krzysztof cichy nic, desy zeuthen, germany adam … · • the famous atiyah-singer index theorem...
TRANSCRIPT
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 1 / 35
Comparison of lattice definitions of the topological charge
Krzysztof CichyNIC, DESY Zeuthen, Germany
Adam Mickiewicz University, Poznań, Poland
in collaboration with:Arthur Dromard, Elena Garcia-Ramos, Karl Jansen, Konstantin Ottnad,
Carsten Urbach, Marc Wagner, Urs Wenger, Falk Zimmermann
Presentation outline
Presentationoutline
Introduction
Results
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 2 / 35
1. Introduction
• Motivation
• Short overview of employed definitions
2. Results
• Comparison at a single lattice spacing
• Increase of correlation towards the continuum limit
• Topological susceptibility
3. Conclusions
Motivation
Presentationoutline
Introduction
Motivation
Definitions
Index
Spectral flow
Spectral projectors
Fermionic disc.
Field theoretic
Lattice setup
Results
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 3 / 35
• theoretical
⋆ there are many definitions of the topological charge
⋆ which of them should one use for different purposes?
⋄ to compute the topological susceptibility
⋄ to sort configurations according to the topological charge
⋄ for weighting results with the topological charge
⋆ what are the pros and cons of different definitions?
⋆ which definitions are theoretically clean and which are“suspicious”?
• practical
⋆ quite a lot of unpublished data for the topological charge fromdifferent definitions and several projects
⋆ missing data rather easy to compute
Definitions
Presentationoutline
Introduction
Motivation
Definitions
Index
Spectral flow
Spectral projectors
Fermionic disc.
Field theoretic
Lattice setup
Results
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 4 / 35
• index of the overlap Dirac operator [K.C., E. García Ramos, K. Jansen]
• spectral flow of the Hermitian Wilson-Dirac operator [U. Wenger]
• spectral projectors [K.C., E. García Ramos, K. Jansen]
• fermionic from disconnected loops [K. Ottnad, C. Urbach,
F. Zimmermann]
• field theoretic with HYP smearing [U. Wenger, F. Zimmermann]
• field theoretic with APE smearing [A. Dromard, M. Wagner,
F. Zimmermann]
• field theoretic with cooling [A. Dromard, M. Wagner]
• field theoretic with gradient flow [U. Wenger]
Index of the overlap Dirac operator
Presentationoutline
Introduction
Motivation
Definitions
Index
Spectral flow
Spectral projectors
Fermionic disc.
Field theoretic
Lattice setup
Results
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 5 / 35
• Chirally symmetric fermionic discretizations allow exact zero modesof the Dirac operator.
• The famous Atiyah-Singer index theorem [M. Atiyah, I.M. Singer, Ann.
Math. 93, 139 (1971) 168] relates the topological charge to the numberof zero modes of the Dirac operator:
Q = n− − n+
• This is quite remarkable, because a property of gauge fields is linkedto a fermionic observable.
• uniqueness: dependence on the s parameter of the kernel
• pros: theoretically clean, integer-valued, no renormalization
• cons: cost, cost, cost...
Spectral flow of
the Hermitian Wilson-Dirac operator
Presentationoutline
Introduction
Motivation
Definitions
Index
Spectral flow
Spectral projectors
Fermionic disc.
Field theoretic
Lattice setup
Results
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 6 / 35
• closely related and actually equivalent to the index
• uniqueness: dependence on the s parameter of the kernel
• pros: theoretically clean, integer-valued, no renormalization, onecomputation leads to the whole s-dependence of the index
• cons: cost (although still cheaper than index), non-trivial to analyzedata
Spectral projectors
Presentationoutline
Introduction
Motivation
Definitions
Index
Spectral flow
Spectral projectors
Fermionic disc.
Field theoretic
Lattice setup
Results
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 7 / 35
• another fermionic definition, introduced in:[L. Giusti, M. Lüscher, 2008], [M. Lüscher, F. Palombi, 2010]
• PM – projector to the subspace of eigenmodes of D†D witheigenvalues below M2, evaluated stochastically
• Q = Tr {γ5PM} for chirally symmetric fermions
• spectral projectors are then also equivalent to the index (=stochasticway of counting the zero modes)
• for non-chirally symmetric fermions it still gives a clean definition,although chirality of modes is no longer ±1 −→ ±1 +O(a2)
• in practice, one evaluates the observable:
C =1
N
N∑k=1
(PMηk, γ5PMηk) ,• uniqueness: dependence on the M of PM• pros: theoretically clean, still rather cheap
• cons: stochastic ingredient, non-integer, ZS/ZP needed
Fermionic from disconnected loops
Presentationoutline
Introduction
Motivation
Definitions
Index
Spectral flow
Spectral projectors
Fermionic disc.
Field theoretic
Lattice setup
Results
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 8 / 35
• another fermionic definition, given by Chris Michael:
NfQ = mq∑
ψ̄γ5ψ =∑ mqγ5
D +mq,
in the limit as mq → 0.
• allows for a Q computation as a by-product of evaluation ofdisconnected loops
• uniqueness: yes
• pros: cheap if treated as a by-product
• cons: unclear to what extent it is valid, stochastic ingredient,non-integer, ZS/ZP needed
Field theoretic
Presentationoutline
Introduction
Motivation
Definitions
Index
Spectral flow
Spectral projectors
Fermionic disc.
Field theoretic
Lattice setup
Results
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 9 / 35
• a very natural definition
• in the continuum:
Q =1
32π2
∫d4x ǫµνρσtr[Fµν(x)Fρσ]
• on the lattice one has to choose some discretization
• renormalization:qR[U ] = round(Zqbare[U ]),
Z is the (non-zero) solution of:
min∑U
(Zqbare − round(Zqbare[U ]))2
• smoothing of gauge fields needed
Field theoretic
Presentationoutline
Introduction
Motivation
Definitions
Index
Spectral flow
Spectral projectors
Fermionic disc.
Field theoretic
Lattice setup
Results
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 10 / 35
• smoothing:
⋆ cooling – an iterative minimization of the lattice action,eliminates rough topological fluctuations while keeping largeinstantons unchanged and decreases renormalization effects bysmoothing out the UV noise[B. Berg, 1981], [Y. Iwasaki et al., 1983], [M. Teper, 1985],
[E. Ilgenfritz et al., 1986]
⋆ HYP/APE smearing [M. Albanese et al., 1987], [A. Hasenfratz,
F. Knechtli, 2001]
⋆ gradient flow (GF) [M. Lüscher, 2010]
• GF is very important from the point of view of validity of the fieldtheoretic approach
• uniqueness: discretization of F , level of smoothing
• pros: very cheap (but: cost of smoothing), theor. clean if GF usedfor smoothing and no renorm. then [M. Lüscher, P. Weisz, 2011]
• cons: HYP/APE or cooling a bit ad hoc (require renorm.)
Lattice setup
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 11 / 35
Ensemble β lattice aµl µR [MeV] κc L [fm] mπLb40.16 3.90 163 × 32 0.004 21 0.160856 1.4 2.5c30.20 4.05 203 × 40 0.003 19 0.157010 1.3 2.4d20.24 4.20 243 × 48 0.002 15 0.154073 1.3 2.4e17.32 4.35 323 × 64 0.00175 16 0.151740 1.5 2.4
Ensemble β lattice aµl µl,R [MeV] κc L [fm] mπLA30.32 1.90 323 × 64 0.0030 13 0.163272 2.8 4.0A40.32 1.90 323 × 64 0.0040 17 0.163270 2.8 4.5A50.32 1.90 323 × 64 0.0050 22 0.163267 2.8 5.1A60.24 1.90 243 × 48 0.0060 26 0.163265 2.1 4.2A80.24 1.90 243 × 48 0.0080 35 0.163260 2.1 4.8B25.32 1.95 323 × 64 0.0025 13 0.161240 2.5 3.4B35.32 1.95 323 × 64 0.0035 18 0.161240 2.5 4.0B55.32 1.95 323 × 64 0.0055 28 0.161236 2.5 5.0B75.32 1.95 323 × 64 0.0075 38 0.161232 2.5 5.8B85.24 1.95 243 × 48 0.0085 45 0.161231 1.9 4.7D20.48 2.10 483 × 96 0.0020 12 0.156357 2.9 3.9D30.48 2.10 483 × 96 0.0030 19 0.156355 2.9 4.7D45.32 2.10 323 × 64 0.0045 29 0.156315 1.9 3.9
MC histories – b40.16
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 12 / 35
-8
-6
-4
-2
0
2
4
6
8
10
12
2000 2500 3000 3500 4000
Q (
index n
onS
mear
s=
0.4
)
conf number
-8
-6
-4
-2
0
2
4
6
8
10
12
1500 2000 2500 3000 3500 4000
Q (
index H
YP
1 s
=0)
conf number
-10
-5
0
5
10
15
1500 2000 2500 3000 3500 4000
Q (
SF
HY
P1 s
=0.7
5)
conf number
-8
-6
-4
-2
0
2
4
6
8
10
12
1500 2000 2500 3000 3500 4000
Q (
SF
HY
P1 s
=0.0
)
conf number
-10
-8
-6
-4
-2
0
2
4
6
8
10
1500 2000 2500 3000 3500 4000
Q (
SF
HY
P5 s
=0.5
)
conf number
-8
-6
-4
-2
0
2
4
6
8
10
1500 2000 2500 3000 3500 4000
Q (
SF
HY
P5 s
=0.0
)
conf number
-3
-2
-1
0
1
2
3
4
5
1500 2000 2500 3000 3500 4000
Q (
spec. pro
j. n
onS
mear)
conf number
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
1500 2000 2500 3000 3500 4000
Q (
ferm
ionic
)
conf number
-8
-6
-4
-2
0
2
4
6
8
500 1000 1500 2000 2500 3000 3500 4000
Q (
GF
flo
w tim
e t0)
conf number
-8
-6
-4
-2
0
2
4
6
8
10
500 1000 1500 2000 2500 3000 3500 4000
Q (
GF
flo
w tim
e 2
t0)
conf number
-8
-6
-4
-2
0
2
4
6
8
10
500 1000 1500 2000 2500 3000 3500 4000
Q (
GF
flo
w tim
e 3
t0)
conf number
-8
-6
-4
-2
0
2
4
6
2000 2500 3000 3500 4000
Q (
field
theor.
nonS
mear)
conf number
-8
-6
-4
-2
0
2
4
6
8
10
1500 2000 2500 3000 3500 4000
Q (
field
theor.
HY
P10)
conf number
-8
-6
-4
-2
0
2
4
6
8
10
1500 2000 2500 3000 3500 4000
Q (
field
theor.
HY
P40)
conf number
-3
-2
-1
0
1
2
3
4
1500 2000 2500 3000 3500 4000
Q (
field
theor.
AP
E10)
conf number
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
2000 2500 3000 3500 4000
Q (
field
theor.
AP
E40)
conf number
-8
-6
-4
-2
0
2
4
6
8
10
1500 2000 2500 3000 3500 4000
Q (
field
theor.
coolin
g tol5
)
conf number
-8
-6
-4
-2
0
2
4
6
8
10
1500 2000 2500 3000 3500 4000
Q (
field
theor.
coolin
g tol1
)
conf number
Histograms – b40.16
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 13 / 35
0
5
10
15
20
25
30
35
40
45
-10 -5 0 5 10 15
num
ber
of confs
Q (index nonSmear s=0.4)
0
10
20
30
40
50
60
70
80
90
-10 -5 0 5 10 15
num
ber
of confs
Q (index HYP1 s=0)
0
10
20
30
40
50
60
70
80
-15 -10 -5 0 5 10 15
num
ber
of confs
Q (SF HYP1 s=0.75)
0
10
20
30
40
50
60
70
80
-10 -5 0 5 10 15
num
ber
of confs
Q (SF HYP1 s=0.0)
0
10
20
30
40
50
60
70
80
90
-10 -5 0 5 10
num
ber
of confs
Q (SF HYP5 s=0.5)
0
10
20
30
40
50
60
70
80
90
-10 -8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (SF HYP5 s=0.0)
0
20
40
60
80
100
120
-4 -3 -2 -1 0 1 2 3 4 5
num
ber
of confs
Q (spec. proj. nonSmear)
0
10
20
30
40
50
60
70
80
90
-3 -2 -1 0 1 2 3
num
ber
of confs
Q (fermionic)
0
10
20
30
40
50
60
70
80
90
-8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (GF flow time t0)
0
10
20
30
40
50
60
70
80
-8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (GF flow time 2t0)
0
10
20
30
40
50
60
70
80
90
-8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (GF flow time 3t0)
0
5
10
15
20
25
30
-8 -6 -4 -2 0 2 4 6
num
ber
of confs
Q (field theor. nonSmear)
0
5
10
15
20
25
-8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (field theor. HYP10)
0
5
10
15
20
25
30
35
40
-8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (field theor. HYP40)
0
5
10
15
20
25
30
35
-3 -2 -1 0 1 2 3 4
num
ber
of confs
Q (field theor. APE10)
0
5
10
15
20
25
30
35
40
45
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
num
ber
of confs
Q (field theor. APE40)
0
5
10
15
20
25
-8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (field theor. cooling tol5)
0
5
10
15
20
25
30
35
40
-8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (field theor. cooling tol1)
Histograms – b40.16
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 14 / 35
0
5
10
15
20
25
-8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (field theor. HYP10)
0
5
10
15
20
25
30
35
40
-8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (field theor. HYP40)
Field theoretic definition
10 HYP iterations 40 HYP iterations
Histograms – b40.16
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 15 / 35
0
5
10
15
20
25
-8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (field theor. cooling tol5)
0
5
10
15
20
25
30
35
40
-8 -6 -4 -2 0 2 4 6 8 10
num
ber
of confs
Q (field theor. cooling tol1)
Field theoretic definition with cooling
5 steps with tol. 5% 10 steps with tol. 1%
Cooling vs. HYP smearing
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 16 / 35
-4
-2
0
2
4
0 20 40 60 80 100 120 140
Q (
field
theor.
coolin
g)
number of cooling iterations
cooling conf 1870cooling conf 1930cooling conf 2090
cooling conf 2240cooling conf 2530cooling conf 2870
-4
-2
0
2
4
6
0 10 20 30 40 50
Q (
field
theor.
HY
Pn)
number of smearing iterations
HYP conf 1600HYP conf 2000
HYP conf 2400HYP conf 2800
HYP conf 3200
Field theoretic definition
cooling HYP smearing
Pairwise comparisons
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 17 / 35
-8
-6
-4
-2
0
2
4
6
8
10
12
2000 2500 3000 3500 4000
Q (
ind
ex n
on
Sm
ea
r s=
0.4
, in
de
x H
YP
1 s
=0
)
conf number
index nonSmear s=0.4
index HYP1 s=0
-10
-8
-6
-4
-2
0
2
4
6
8
10
12
-8 -6 -4 -2 0 2 4 6 8 10 12
Q (
ind
ex H
YP
1 s
=0
)
Q (index nonSmear s=0.4)
correlation=0.950385 (0.007323)index nonSmear s=0.4 + index HYP1 s=0
-8
-6
-4
-2
0
2
4
6
8
10
12
2000 2500 3000 3500 4000
Q (
ind
ex n
on
Sm
ea
r s=
0.4
, sp
ec.
pro
j. n
on
Sm
ea
r)
conf number
index nonSmear s=0.4
spec. proj. nonSmear
-8
-6
-4
-2
0
2
4
6
8
10
12
-8 -6 -4 -2 0 2 4 6 8 10 12
Q (
sp
ec.
pro
j. n
on
Sm
ea
r)
Q (index nonSmear s=0.4)
correlation=0.502528 (0.045219)index nonSmear s=0.4 + spec. proj. nonSmear
0.950(7)
index
HY
P1
index nonSmear
0.503(45)
spec
.pro
j.non
Sm
ear
index nonSmear
Pairwise comparisons
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 18 / 35
-8
-6
-4
-2
0
2
4
6
8
10
12
2000 2500 3000 3500 4000
Q (
ind
ex n
on
Sm
ea
r s=
0.4
, fe
rmio
nic
)
conf number
index nonSmear s=0.4
fermionic
-10
-8
-6
-4
-2
0
2
4
6
8
10
12
-10 -8 -6 -4 -2 0 2 4 6 8 10 12
Q (
ferm
ion
ic)
Q (index nonSmear s=0.4)
correlation=0.472707 (0.045937) index nonSmear s=0.4 + fermionic
-8
-6
-4
-2
0
2
4
6
8
10
12
2000 2500 3000 3500 4000
Q (
ind
ex n
on
Sm
ea
r s=
0.4
, G
F f
low
tim
e t
0)
conf number
index nonSmear s=0.4
GF flow time t0
-10
-8
-6
-4
-2
0
2
4
6
8
10
12
-10 -8 -6 -4 -2 0 2 4 6 8 10 12
Q (
GF
flo
w t
ime
t0
)
Q (index nonSmear s=0.4)
correlation=0.863050 (0.016853) index nonSmear s=0.4 + GF flow time t0
0.473(46)
ferm
ionic
(dis
c.)
index nonSmear
0.863(17)
GF
flow
tim
et 0
index nonSmear
Pairwise comparisons
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 19 / 35
-8
-6
-4
-2
0
2
4
6
8
10
12
2000 2500 3000 3500 4000
Q (
ind
ex n
on
Sm
ea
r s=
0.4
, fie
ld t
he
or.
no
nS
me
ar)
conf number
index nonSmear s=0.4
field theor. nonSmear
-10
-8
-6
-4
-2
0
2
4
6
8
10
12
-10 -8 -6 -4 -2 0 2 4 6 8 10 12
Q (
fie
ld t
he
or.
no
nS
me
ar)
Q (index nonSmear s=0.4)
correlation=0.183216 (0.059880)index nonSmear s=0.4 + field theor. nonSmear
-8
-6
-4
-2
0
2
4
6
8
10
12
1500 2000 2500 3000 3500 4000
Q (
ind
ex H
YP
1 s
=0
, fie
ld t
he
or.
HY
P1
0)
conf number
index HYP1 s=0
field theor. HYP10
-10
-8
-6
-4
-2
0
2
4
6
8
10
12
-10 -8 -6 -4 -2 0 2 4 6 8 10 12
Q (
fie
ld t
he
or.
HY
P1
0)
Q (index HYP1 s=0)
correlation=0.924477 (0.007851) index HYP1 s=0 + field theor. HYP10
0.183(60)
fiel
dth
eor.
non
Sm
ear
index nonSmear
0.924(8)
fiel
dth
eor.
HY
P10
index HYP1
Pairwise comparisons
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 20 / 35
-8
-6
-4
-2
0
2
4
6
8
10
12
1500 2000 2500 3000 3500 4000
Q (
ind
ex H
YP
1 s
=0
, fie
ld t
he
or.
co
olin
g t
ol5
)
conf number
index HYP1 s=0
field theor. cooling tol5
-10
-8
-6
-4
-2
0
2
4
6
8
10
12
-10 -8 -6 -4 -2 0 2 4 6 8 10 12
Q (
fie
ld t
he
or.
co
olin
g t
ol5
)
Q (index HYP1 s=0)
correlation=0.837581 (0.017606)index HYP1 s=0 + field theor. cooling tol5
-3
-2
-1
0
1
2
3
4
5
1500 2000 2500 3000 3500 4000
Q (
sp
ec.
pro
j. n
on
Sm
ea
r, f
erm
ion
ic)
conf number
spec. proj. nonSmear
fermionic
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
3
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3
Q (
ferm
ion
ic)
Q (spec. proj. nonSmear)
correlation=0.846143 (0.016841) spec. proj. nonSmear + fermionic
0.838(18)
fiel
dth
eor.
cool
.to
l5
index HYP1
0.846(17)
ferm
ionic
(dis
c.)
spec. proj. nonSmear
Pairwise comparisons
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 21 / 35
-8
-6
-4
-2
0
2
4
6
8
10
500 1000 1500 2000 2500 3000 3500 4000
Q (
GF
flo
w t
ime
t0
, G
F f
low
tim
e 3
t0)
conf number
GF flow time t0
GF flow time 3t0
-8
-6
-4
-2
0
2
4
6
8
10
-8 -6 -4 -2 0 2 4 6 8
Q (
GF
flo
w t
ime
3t0
)
Q (GF flow time t0)
correlation=0.961467 (0.004643) GF flow time t0 + GF flow time 3t0
-8
-6
-4
-2
0
2
4
6
8
10
2000 2500 3000 3500 4000
Q (
GF
flo
w t
ime
t0
, S
F H
YP
5 s
=0
.0)
conf number
GF flow time t0
SF HYP5 s=0.0
-8
-6
-4
-2
0
2
4
6
8
10
-8 -6 -4 -2 0 2 4 6 8
Q (
SF
HY
P5
s=
0.0
)
Q (GF flow time t0)
correlation=0.922854 (0.008417) GF flow time t0 + SF HYP5 s=0.0
0.961(5)
GF
flow
tim
e3t
0
GF flow time t0
0.923(8)
SF
HY
P5s
=0
GF flow time t0
Pairwise comparisons
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 22 / 35
-8
-6
-4
-2
0
2
4
6
8
2000 2500 3000 3500 4000
Q (
GF
flo
w t
ime
t0
, fie
ld t
he
or.
HY
P1
0)
conf number
GF flow time t0
field theor. HYP10
-8
-6
-4
-2
0
2
4
6
8
-8 -6 -4 -2 0 2 4 6 8
Q (
fie
ld t
he
or.
HY
P1
0)
Q (GF flow time t0)
correlation=0.970876 (0.003396) GF flow time t0 + field theor. HYP10
-8
-6
-4
-2
0
2
4
6
8
10
2000 2500 3000 3500 4000
Q (
GF
flo
w t
ime
3t0
, fie
ld t
he
or.
co
olin
g t
ol1
)
conf number
GF flow time 3t0
field theor. cooling tol1
-8
-6
-4
-2
0
2
4
6
8
10
-8 -6 -4 -2 0 2 4 6 8
Q (
fie
ld t
he
or.
co
olin
g t
ol1
)
Q (GF flow time 3t0)
correlation=0.982992 (0.003395)GF flow time 3t0 + field theor. cooling tol1
0.971(3)
fiel
dth
eor.
HY
P10
GF flow time t0
0.983(3)
fiel
dth
eor.
cool
.to
l1
GF flow time 3t0
Pairwise comparisons
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 23 / 35
-8
-6
-4
-2
0
2
4
6
8
10
1500 2000 2500 3000 3500 4000
Q (
fie
ld t
he
or.
HY
P1
0,
fie
ld t
he
or.
HY
P4
0)
conf number
field theor. HYP10
field theor. HYP40
-8
-6
-4
-2
0
2
4
6
8
10
-8 -6 -4 -2 0 2 4 6 8 10
Q (
fie
ld t
he
or.
HY
P4
0)
Q (field theor. HYP10)
correlation=0.986610 (0.001899) field theor. HYP10 + field theor. HYP40
-8
-6
-4
-2
0
2
4
6
8
10
1500 2000 2500 3000 3500 4000
Q (
fie
ld t
he
or.
HY
P4
0,
fie
ld t
he
or.
co
olin
g t
ol5
)
conf number
field theor. HYP40
field theor. cooling tol5
-8
-6
-4
-2
0
2
4
6
8
10
-8 -6 -4 -2 0 2 4 6 8 10
Q (
fie
ld t
he
or.
co
olin
g t
ol5
)
Q (field theor. HYP40)
correlation=0.940834 (0.007505)field theor. HYP40 + field theor. cooling tol5
0.987(2)
fiel
dth
eor.
HY
P40
field theor. HYP10
0.941(8)
fiel
dth
eor.
cool
.to
l5
field theor. HYP40
Pairwise comparisons
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 24 / 35
-8
-6
-4
-2
0
2
4
6
8
10
1500 2000 2500 3000 3500 4000
Q (
fie
ld t
he
or.
co
olin
g t
ol5
, fie
ld t
he
or.
co
olin
g t
ol1
)
conf number
field theor. cooling tol5
field theor. cooling tol1
-8
-6
-4
-2
0
2
4
6
8
10
-8 -6 -4 -2 0 2 4 6 8 10
Q (
fie
ld t
he
or.
co
olin
g t
ol1
)
Q (field theor. cooling tol5)
correlation=0.939104 (0.008230)field theor. cooling tol5 + field theor. cooling tol1
-10
-5
0
5
10
15
1500 2000 2500 3000 3500 4000
Q (
SF
HY
P1
s=
0.0
, S
F H
YP
1 s
=0
.75
)
conf number
SF HYP1 s=0.0
SF HYP1 s=0.75
-10
-5
0
5
10
15
-10 -8 -6 -4 -2 0 2 4 6 8 10 12
Q (
SF
HY
P1
s=
0.7
5)
Q (SF HYP1 s=0.0)
correlation=0.907585 (0.009311) SF HYP1 s=0.0 + SF HYP1 s=0.75
0.939(8)
fiel
dth
eor.
cool
.to
l1
field theor. cool. tol5
0.908(9)
SF
HY
P1s
=0.
75
SF HYP1 s = 0
Pairwise comparisons
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 25 / 35
-8
-6
-4
-2
0
2
4
6
8
10
12
1500 2000 2500 3000 3500 4000
Q (
SF
HY
P1
s=
0.0
, S
F H
YP
5 s
=0
.0)
conf number
SF HYP1 s=0.0
SF HYP5 s=0.0
-10
-8
-6
-4
-2
0
2
4
6
8
10
12
-10 -8 -6 -4 -2 0 2 4 6 8 10 12
Q (
SF
HY
P5
s=
0.0
)
Q (SF HYP1 s=0.0)
correlation=0.928337 (0.006411) SF HYP1 s=0.0 + SF HYP5 s=0.0
-10
-8
-6
-4
-2
0
2
4
6
8
10
1500 2000 2500 3000 3500 4000
Q (
SF
HY
P5
s=
0.0
, S
F H
YP
5 s
=0
.5)
conf number
SF HYP5 s=0.0
SF HYP5 s=0.5
-10
-8
-6
-4
-2
0
2
4
6
8
10
-8 -6 -4 -2 0 2 4 6 8 10
Q (
SF
HY
P5
s=
0.5
)
Q (SF HYP5 s=0.0)
correlation=0.994330 (0.000735) SF HYP5 s=0.0 + SF HYP5 s=0.5
0.928(6)
SF
HY
P5s
=0
SF HYP1 s = 0
0.994(1)
SF
HY
P5s
=0.
5
SF HYP5 s = 0
Pairwise comparisons
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 26 / 35
-8
-6
-4
-2
0
2
4
6
8
10
1500 2000 2500 3000 3500 4000
Q (
SF
HY
P5
s=
0.0
, fie
ld t
he
or.
HY
P1
0)
conf number
SF HYP5 s=0.0
field theor. HYP10
-8
-6
-4
-2
0
2
4
6
8
10
-8 -6 -4 -2 0 2 4 6 8 10
Q (
fie
ld t
he
or.
HY
P1
0)
Q (SF HYP5 s=0.0)
correlation=0.972357 (0.003023) SF HYP5 s=0.0 + field theor. HYP10
-4
-3
-2
-1
0
1
2
3
4
5
2000 2500 3000 3500 4000
Q (
sp
ec.
pro
j. n
on
Sm
ea
r, s
pe
c.
pro
j. H
YP
1)
conf number
spec. proj. nonSmear
spec. proj. HYP1
-3
-2
-1
0
1
2
3
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3
Q (
sp
ec.
pro
j. H
YP
1)
Q (spec. proj. nonSmear)
correlation=-0.050581 (0.049186) spec. proj. nonSmear + spec. proj. HYP1
0.972(3)
fiel
dth
eor.
HY
P10
SF HYP5 s = 0
-0.051(49)
spec
.pro
j.H
YP
1
spec. proj. nonSmear
Table of correlations
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 27 / 35
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1 1.000(0) 0.950(7) 0.877(14) 0.951(4) 0.929(9) 0.925(10) 0.506(44) 0.474(49) 0.863(16) 0.827(21) 0.813(20) 0.183(59) 0.907(12) 0.888(14) 0.703(33) 0.640(41) 0.830(21) 0.810(22)2 0.950(7) 1.000(0) 0.897(10) 0.990(2) 0.937(6) 0.933(6) 0.474(46) 0.462(50) 0.863(16) 0.820(21) 0.804(23) 0.167(43) 0.924(8) 0.906(9) 0.667(34) 0.585(36) 0.838(18) 0.808(20)3 0.877(14) 0.897(10) 1.000(0) 0.907(9) 0.854(12) 0.847(12) 0.404(50) 0.394(49) 0.774(26) 0.734(30) 0.718(32) 0.182(40) 0.834(15) 0.813(16) 0.652(32) 0.550(40) 0.752(23) 0.726(26)4 0.951(4) 0.990(2) 0.907(9) 1.000(0) 0.934(6) 0.928(6) 0.474(46) 0.452(52) 0.865(16) 0.821(22) 0.810(22) 0.172(44) 0.919(8) 0.900(10) 0.664(33) 0.580(40) 0.836(16) 0.806(20)5 0.929(9) 0.937(6) 0.854(12) 0.934(6) 1.000(0) 0.994(0) 0.493(41) 0.457(46) 0.915(9) 0.868(15) 0.853(17) 0.151(44) 0.967(3) 0.945(5) 0.647(34) 0.572(36) 0.879(11) 0.835(15)6 0.925(10) 0.933(6) 0.847(12) 0.928(6) 0.994(0) 1.000(0) 0.490(43) 0.455(45) 0.924(8) 0.877(15) 0.859(17) 0.154(46) 0.972(2) 0.951(5) 0.644(35) 0.575(36) 0.889(11) 0.842(15)7 0.506(44) 0.474(46) 0.404(50) 0.474(46) 0.493(41) 0.490(43) 1.000(0) 0.847(16) 0.620(39) 0.639(37) 0.647(38) 0.074(47) 0.556(42) 0.547(41) 0.464(44) 0.562(47) 0.556(41) 0.600(37)8 0.474(49) 0.462(50) 0.394(49) 0.452(52) 0.457(46) 0.455(45) 0.847(16) 1.000(0) 0.545(45) 0.558(42) 0.568(38) 0.051(42) 0.517(49) 0.502(50) 0.452(47) 0.534(41) 0.508(43) 0.553(40)9 0.863(16) 0.863(16) 0.774(26) 0.865(16) 0.915(9) 0.924(8) 0.620(39) 0.545(45) 1.000(0) 0.976(3) 0.962(4) 0.153(58) 0.971(3) 0.967(4) 0.663(39) 0.640(44) 0.974(4) 0.954(7)
10 0.827(21) 0.820(21) 0.734(30) 0.821(22) 0.868(15) 0.877(15) 0.639(37) 0.558(42) 0.976(3) 1.000(0) 0.992(1) 0.132(60) 0.934(9) 0.937(8) 0.632(43) 0.614(44) 0.965(6) 0.978(4)11 0.813(20) 0.804(23) 0.718(32) 0.810(22) 0.853(17) 0.859(17) 0.647(38) 0.568(38) 0.962(4) 0.992(1) 1.000(0) 0.138(57) 0.919(9) 0.921(9) 0.624(43) 0.615(46) 0.952(7) 0.983(3)12 0.183(59) 0.167(43) 0.182(40) 0.172(44) 0.151(44) 0.154(46) 0.074(47) 0.051(42) 0.153(58) 0.132(60) 0.138(57) 1.000(0) 0.170(44) 0.166(46) 0.256(37) 0.171(39) 0.178(45) 0.124(45)13 0.907(12) 0.924(8) 0.834(15) 0.919(8) 0.967(3) 0.972(2) 0.556(42) 0.517(49) 0.971(3) 0.934(9) 0.919(9) 0.170(44) 1.000(0) 0.987(1) 0.653(36) 0.611(33) 0.938(8) 0.904(11)14 0.888(14) 0.906(9) 0.813(16) 0.900(10) 0.945(5) 0.951(5) 0.547(41) 0.502(50) 0.967(4) 0.937(8) 0.921(9) 0.166(46) 0.987(1) 1.000(0) 0.629(35) 0.589(35) 0.941(7) 0.908(11)15 0.703(33) 0.667(34) 0.652(32) 0.664(33) 0.647(34) 0.644(35) 0.464(44) 0.452(47) 0.663(39) 0.632(43) 0.624(43) 0.256(37) 0.653(36) 0.629(35) 1.000(0) 0.776(22) 0.587(39) 0.580(40)16 0.640(41) 0.585(36) 0.550(40) 0.580(40) 0.572(36) 0.575(36) 0.562(47) 0.534(41) 0.640(44) 0.614(44) 0.615(46) 0.171(39) 0.611(33) 0.589(35) 0.776(22) 1.000(0) 0.573(35) 0.581(37)17 0.830(21) 0.838(18) 0.752(23) 0.836(16) 0.879(11) 0.889(11) 0.556(41) 0.508(43) 0.974(4) 0.965(6) 0.952(7) 0.178(45) 0.938(8) 0.941(7) 0.587(39) 0.573(35) 1.000(0) 0.939(8)18 0.810(22) 0.808(20) 0.726(26) 0.806(20) 0.835(15) 0.842(15) 0.600(37) 0.553(40) 0.954(7) 0.978(4) 0.983(3) 0.124(45) 0.904(11) 0.908(11) 0.580(40) 0.581(37) 0.939(8) 1.000(0)
1 = index nonSmear s = 0.4, 2 = index HYP1 s = 0, 3 = SF HYP1 s = 0.75,4 = SF HYP1 s = 0, 5 = SF HYP5 s = 0.5, 6 = SF HYP5 s = 0,7 = spectral projectors nonSmear, 8 = fermionic (from disconnected loops),9 = GF flow time t0, 10 = GF flow time 2t0, 11 = GF flow time 3t0,12 = field theor. nonSmear, 13 = field. theor. HYP10, 14 = field. theor. HYP40,15 = field. theor. APE10, 16 = field. theor. APE40, 17 = field theor. cooling tol. 5%,
18 = field theor. cooling tol. 1%
Plot of correlations
Presentationoutline
Introduction
Results
MC histories
Histograms
Cool. vs. HYP
Pair comparisons
Correlations
Topo. susc.
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 28 / 35
index nonSmear s=0.4
index nonSmear s=0
index HYP1 s=0
SF HYP1 s=0.75
SF HYP1 s=0.0
SF HYP5 s=0.5
SF HYP5 s=0.0
spec. proj. nonSmear
spec. proj. HYP1
fermionic
GF flow time t0
GF flow time 2t0
GF flow time 3t0
field theor. nonSmear
field theor. HYP10
field theor. HYP40
field theor. APE10
field theor. APE40
field theor. cooling tol5
field theor. cooling tol1
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
me
tho
d 2
method 1
-0.2
0
0.2
0.4
0.6
0.8
1
Correlation towards the continuum limit
Presentationoutline
Introduction
Results
MC histories
Histograms
Cool. vs. HYP
Pair comparisons
Correlations
Topo. susc.
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 29 / 35
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
corr
ela
tion
a/r0
index HYP1 s=0 vs. spec. proj. nonSmearindex HYP1 s=0 vs. GF flow time t0
index HYP1 s=0 vs. field theor. cooling tol1index HYP1 s=0 vs. field theor. cooling tol5
spec. proj. nonSmear vs. GF flow time t0spec. proj. nonSmear vs. field theor. cooling tol1spec. proj. nonSmear vs. field theor. cooling tol5
GF flow time t0 vs. field theor. cooling tol1GF flow time t0 vs. field theor. cooling tol5
field theor. cooling tol1 vs. field theor. cooling tol5
Topological susceptibility – b40.16
Presentationoutline
Introduction
Results
MC histories
Histograms
Cool. vs. HYP
Pair comparisons
Correlations
Topo. susc.
Conclusions
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 30 / 35
0
1e-05
2e-05
3e-05
4e-05
5e-05
6e-05
7e-05
8e-05
9e-05
0.0001
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
a4 χ
ren
definition ID
noSmears=0.4
noSmears=0
HYP1s=0
HYP1s=0
HYP1s=0.75
HYP5s=0
HYP5s=0.5
noSmear
HYP1
noSmear
flow time: t0 2t0 3t0
HYP1
HYP2
HYP5
HYP10
HYP20
HYP30
HYP40
HYP50
tol.5
tol.1
indexspectral flow
spec. projectorsfermionic (disc.)
field theor. noSmearfield theor. GF
field theor. HYPfield theor. cool.
Topological susceptibility
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 31 / 35
used only pion masses mπ ≤ 400 MeV
spectral projectors: r0Σ1/3 = 0.646(59) fermionic: r0Σ1/3 = 0.644(20)
Tree-level χPT fit: r40χ =r3
0Σ·r0µR
2
compare to direct determination[K.C., E. García Ramos, K. Jansen, JHEP 10(2013)175]
r0Σ1/3cont,Nf=2+1+1 = 0.680(29)
Topological susceptibility
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 32 / 35
used all pion masses (mπ = [260, 500] MeV)
spectral projectors: r0Σ1/3 = 0.612(57) fermionic: r0Σ1/3 = 0.622(19)
Tree-level χPT fit: r40χ =r3
0Σ·r0µR
2
compare to direct determination[K.C., E. García Ramos, K. Jansen, JHEP 10(2013)175]
r0Σ1/3cont,Nf=2+1+1 = 0.680(29)
My subjective ranking of costs
Presentationoutline
Introduction
Results
Conclusions
Costs
Cleanness
Which definition?
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 33 / 35
From the most expensive to the cheapest:
1. index of the overlap Dirac operator
2. spectral flow
3. spectral projectors
4. fermionic from disconnected loops (but...)
5. field theoretic (with smearing or cooling)
6. field theoretic with gradient flow
My subjective ranking
of theoretical cleanness
Presentationoutline
Introduction
Results
Conclusions
Costs
Cleanness
Which definition?
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 34 / 35
From the cleanest to the problematic ones:
1. index of the overlap Dirac operatorspectral flowfield theoretic with gradient flow
2. spectral projectors (fully clean for χ)
3. field theoretic with smearing
4. field theoretic with cooling
5. fermionic from disconnected loops
Which definition to use?
Presentationoutline
Introduction
Results
Conclusions
Costs
Cleanness
Which definition?
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 35 / 35
The answer to this question is not unique:
• for the computation of topological susceptibility:spectral projectors, field theoretic with gradient flow
• for sorting configurations:field theoretic
• for weighting results with the topological charge:field theoretic
And the winner is: ,
field theoretic with gradient flow
Thank you for your attention!
Presentationoutline
Introduction
Results
Conclusions
Thank you for yourattention!
Krzysztof Cichy ETMC Meeting Grenoble – 26-28 May 2014 – 36 / 35