Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Ab Initio Theory
Outside the Box
Robert Roth
1
Inside the Box
Robert Roth – TU Darmstadt – 02/2014
this workshop has provided an impres-sive snapshot of the progress andperspectives in ab initio nucleartheory and its links to experiment
definition: everything we have heardso-far is inside the ab initio box
2
Inside the Box
Robert Roth – TU Darmstadt – 02/2014
ab initio theory is entering new territory...
QCD frontiernuclear structure connected systematically to QCD via chiral EFT
accuracy frontiercontrol uncertainties, improve convergence, inform extrapolations
mass frontierab initio calculations up to heavy nuclei with quantified uncertainties
open-shell frontierextend to medium-mass open-shell nuclei and their excitation spectrum
continuum & clustering frontierinclude continuum & clustering effects for threshold states & nuclei
reaction frontierdescribe structure & reaction observables on the same footing
...providing a coherent theoretical framework for nuclearstructure & reactions and linking it to experiment
3
Outside the Box
Robert Roth – TU Darmstadt – 02/2014
two more things that are not yet insidethe ab initio box:
ab initio hypernuclear structurecan we describe the spectroscopy ofp-shell hypernuclei ab initio ?
perturbation theory — ab initio ?wouldn’t it be great if MBPT wouldqualify as ab initio approach ?
4
Ab Initio Hypernuclear Structure
with
Roland Wirth, Daniel Gazda, Petr Navrátil
5
Ab Initio Hypernuclear Structure
Robert Roth – TU Darmstadt – 02/2014
precise data on ground states &spectroscopy of hypernuclei
ab initio few-body (A ≲ 4) andphenomenological shell modelor cluster calculations
chiral YN & YY interactions at(N)LO are available
constrain YN & YY interactionsby ab initio hypernuclear struc-ture calculations
6
YN Interaction — A Problem
Robert Roth – TU Darmstadt – 02/2014
Haidenbauer et al., NPA 915, 24 (2013), Polinder et al., NPA 779, 244 (2006), Haidenbauer et al., PRC 72, 044005 (2005)
100 200 300 400 500 600 700 800 900
plab (MeV/c)
0
100
200
300
σ (m
b)
Sechi-Zorn et al.Kadyk et al.Alexander et al.
Λp -> Λp
100 120 140 160 180
plab (MeV/c)
0
50
100
150
200
250
300
σ (m
b)
Engelmann et al.
Σ−p -> Λn
100 120 140 160 180
plab (MeV/c)
0
100
200
300
400
500
σ (m
b)
Engelmann et al.
Σ−p -> Σ0
n
100 120 140 160 180
plab (MeV/c)
0
50
100
150
200
250
300
σ (m
b)
Eisele et al.
Σ−p -> Σ−
p
Jülich’04 LO chiral NLO chiral
experimental YN scattering data is scarce and has large uncertainties
fit of interactions not well constrained (invoke symmetries)
scattering data cannot discriminate between different YN potentials
7
Ab Initio Toolbox
Robert Roth – TU Darmstadt – 02/2014
Hamiltonian from chiral EFT
NN: chiral N3LO by Entem & Machleidt, ΛNN = 500 MeV
3N: chiral N2LO by Navrátil, Λ3N = 500 MeV, A = 3 fit
YN: chiral LO by Polinder, Haidenbauer & Meißner, ΛYN = 600,700 MeVJülich’04 by Haidenbauer & Meißner
Similarity Renormalization Group
consistent SRG-evolution of NN, 3N, YN interactions
using particle basis and including Λ--coupling (larger matrices)
Λ- mass difference and p± Coulomb included consistently
Importance Truncated No-Core Shell Model
include explicit (p, n,Λ,+,0,−) with physical masses
larger model spaces easily tractable with importance truncation
all p-shell single-Λ hypernuclei are accessible
8
Application: Λ7Li
Robert Roth – TU Darmstadt – 02/2014
1+
3+
-40
-35
-30
-25
-20
.
E[M
eV]
1+
3+
4 6 8 10 12 Exp.Nmx
-1
0
1
2
3
.
E[M
eV]
6Li NN @ N3LOΛNN = 500 MeV
Entem&Machleidt
3N @ N2LOΛ3N = 500 MeV
NavratilA = 3 fit
Jülich’04Haidenbauer et al.scatt. & hypertriton
αN = 0.08 fm4
αY = 0.00 fm4
hΩ = 20 MeV
9
Application: Λ7Li
Robert Roth – TU Darmstadt – 02/2014
1+
3+
-40
-35
-30
-25
-20
.
E[M
eV]
1+
3+
4 6 8 10 12 Exp.Nmx
-1
0
1
2
3
.
E[M
eV]
12+32+
52+72+
12+
32+
52+
72+
4 6 8 10 12 Exp.Nmx
6Li 7ΛLi NN @ N3LO
ΛNN = 500 MeV
Entem&Machleidt
3N @ N2LOΛ3N = 500 MeV
NavratilA = 3 fit
Jülich’04Haidenbauer et al.scatt. & hypertriton
αN = 0.08 fm4
αY = 0.00 fm4
hΩ = 20 MeV
10
Application: Λ7Li
Robert Roth – TU Darmstadt – 02/2014
1+
3+
-40
-35
-30
-25
-20
.
E[M
eV]
1+
3+
4 6 8 10 12 Exp.Nmx
-1
0
1
2
3
.
E[M
eV]
12+32+
52+72+
12+
32+
52+
72+
4 6 8 10 12 Exp.Nmx
6Li 7ΛLi NN @ N3LO
ΛNN = 500 MeV
Entem&Machleidt
3N @ N2LOΛ3N = 500 MeV
NavratilA = 3 fit
Jülich’04Haidenbauer et al.scatt. & hypertriton
αN = 0.08 fm4
αY = 0.00 fm4
hΩ = 20 MeV
11
Application: Λ7Li
Robert Roth – TU Darmstadt – 02/2014
1+
3+
-40
-35
-30
-25
-20
.
E[M
eV]
1+
3+
4 6 8 10 12 Exp.Nmx
-1
0
1
2
3
.
E[M
eV]
12+32+
52+72+
12+
32+
52+
72+
4 6 8 10 12 Exp.Nmx
6Li 7ΛLi NN @ N3LO
ΛNN = 500 MeV
Entem&Machleidt
3N @ N2LOΛ3N = 500 MeV
NavratilA = 3 fit
YN @ LOΛYN = 600 MeV
Polinder et al.scatt. & hypertriton
αN = 0.08 fm4
αY = 0.00 fm4
hΩ = 20 MeV
12
Application: Λ7Li
Robert Roth – TU Darmstadt – 02/2014
1+
3+
-40
-35
-30
-25
-20
.
E[M
eV]
1+
3+
4 6 8 10 12 Exp.Nmx
-1
0
1
2
3
.
E[M
eV]
12+32+
52+72+
12+
32+
52+
72+
4 6 8 10 12 Exp.Nmx
6Li 7ΛLi NN @ N3LO
ΛNN = 500 MeV
Entem&Machleidt
3N @ N2LOΛ3N = 500 MeV
NavratilA = 3 fit
YN @ LOΛYN = 700 MeV
Polinder et al.scatt. & hypertriton
αN = 0.08 fm4
αY = 0.00 fm4
hΩ = 20 MeV
13
Application: Λ9Be
Robert Roth – TU Darmstadt – 02/2014
0+
2+
-65
-60
-55
-50
-45
-40
.
E[M
eV]
0+
2+
2 4 6 8 10 Exp.Nmx
0
1
2
3
4
.
E[M
eV]
12+
32+52+
12+
32+52+
2 4 6 8 10 Exp.Nmx
8Be 9ΛBe NN @ N3LO
ΛNN = 500 MeV
Entem&Machleidt
3N @ N2LOΛ3N = 500 MeV
NavratilA = 3 fit
Jülich’04Haidenbauer et al.scatt. & hypertriton
αN = 0.08 fm4
αY = 0.00 fm4
hΩ = 20 MeV
14
Application: Λ9Be
Robert Roth – TU Darmstadt – 02/2014
0+
2+
-65
-60
-55
-50
-45
-40
.
E[M
eV]
0+
2+
2 4 6 8 10 Exp.Nmx
0
1
2
3
4
.
E[M
eV]
12+
32+52+
12+
32+52+
2 4 6 8 10 Exp.Nmx
8Be 9ΛBe NN @ N3LO
ΛNN = 500 MeV
Entem&Machleidt
3N @ N2LOΛ3N = 500 MeV
NavratilA = 3 fit
YN @ LOΛYN = 600 MeV
Polinder et al.scatt. & hypertriton
αN = 0.08 fm4
αY = 0.00 fm4
hΩ = 20 MeV
15
Application: Λ9Be
Robert Roth – TU Darmstadt – 02/2014
0+
2+
-65
-60
-55
-50
-45
-40
.
E[M
eV]
0+
2+
2 4 6 8 10 Exp.Nmx
0
1
2
3
4
.
E[M
eV]
12+
32+52+
12+
32+52+
2 4 6 8 10 Exp.Nmx
8Be 9ΛBe NN @ N3LO
ΛNN = 500 MeV
Entem&Machleidt
3N @ N2LOΛ3N = 500 MeV
NavratilA = 3 fit
YN @ LOΛYN = 700 MeV
Polinder et al.scatt. & hypertriton
αN = 0.08 fm4
αY = 0.00 fm4
hΩ = 20 MeV
16
Application: Λ13C
Robert Roth – TU Darmstadt – 02/2014
0+
2+
-110
-105
-100
-95
-90
-85
-80
-75
.
E[M
eV]
0+
2+
2 4 6 8 10 Exp.Nmx
0
2
4
6
.
E[M
eV]
12+
32+
12+
32+
2 4 6 8 10 Exp.Nmx
12C 13ΛC NN @ N3LO
ΛNN = 500 MeV
Entem&Machleidt
3N @ N2LOΛ3N = 500 MeV
NavratilA = 3 fit
Jülich’04Haidenbauer et al.scatt. & hypertriton
αN = 0.08 fm4
αY = 0.00 fm4
hΩ = 20 MeV
17
Application: Λ13C
Robert Roth – TU Darmstadt – 02/2014
0+
2+
-110
-105
-100
-95
-90
-85
-80
-75
.
E[M
eV]
0+
2+
2 4 6 8 10 Exp.Nmx
0
2
4
6
.
E[M
eV]
12+
32+
12+
32+
2 4 6 8 10 Exp.Nmx
12C 13ΛC NN @ N3LO
ΛNN = 500 MeV
Entem&Machleidt
3N @ N2LOΛ3N = 500 MeV
NavratilA = 3 fit
YN @ LOΛYN = 600 MeV
Polinder et al.scatt. & hypertriton
αN = 0.08 fm4
αY = 0.00 fm4
hΩ = 20 MeV
18
Application: Λ13C
Robert Roth – TU Darmstadt – 02/2014
0+
2+
-110
-105
-100
-95
-90
-85
-80
-75
.
E[M
eV]
0+
2+
2 4 6 8 10 Exp.Nmx
0
2
4
6
.
E[M
eV]
12+
32+
12+
32+
2 4 6 8 10 Exp.Nmx
12C 13ΛC NN @ N3LO
ΛNN = 500 MeV
Entem&Machleidt
3N @ N2LOΛ3N = 500 MeV
NavratilA = 3 fit
YN @ LOΛYN = 700 MeV
Polinder et al.scatt. & hypertriton
αN = 0.08 fm4
αY = 0.00 fm4
hΩ = 20 MeV
19
SRG Evolution of YN Channels
Robert Roth – TU Darmstadt – 02/2014
6 8 10 12 14 16 18Nmx
-44
-42
-40
-38
-36
-34
.
E[M
eV]
αY [fm4]
0.000
0.005
0.010
0.0200.0400.080
7ΛLi
SRG evolution of YN chan-nels improves conver-gence as expected
significant αY dependenceindicates SRG-inducedYNN interactions
YN @ LOΛYN = 600 MeV
αN = 0.08 fm4
hΩ = 20 MeV
20
SRG Evolution of YN Channels
Robert Roth – TU Darmstadt – 02/2014
0 0.04 0.08 0.12 0.16
αY [fm4]
-44
-43
-42
-41
-40
-39
.
E[M
eV]
7ΛLi
SRG evolution of YN chan-nels improves conver-gence as expected
significant αY dependenceindicates SRG-inducedYNN interactions
YN @ LOΛYN = 600 MeV
αN = 0.08 fm4
hΩ = 20 MeV
3.4 MeVinduced YNN
8.0 MeVinitial YN
21
Ab Initio Hypernuclear Structure
Robert Roth – TU Darmstadt – 02/2014
ab initio hypernuclear structure in the IT-NCSM now possible for allsingle-Λ p-shell hypernuclei
LO chiral YN interactions provide spectra that agree with experimentwithin cutoff uncertainties
hypernuclear structure sets tight constraints on YN interaction
significant SRG-induced YNN interactions, implications for mean-fieldtype models and the hyperon puzzle ?
NLO chiral YN interactions are expected to reduce cutoff dependence,but fit is difficult...(13 instead of 5 LECs in S/SD-waves assuming SU(3)ƒ and neglecting P-waves; fit to 36 data)
lots of applications are waiting...
22
Perturbation Theory — Ab Initio ?
with
Alexander Tichai, Christina Stumpf, Joachim Langhammer
23
Many-Body Perturbation Theory
Robert Roth – TU Darmstadt – 02/2014
low-order many-body perturbation theory is a cheap and sim-ple tool to access nuclear observables
wouldn’t it be great if low-order MBPT would qualify as ab initioapproach ?
problem: convergence behavior of perturbation series unclear
how to quantify uncertainties?
which factors influence the order-by-order convergence?
how to restore or accelerate the convergence?
strategy: study convergence behavior with explicit high-ordercalculations
24
Explicit High-Order MBPT
Robert Roth – TU Darmstadt – 02/2014
partitioning: definition of unperturbed basis ∣n⟩H(λ) = H0 + λW H0 ∣n⟩ = εn ∣n⟩
power-series ansatz for energy and eigenstates
En(λ) = ∞∑p=0
λp E(p)n ∣Ψn(λ)⟩ = ∞∑
p=0
λp ∣Ψ(p)n ⟩
recursive relations for energy E(p)n and states ∣Ψ(p)n ⟩ = ∑mC
(p)n,m ∣m⟩
E(p)n = ∑
m
⟨n∣W ∣m⟩ C(p−1)n,m
C(p)n,m = 1
εn − εm(∑m′⟨m∣W ∣m′⟩ C(p−1)n,m′ −p∑j=1
E(j)n C
(p−j)n,m )
easy to evaluate to ’arbitrary’ order with NCSM technology...
25
Summation and Resummation
Robert Roth – TU Darmstadt – 02/2014
partial sum: starting point for convergence study
Esum(p) = E(0) + λE(1) + λ2E(2) +⋯λpE(p)∣λ=1 Padé approximant: map power series of order p to a quotient ofpolynomials of orders M and N
EPadé(M/N) = A(0) + λA(1) + λ2A(2) +⋯λMA(M)B(0) + λB(1) + λ2B(2) +⋯λNB(N) ∣λ=1= Esum(M +N) +O(M +N + 1)
focus on Padé main sequence: EPadé(M/M) and EPadé(M/M − 1)
powerful convergence theory for special power series (e.g. Stieltjes)...
additional sequence transformations on top of Padé can further accel-erate convergence (Shanks, Levin-Weniger)...
26
16O: MBPT Convergence
Robert Roth – TU Darmstadt – 02/2014
HO – Partial Sum
0 5 10 15 20 25p
-200
-180
-160
-140
-120
-100
.
E(p)[M
eV]
partial sum exhibitsexponential oscillatory
divergence
emx = 2 emx = 4 emx = 6 ∎ emx = 8
NNonly , α = 0.08 fm4 , hΩ = 20 MeV , 2p2h truncated
27
16O: MBPT Convergence
Robert Roth – TU Darmstadt – 02/2014
HO – Partial Sum
0 5 10 15 20 25p
-200
-180
-160
-140
-120
-100
.
E(p)[M
eV]
HO – Padé Approx.
0 5 10 15 20 25 30p
Padé resummationleads to rapidly
converging series
converged resultsare identical to exact
eigenvalues
emx = 2 emx = 4 emx = 6 ∎ emx = 8
NNonly , α = 0.08 fm4 , hΩ = 20 MeV , 2p2h truncated
28
16O: MBPT Convergence
Robert Roth – TU Darmstadt – 02/2014
HO – Partial Sum
0 5 10 15 20 25p
-200
-180
-160
-140
-120
-100
.
E(p)[M
eV]
HF – Partial Sum
0 5 10 15 20 25 30p
HF as unperturbedbasis also yields rapid
convergence
wrong long-rangebehavior of HO drives
divergence ?
emx = 2 emx = 4 emx = 6 ∎ emx = 8
NNonly , α = 0.08 fm4 , hΩ = 20 MeV , 2p2h truncated
29
16O: MBPT Convergence
Robert Roth – TU Darmstadt – 02/2014
HO – Padé Approx.
0 5 10 15 20 25p
-200
-180
-160
-140
-120
-100
.
E(p)[M
eV]
HF – Padé Approx.
0 5 10 15 20 25 30p
Padé always yieldsconvergence
emx = 2 emx = 4 emx = 6 ∎ emx = 8
NNonly , α = 0.08 fm4 , hΩ = 20 MeV , 2p2h truncated
30
16O: MBPT Convergence
Robert Roth – TU Darmstadt – 02/2014
HO – Partial Sum
0 5 10 15 20 25p
-200
-180
-160
-140
-120
-100
.
E(p)[M
eV]
HF – Partial Sum
0 5 10 15 20 25 30p
convergencein HF basis depends onsoftness of interaction
α = 0.02 fm4
α = 0.04 fm4
α = 0.08 fm4
NNonly , emx = 8 , hΩ = 20 MeV , 2p2h truncated
31
16O: MBPT Convergence
Robert Roth – TU Darmstadt – 02/2014
HO – pth Order
0 5 10 15 20 25p
10−3
10−2
0.1
1
10
102
103
104
105
.
∣E(p)∣[M
eV]
HF – pth Order
0 5 10 15 20 25 30p
convergence &divergence is always
exponential
α = 0.02 fm4
α = 0.04 fm4
α = 0.08 fm4
NNonly , emx = 8 , hΩ = 20 MeV , 2p2h truncated
32
16O: MBPT Convergence
Robert Roth – TU Darmstadt – 02/2014
HO – Padé Approx.
0 5 10 15 20 25p
-200
-180
-160
-140
-120
-100
.
E(p)[M
eV]
HF – Padé Approx.
0 5 10 15 20 25 30p
α = 0.02 fm4
α = 0.04 fm4
α = 0.08 fm4
NNonly , emx = 8 , hΩ = 20 MeV , 2p2h truncated
33
MBPT Convergence Heuristics
Robert Roth – TU Darmstadt – 02/2014
in many cases partial sums do not converge, but there are sys-tematic exceptions
factors causing the non-convergence
unperturbed basis (partitioning) is the primary factor
softness of the interaction is a secondary factor
many-body truncation also has some influence
Padé resummation robustly provides convergence at intermedi-ate orders and agrees with exact diagonalization
can we rely on low-order MBPT ?
34
Low-Order MBPT
Robert Roth – TU Darmstadt – 02/2014
-25
-20
-15
-10
-5
.
E/A[M
eV]
4He16O
24O34Si
40Ca48Ca
48Ni56Ni
78Ni88Sr
90Zr100Sn
114Sn132Sn
146Gd208Pb
-6
-5
-4
-3
-2
-1
.
Ecorr/A[M
eV]
NNonly
α = 0.08 fm4
HF basisemx = 10hΩ = 20 MeV
Exp.
EHF
∎
EHF + E(2)
EHF + E(2)+ E(3)
35
Low-Order MBPT vs. Coupled-Cluster
Robert Roth – TU Darmstadt – 02/2014
-25
-20
-15
-10
-5
.
E/A[M
eV]
4He16O
24O34Si
40Ca48Ca
48Ni56Ni
78Ni88Sr
90Zr100Sn
114Sn132Sn
146Gd208Pb
-6
-5
-4
-3
-2
-1
.
Ecorr/A[M
eV]
NNonly
α = 0.08 fm4
HF basisemx = 10hΩ = 20 MeV
Exp.
EHF
∎
EHF + E(2)
EHF + E(2)+ E(3)
CR-CC(2,3)
36
Low-Order MBPT vs. Coupled-Cluster
Robert Roth – TU Darmstadt – 02/2014
-16
-14
-12
-10
-8
-6
-4
-2
.
E/A[M
eV]
4He16O
24O34Si
40Ca48Ca
48Ni56Ni
78Ni88Sr
90Zr100Sn
114Sn132Sn
146Gd208Pb
-6
-5
-4
-3
-2
-1
.
Ecorr/A[M
eV]
NNonly
α = 0.02 fm4
HF basisemx = 10hΩ = 20 MeV
Exp.
EHF
∎
EHF + E(2)
EHF + E(2)+ E(3)
CR-CC(2,3)
low-order MBPTseems to work nicely,
but...
37
Perspectives: Degenerate MBPT
Robert Roth – TU Darmstadt – 02/2014
PRC 86, 054315 (2012), PLB 683, 272 (2010)
-40
-30
-20
-10
0
.
E[M
eV]
d = 0 d = 1 d = 2 d = 3 d = 4
0 4 8 12 16 20 24 28pth order
-40
-30
-20
-10
E[M
eV]
d = 5
0 4 8 12 16 20 24 28pth order
6LiNmax = 8~Ω = 20 MeV
d = 6
0 4 8 12 16 20 24 28pth order
d = 7
0 4 8 12 16 20 24 28pth order
d = 8
0 4 8 12 16 20 24 28pth order
d = 9
-35
-30
-25
-20
-15
.
E∗
[MeV
]
d = 0 d = 1 d = 2 d = 3 d = 4
0 4 8 12 16 20 24 28L + M
-35
-30
-25
-20
.
E∗
[MeV
]
d = 5
0 4 8 12 16 20 24 28
6LiNmax = 8~Ω = 20 MeV
L + M
d = 6
0 4 8 12 16 20 24 28L + M
d = 7
0 4 8 12 16 20 24 28L + M
d = 8
0 4 8 12 16 20 24 28L + M
d = 9
38
Epilogue
Robert Roth – TU Darmstadt – 02/2014
thanks to my group & my collaborators
S. Binder, J. Braun, A. Calci, S. Fischer,E. Gebrerufael, H. Spiess, J. Langhammer, S. Schulz,C. Stumpf, A. Tichai, R. Trippel, R. Wirth, K. VobigInstitut für Kernphysik, TU Darmstadt
P. NavrátilTRIUMF Vancouver, Canada
J. Vary, P. MarisIowa State University, USA
S. Quaglioni, G. HupinLLNL Livermore, USA
P. PiecuchMichigan State University, USA
H. HergertOhio State University, USA
P. PapakonstantinouIBS/RISP, Korea
C. ForssénChalmers University, Sweden
H. Feldmeier, T. NeffGSI Helmholtzzentrum
JUROPA LOEWE-CSC HOPPER
COMPUTING TIME
39