Klabucar Zimany 2019

UA (1) symmetry breaking is why η0 ≈ η ′ has an anomalous piece of mass Examples of UA (1) and Chiral symmetry interplay: Leutwyler-Smilga
ga relation a
How, how much, and when are UA (1) and chiral

symmetry restored: T -dependence of axions, η and η ′
Talk presented at
ZIMÁNYI SCHOOL’19, Winter Workshop on Heavy Ion Physics
Budapest, Hungary, 2. – 6. of December 2019.
Dubravko Klabučar(1) in collaboration with Davor Horvatić(1) & Dalibor Kekez(2)
(1) Physics Department, Faculty of Science – PMF, University of Zagreb, Croatia

(2) Rudjer Bošković Institute, Zagreb, Croatia
UA (1) symmetry breaking is why η0 ≈ η ′ has an anomalous piece of mass Examples of UA (1) and Chiral symmetry interplay: Leutwyler-Smilga relation a
The issue of the (effective) restoration of the UA (1) symmetry

● In QCD, UA (1) and SUA (3) chiral symmetry are explicitly broken by current
quark masses: only slightly by mu and md & not too badly by s-quark mass ms
Ð→ chiral limit(s) make sense [with 3 (or 2) mq → 0].
● But approximate chiral SUA (3) symmetry = absent due to DChSB,
signaled by ⟨q̄q⟩ condensates and by the octet of very light (almost)
Goldstone bosons: π 0,± , K 0,± , K¯0 , η.
... But as lattice now agrees, chiral symmetry should be restored as a
crossover (for µ ∼ 0) around TCh ∼ 155 MeV: ⟨q̄q⟩(T ) → 0.
● η ′ very massive, as even in chiral limit, mq → 0, UA (1) is broken
explicitly on the quantum level by nonabelian (”gluon”) axial anomaly:
λ0 ̃ a (x) ≡ F a (x) 1 µνρσ F a (x) ≠ 0 ,
∂α ψ̄(x)γ α γ5 ψ(x) ∝ F a (x) ⋅ F µν 2 ρσ
2
which holds at any E and T ⇒ ? Does UA (1) remains unrestored ?!?!
NO, since DChSB and UA (1) anomaly are tied through quark
bilinears such as ⟨q̄q⟩ and QCD topological susceptibility χ ⇒
Expect an effective restoration signaled by vanishing or diminishing
of UA (1)-violating quantities (e.g., large Mη′ , difference π-a0 (980), ...)
over the chiral symmetry crossover ... BUT ...
... still debatable what happens with UA (1) symmetry restoration!

● Presently, no consensus within lattice community whether UA (1) is badly
broken or effectively restored at the chiral crossover critical temper. T = TCh
[Sharma for HotQCD collaboration, e-Print: arXiv:1801.08500]
● Already older works found sizable UA (1) breaking above TCh [Bernard+al, PRL78
(1997)598, Chanrasekharan+al, PRL82(1999)2463, Ohno+al, PoS LATTICE 2011(2011)210 arXiv:1111.1939]
... and, this is confirmed by some recent works: notably by HotQCD collab.
[Bazavov+al,PRD86(2012)9094503] and by Karsch & collaborators [Buchoff+al,PRD89(2014)
054514, Sharma+al, NPA956(2016)793, Dick+al,PRD91(2015)095504] as high as T = 1.5 TCh .
● BUT, some recent works claim that UA (1) breaking above TCh is
overestimated in the continuum limit (blaming lattice artifacts near ChLim).
Some then conclude that UA (1) anomaly is consistent with zero above TCh ,
including also Graz group Rohrhofer+al, Phys.Rev.D96(2017)094501 arXiv 1707.01881, but
most vocal were researchers around JLQCD collaboration [Cossu+al,
PRD93(2016)034507 arXiv:1510.07395, PRD87&88 (2013)114514&019901 ....
These disappearances of UA (1) anomaly seem to be associated with the chiral
limit - see, e.g., Tomiya+al, PRD96(2017)034509.
● Then our model approach to η-η ′ may show that these two kinds of results
can be reconciled, since it is consistent with both - depending whether one uses
“massless” ⟨q̄q⟩0 or “massive” qq̄ condensates: Horvatić, Kekez & D.K.,
Phys.Rev. D99 (2019) 014007, and spinoff for axions in Universe 5 (2019) 208.
What happens with UA (1) symmetry restoration matters a lot - see Columbia plot!
Left: UA (1) broken by anomaly, right: UA (1) restored (C.Fischer arXiv1810.12938)
General renorm-group arguments (Pisarski:1983ms) ⇒ QCD with 3 degenerate

light flavors has a 1st order phase transition in chiral limit, whereas in QCD with
(2+1) flavors (i.e., s-quark significantly more massive), a 2nd order chiral-limit
transition is also possible and even more likely (e.g., Ejiri:2009ac,Ding:2019fzc).
A 2nd order chiral-limit transition is exhibited by most DSE models – e.g.,
clearly through the characteristic drop of their “massless” condensates ⟨q̄q⟩0 .
Quantum-level breaking of UA (1) causes anomalously high η ′ ≈ η0 mass

QCD chiral behavior (reproduced by, e.g., DS approach) of the non-anomalous
′ = const (mq + mq ′ ).
2
parts of masses of light qq̄ ′ pseudoscalars: Mqq̄
⇒ non-anomalous parts of the masses cancel in Witten-Veneziano rel. (WVR):
Mη′ 2 + Mη 2 − 2 MK 2 = 2Nf
fπ2
χYM = anomalous mass2 ≡ MUA (1) 2 ≈ ∆Mη0 2 ,
g2 a ̃ a (x)
χ ≡ ∫ d 4 x ⟨0∣Q(x)Q(0)∣0⟩ = ma2 fa2 , Q(x) = F (x)F
32 π 2 µν µν
QCD topological susceptibility χ = a direct measure of UA (1) breaking ⇒

(partial) UA (1) restoration is indicated by vanishing or reduction of χ and
related quantities, like MUA (1) ≈ ∆Mη0 ≈ ∆Mη′ .
● Q(x) = topological charge density operator
● In WVR, χ is pure-gauge, YM one, χYM ↔ χquench , obtained long ago by
lattice - harder for χ of light-flavorQCD, but can use DiVecchia-Veneziano
− ⟨q̄q⟩0
relation ∶ χ = + C m (unknown corrections, higher O in small mq )
∑ m1q
q=u,d,s
=
. 1 st
& simplest example of UA (1) breaking given by chiral symmetry breaking
The 2nd example of tied breaking of UA (1) and chiral symmetries:

Leutwyler-Smilga relation (LS), also connecting YM and full QCD
quantities (like WVR), “making” χYM out of much smaller χ:
χ
At T = 0 χYM = ̃→χ
≡χ ̃(T )
1+ χ
⟨q̄q⟩0 ∑ 1
mq
q=u,d,s
−1
where for the light quarks χ= 1
+ Cm
∑ mq ⟨q̄q⟩0
q=u,d,s
.
● Cm = small corrections of higher orders in small mq .
However, neglecting it, i.e., Cm = 0, would imply χYM = ∞.
Conversely, χYM = ∞ in LS returns the leading term of χ.
For axions, χYM is not needed ⇒ the leading term of χ will suffice.
● LS relation fixes the value of the correction at T = 0:
2
1 1 ⎛ 1 ⎞
= ∑ − χYM (0) ∑ .
Cm q=u,d,s mq ⟨q̄q⟩0 ⎝q=u,d,s mq ⟨q̄q⟩0 ⎠
● The conjecture on χ̃(T ) supported by Shore’s generalization of WV relation.
̃(T )
Chiral-limit condensate ⟨q̄q⟩0 (T ) and resulting χ
0.25
χ1/4 , δ =0
χ1/4 , δ =1
0.20
−
qq̄ 1/3
0
All values in [GeV]
0.15
0.10
0.05
0.00
0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
T/Tc
At density = 0, 2nd order chiral transition forces at T = Tc ≡ TCh the abrupt

transition to the NS-S asymptotic regime of vanishing UA (1) anomaly influence:
Mη′ (T ) → Mss̄ (T ), and Mη (T ) → MNS (T ) → Mπ (T ), and φ(T ) → 0.
Acceptable or even good for η ′ , but η would be in conflict with experiment.
Prediction good for η ′ , but for η not supported by any experiment

[Benić, Horvatić, Kekez and Klabučar, Phys. Rev. D 84 (2011) 016006.]:
Anomalous contribution from WVR:
ΧYM=H0.1757 GeVL4, ∆=0
1.2
1.0 Η' 2ΠT

Η0
0.8 ΗΗNS
S
MP @GeVD
ss
0.6 Η8
Η
0.4
0.2 Π
0.0
0.0 0.5 1.0 1.5
TTch
Shore’s generalized WV = 3rd example of tying UA (1) and SUA (3)

(fη0′ )2 Mη2′ + (fη0 )2 Mη2 = 1
(f 2 Mπ2
3 π
+ 2fK2 MK2 ) + 2Nf A (1)
√
fη0′ fη8′ Mη2′ + fη0 fη8 Mη2 = 2 2
3
(fπ2 Mπ2 − fK2 MK2 ) (2)
(fη8′ )2 Mη2′ + (fη8 )2 Mη2 = − 13 (fπ2 Mπ2 − 4fK2 MK2 ) (3)
The role of χYM taken over by the full QCD topological charge parameter A ,
χ
A= (4)
1 + χ( ⟨ūu⟩1 mu + 1
⟨d̄d⟩ md
+ ⟨s̄s⟩ ms )
1
A behaves with T as a full QCD quantity, but, at T = 0, A = χYM + O( N1c )

Again, A = ∞ returns the leading term of
−1 ′
χ= + Cm (5)
1
mu ⟨ūu⟩ + 1
md ⟨d̄d⟩
+ 1
ms ⟨s̄s⟩
Massive-quark condensates employed ⇒ crossover around T ∼ TCh

(Large Nc limit & approximating 3 condensates by ⟨q̄q⟩0 , returns the LS relation.)
√
QCD topological susceptibility χ(T ) gives axion mass × fa
● For all temperatures: ma2 (T ) fa2 = χ(T ) = full QCD topological susceptibility
mu md isospin
● At T = 0, ma2 fa2 = fπ2 Mπ2 → → (78.9 MeV)4
( mu + md )2 limit
● This agrees well with results, including χ(T ), from “our” DS-BSE chirally
well-behaved model (separable: simplified, but phenomenologically successful)
100
● Agrees well with χ(T ) from
1/4
χlatt.P
1/4
lattice studies of axion mass: χlatt.B
80 P
Petreczky & al. PLB (2016) and (− q (mq hq̄qi)−1)−1/4
P
All values in [MeV]
Borsany & al. Nature (2016) (− q (mq hq̄qi0)−1)−1/4
60
● χ(T ) from our usual DS-BSE
model: successful at T = 0,
no additional fitting for T > 0: 40
condensates ⟨q̄q⟩(T ) of massive

q = u, d, s essential to yield 20
good crossover T -dependence
of χ(T ) for good T -dependence
of η and η ′ masses. 0
0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25
T /Tc
Briefly on axions as solutions for Strong CP problem

● QCD has the Strong CP problem: no experimental evidence of any
CP-symmetry violation in strong interactions, in spite of its θ-term:
g2 ̃ bµν ̃ bµν ≡
LQCD = LQCD
CPsymmetric + θ̄ Fb F (F Fρσ )
1 µνρσ b

32π 2 µν 2
● The θ-term is a total divergence, but it cannot be discarded. It

contributes anyway (unlike in QED) due to nontrivial topological
structures in QCD – e.g., instantons (probably yielding, e.g.,
anomalously large Mη′ ⇒ important for solving the UA (1) problem).
● The experimental bound is mysteriosly small: ∣θ̄∣ < 10−10 . Why?!?
θ̄ = θQCD + Arg Det M̂q ⇒ setting θ̄ = 0 just “by hand” is fine-tuning.
How to get θ̄ ≈ 0 ?!?

● Nowadays preferred solution: a new particle beyond SM: axion a
● Axions are very interesting also for cosmology as candidates for
dark matter.
Axions as quasi-Goldstone bosons

● Peccei & Quinn introduced a new axial global symmetry U(1)PQ
which is broken spontaneously at some scale fa
(fa = free parameter of axion theories, determines absolute value of the
axion mass ma , but cancels from combinations such as ma (T )/ma (0).)
● the pseudoscalar axion field a(x) is the (would-be massless) Goldstone
boson of this spontaneous breaking. Then,
a g2 ̃ bµν + 1 ∂µ a ∂ µ a + La ψ
LQCD+ QCD
axion = LCPsymmetric + ( θ̄ + ) Fb F
fa 32π 2 µν 2 int
● But, the U(1)PQ symmetry is also broken explicitly by the gluon axial
anomaly through axion’s coupling with gluons ⇒ ma ≠ 0.
● Gluons generate an effective axion potential, which leads to the axion
expectation value ⟨a⟩ such that (θ̄ + ⟨a⟩/fa ) = 0, minimizing the potential
⇒ strong CP problem solved, irrespective of the initial θ̄.
(”Misalignment production” is relaxation from any value in the early
Universe towards the effective potential minimum at θ̄ = −⟨a⟩/fa . The
resulting axion oscillation energy is a ”cold dark matter” candidate.)
Evaluation of qq̄ condensates from propagators

Solving the gap SD equation ⇒ dressed propagators Sq (p)
The usual expression for the condensate of the flavor q for T > 0 becomes
d 3p
⟨q̄q⟩ = − Nc ⨋ Tr [Sq (p)] ≡ − Nc T ∑ ∫ Tr [Sq (p)]
p nq ∈Z (2π)3
Tr = trace in Dirac space. The combined integral-sum symbol says: when

T > 0, the 4-momentum integration Ð→ 3-momentum integr. & sum over
Matsubaras ωq = (2nq + 1)πT , nq ∈ Z.
● Well known: qq̄ condensates are finite only for massless quarks, mq = 0.
“Massive” condensates must be subtracted of their divergences.
● The arbitrariness of sensible procedures is in practice slight, i.e., only small
differences between the results of various sensible subtractions.
● First consider the (normalized) subtraction proposed on lattice by Burger+al
(2011), here applied to our condensates of light (u- and d-) quarks:
⟨ū u⟩(T ) − ⟨ū u⟩(0) + ⟨q̄ q⟩0 (0)

R⟨ψ̄ψ⟩ (T ) = R⟨ūu⟩ (T ) = .
⟨q̄ q⟩0 (0)
Comparison of subtracted & normalized lattice- and DS-condensates

1.0
Relative T -dependence Rhψ̄ψi
of the subtracted (and hūui(T )/hūui(0)
normalized) condensate 0.8
R⟨ψ̄ψ⟩ . The lattice data
points are from Fig. 6
0.6
of Kotov, Lombardo &
Rhψ̄ψi
Trunin, PLB794 (2019),
scaled for the critical 0.4
temperatures Tχ from
their Table 2, which
0.2
are different for the
“crosses” (lattice data
for mπ ≈ 370 MeV) and 0.0
“bars” (lattice data for
mπ ≈ 210 MeV). 0.6 0.8 1.0 1.2
T /Tc
1.4 1.6 1.8
The lower, green curve results from only the R⟨ψ̄ψ⟩ -subtraction of our u-quark
condensate. The upper, red curve is R⟨ūu⟩ (T ) when our u-quark condensate is
regularized in the usual way, by subtracting the current quark mass parameter
mu from the numerator of the dressed quark propagator.
Compare ∆l,s (T ) regularization of the lattice and DS condensates

⟨l̄ l(T )⟩ − ml
ms
⟨s̄ s(T )⟩
∆l,s (T ) = (l = u or d in isosymmetric limit)
⟨l̄ l(0)⟩ − ml
ms
⟨s̄ s(0)⟩
This is the most

1.0
usual (normalized) ∆l,s - with mass subtraction
subtraction on ∆l,s - w/o mass subtraction
the lattice. Very 0.8
good agreement
with Isserstedt+al
2019 & the lat- 0.6
tice (Borsanyi+al
∆l,s
2010). Red and

0.4
green curve, respec-
tively, again result
from our model 0.2
DS condensate
⟨ū u(T )⟩ with and
without subtraction 0.0
of mu from the
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
quark propagator T /Tc
numerator.
T -dependence of ⟨q̄q⟩ & decay const’s fP with χ & A

300
1/3
A1/4 −hūui1/3 fπ −hq̄qi0
χ1/4 −hs̄si1/3 fss̄
All values in [MeV] 250
200
150
100
50
0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
T /TCh
FKS scheme on Shore ⇒ how fP influence elements of the η-η ′ mass matrix:
√
fπ 2 4A 2 2A 2A
X= , MNS = Mπ2 + 2 , 2
MNSS = , MS2 = Mss̄
2
+ 2
fss̄ fπ fπ fss̄ fss̄
Zoomed η-η ′ complex

Mπ
Mss̄
1.0
Mη
Mη 0
● Mη′ (T ) is not changed much M η8
as condensates are changed M η0
from chiral to massive: Mη′ (T ) 0.9 MNS
falls again around TCh by 300
All values in [GeV]

MS
to 200 MeV, corresponding to MUA(1)
melting of ∼ 13 MUA (1) .
0.8
● But η does not exhibit any
mass drop at all, now. It
stays predominantly η8 till
anticrossing at ∼ 1.5 TCh . 0.7
Similarly η ∼ η0 long after TCh ,

′
and only after this anti-X with

η, η ′ tends to a pure ss̄. 0.6
0.6 0.8 1.0 1.2 1.4 1.6

T /TCh
T -dependence of MP (T ) up to T = 1.8TCh [Horvatić& al. Phys.Rev. D99 (2019) 014007]
● C(T ) ≠ const, adjusted to en-

able reaching arbitrary high T ’s, Mπ M η8 MNSS
1.2 Mss̄ M η0 A1/4
results otherwise very similar to
Mη MNS MUA(1)/2
previous case with C(T ) = C(0). MS
Mη 0
● Other limitations of rank-2 1.0
separable model make it hard to
find solutions beyond ∼ 1.8Tc .
All values in [GeV]

0.8
But it is enough to exhibit
cleanly the asymptotic regime
beyond anticrossing at 1.5TCh .
0.6
Along with A, influence on

anomalous masses is given by 0.4
MNS S and ( 21 )MUA (1) .
Utopistic in practice? - but in
0.2
principle, accurate experimental
knowledge of Mη′ (T ) would tell
us about A(T ) and thus about
0.0
χ(T ) ∝ ma (T )2 . 0.6 0.8 1.0 1.2 1.4 1.6 1.8
T /TCh
Summary of model results
● Our approach ties the UA (1) SB to the DChSB so closely, expressing

χ & A through mq ⟨q̄q⟩, q = u, d, s, that the restoration of the chiral
symmetry must lead to the restoration of the UA (1) symmetry at
least partially, and surely on the level of the η ′ & η masses.
● Full UA (1) restoration occurs together with the chiral one at
T = TCh only for the chiral-limit condensate ⟨qq̄⟩0 exhibiting
sharp phase transition. However, such an abrupt restoration is
in the real world excluded by the behavior of η.
● Condensates with explicit ChSB exhibit crossover, i.e., fall with T
much more slowly than ⟨qq̄⟩0 . Our “massive” condensates yield
χ(T ) in reasonable agreement with χ(T ) from lattice studies of the
T -dependence of axion mass.
Now, η does not exhibit any mass drop at all, while the significant
drop of the η ′ mass signals only a partial restoration of UA (1)
symmetry, consuming only about 31 MUA (1) in the vicinity of
T = TCh .
Conclusion: our results consistent with “Left C-plot”, with broken UA (1)
● DS ⟨q̄q⟩ consistent with chiral limit enabling 2nd order transition,

i.e., mu = 0 = md only for the 2 lightest flavors, but ms > mstri .
● For realistic explicit chiral breaking, the crossover creates an
intermediate region between the chiral restoration at T = TCh and
the η-η ′ anticrossing at T = 1.5 TCh which marks the effective UA (1)
restoration. The anomalous contributions then become so weak, that
the η-η ′ complex enters the NS-S asymptotic regime: φ(T ) → 0 and
Mη′ (T ) → MS (T ) → Mss̄ (T ) and Mη (T ) → MNS (T ) → Mπ (T ) .

Klabucar Zimany 2019

Uploaded by

Copyright:

Available Formats

Klabucar Zimany 2019

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Klabucar Zimany 2019

Uploaded by

Copyright:

Available Formats

UA (1) symmetry breaking is why η0 ≈ η ′ has an anomalous piece of mass Examples of UA (1) and Chiral symmetry interplay: Leutwyler-Smilga

How, how much, and when are UA (1) and chiral

(1) Physics Department, Faculty of Science – PMF, University of Zagreb, Croatia

The issue of the (effective) restoration of the UA (1) symmetry

... still debatable what happens with UA (1) symmetry restoration!

General renorm-group arguments (Pisarski:1983ms) ⇒ QCD with 3 degenerate

Quantum-level breaking of UA (1) causes anomalously high η ′ ≈ η0 mass

⇒ non-anomalous parts of the masses cancel in Witten-Veneziano rel. (WVR):

QCD topological susceptibility χ = a direct measure of UA (1) breaking ⇒

The 2nd example of tied breaking of UA (1) and chiral symmetries:

At density = 0, 2nd order chiral transition forces at T = Tc ≡ TCh the abrupt

Prediction good for η ′ , but for η not supported by any experiment

1.0 Η' 2ΠT

Shore’s generalized WV = 3rd example of tying UA (1) and SUA (3)

(fη8′ )2 Mη2′ + (fη8 )2 Mη2 = − 13 (fπ2 Mπ2 − 4fK2 MK2 ) (3)

A behaves with T as a full QCD quantity, but, at T = 0, A = χYM + O( N1c )

Massive-quark condensates employed ⇒ crossover around T ∼ TCh

condensates ⟨q̄q⟩(T ) of massive

Briefly on axions as solutions for Strong CP problem

● The θ-term is a total divergence, but it cannot be discarded. It

How to get θ̄ ≈ 0 ?!?

Axions as quasi-Goldstone bosons

Evaluation of qq̄ condensates from propagators

Tr = trace in Dirac space. The combined integral-sum symbol says: when

⟨ū u⟩(T ) − ⟨ū u⟩(0) + ⟨q̄ q⟩0 (0)

Comparison of subtracted & normalized lattice- and DS-condensates

Compare ∆l,s (T ) regularization of the lattice and DS condensates

This is the most

2010). Red and

T -dependence of ⟨q̄q⟩ & decay const’s fP with χ & A

Zoomed η-η ′ complex

All values in [GeV]

Similarly η ∼ η0 long after TCh ,

and only after this anti-X with

0.6 0.8 1.0 1.2 1.4 1.6

T -dependence of MP (T ) up to T = 1.8TCh [Horvatić& al. Phys.Rev. D99 (2019) 014007]

● C(T ) ≠ const, adjusted to en-

All values in [GeV]

Along with A, influence on

Summary of model results

● Our approach ties the UA (1) SB to the DChSB so closely, expressing

● DS ⟨q̄q⟩ consistent with chiral limit enabling 2nd order transition,

You might also like