External Data Fits CCQE fits CC1pi fits FSI fits

4

Summary and Outlook

  • T. Feusels (UBC)

External data fits 13/08/2015 21 / 27

slide-22
SLIDE 22

Summary

Outlook

Use MiniBooNE covariance matrix to properly account for correlations. Add more samples: MINERνA proton sample, K2K single pion data, SciBoone,... Include FSI parameters (absorption, charge exchange and pion production cross sections) in pion data fitting. Fit CC0pi data instead of CCQE(-corrected) data. Joint CC0pi + CC1pi fits Revisit FSI external data fits.

Summary

RFG + RPA + MEC is current default T2K model for oscillation analysis. Poor agreement between MiniBooNE and MINERνA CCQE datasets. Good agreement between neutrino and antineutrino datasets. Best fit parameters still “effective”, errors inflated to cover all external datasets. Tuned pion production model provides reasonable description of MiniBooNE and MINERνA data.

  • T. Feusels (UBC)

External data fits 13/08/2015 22 / 27

slide-23
SLIDE 23

Backup

  • T. Feusels (UBC)

External data fits 13/08/2015 23 / 27

slide-24
SLIDE 24

NEUT: future plans

np/nn/pp fractions and its uncertainties for 2p2h models for proper C/O extrapolation. SF model: leptonic FSI (Ankowski et al.) Local Fermi Gas for CCQE Update Bodek-Yang correction for DIS. Martini 2p2h model and hadronic tensor. Radiative CCQE.

  • T. Feusels (UBC)

External data fits 13/08/2015 24 / 27

slide-25
SLIDE 25

NEUT model defaults: RFG + RPA + MEC

Model parameter NEUT default value MQE

A

1.21 GeV/c2 Fermi momentum pRFG

F

217 MeV/c RPA Nieves model MEC normalization 100% Nieves model Axial Form Factor Dipole Vector Form Factor BBBA05

  • T. Feusels (UBC)

External data fits 13/08/2015 25 / 27

slide-26
SLIDE 26

NEUT model defaults: SF + MEC

Model parameter NEUT default value MQE

A

1.21 GeV/c2 Fermi momentum pSF

F

209 MeV/c Mean-field width 200 MeV (Benhar SF nominal) Normalization of correlated term correlated term ∼20% of total CCQE cross-section (Benhar SF nominal) MEC normalization 100% Nieves model Axial Form Factor Dipole Vector Form Factor BBBA05

  • T. Feusels (UBC)

External data fits 13/08/2015 26 / 27

slide-27
SLIDE 27

Fit results with MB normalization

Fit type χ2/DOF MA (GeV) MEC (%) pF (MeV) λMB

ν

λMB

¯ ν

  • Rel. RPA

97.8/195 1.15±0.03 27 ±12 223±5 0.79±0.03 0.78±0.03 Non-rel. RPA 117.9/195 1.07±0.03 34 ±12 225±5 0.80±0.03 0.75±0.03 SF + MEC 97.5/195 1.33±0.03 0 (at limit) 234±4 0.81±0.03 0.86±0.03

  • T. Feusels (UBC)

External data fits 13/08/2015 27 / 27