Going beyond Mahabaleshwar: Search for

CPT Violation

Anirban Kundu

University of Calcutta
February 20, 2013
PCPV 2013, Mahabaleshwar

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Part I: Introduction to CPT violation

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Parity violation can be incorporated through the current (Lee and Yang)

P: ψ̄γ µ (1 − γ5 )ψ =⇒ ψ̄γ µ (1 + γ5 )ψ

Maximal P violation for weak interaction

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

This is not enough for CP violation, you need the coupling to be complex
too (Kobayashi and Maskawa)

CP : g ψ̄1 γ µ (1 − γ5 )ψ2 + h.c. =⇒ g ψ̄2 γ µ (1 − γ5 )ψ1 + h.c.

H.c. involves g ∗ , but gauge couplings are real. Introduce quark mixing.
Large CP violation for B systems, but too small to explain nb /nγ

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

CPT, taken in any order, is the only combination of C,P,T that is still
conserved.
I Pauli (1940): Spin-statistics theorem, requires Lorentz invariance
I Schwinger (1951): Spin-statistics theorem, implicit use of CPT
theorem
I Lüders, Pauli, Bell (1954-55): Proof of CPT theorem
I Jost (1958): General proof for axiomatic QFT

Why, then, should one look for CPT violation?

Motivation 1: George Mallory about Mt. Everest: Because it is there.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

CPT, taken in any order, is the only combination of C,P,T that is still
conserved.
I Pauli (1940): Spin-statistics theorem, requires Lorentz invariance
I Schwinger (1951): Spin-statistics theorem, implicit use of CPT
theorem
I Lüders, Pauli, Bell (1954-55): Proof of CPT theorem
I Jost (1958): General proof for axiomatic QFT

Why, then, should one look for CPT violation?

Motivation 1: George Mallory about Mt. Everest: Because it is there.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

CPT transformation on Fermion current:

ψ̄aCPT (t, x)Γi ψbCPT (t, x) = ψ̄b (−t, −x)ΓCPT

i ψa (−t, −x)

Γi : {1, γ5 , γµ , γµ γ5 , σµν } =⇒ {1, −γ5 , −γµ , −γµ γ5 , σµν }

CPT is a good symmetry if

(CPT )L(t, x)(CPT )−1 = L(−t, −x)

That gives you an idea of what terms can potentially violate CPT.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Theorem
CPT is a good symmetry of any local Lorentz-invariant axiomatic
quantum field theory with a unique vacuum state.

You can never construct a Lorentz-invariant QFT with a hermitian

Hamiltonian that violates CPT.

Proof of CPT theorem is not straightforward

(see, e.g., Streater and Wightman)

Proof.
Consider real scalar field −→ C is conserved
PT is x µ → −x µ , proper LT, continuously connected to identity
In Euclidean space, just like a 4-d rotation — must be conserved
PT is always a good symmetry for real scalar field

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Theorem
CPT is a good symmetry of any local Lorentz-invariant axiomatic
quantum field theory with a unique vacuum state.

You can never construct a Lorentz-invariant QFT with a hermitian

Hamiltonian that violates CPT.

Proof of CPT theorem is not straightforward

(see, e.g., Streater and Wightman)

Proof.
Consider real scalar field −→ C is conserved
PT is x µ → −x µ , proper LT, continuously connected to identity
In Euclidean space, just like a 4-d rotation — must be conserved
PT is always a good symmetry for real scalar field

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Consequences of CPT conservation
I Particle and antiparticle must have same mass and opposite electric
charge
I Particle and antiparticle, if unstable, must have same decay width
Not true if stationary states are particle-antiparticle combinations
1 1
KL ≈ √ (K 0 + K 0 ) , KS ≈ √ (K 0 − K 0 ) ,
2 2
MKL 6= MKS , ΓKL 6= ΓKS

I Particle and antiparticle must have equal and opposite mag. moment
I Hydrogen and antihydrogen must have identical spectra ....

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Consequences of CPT conservation
I Particle and antiparticle must have same mass and opposite electric
charge
I Particle and antiparticle, if unstable, must have same decay width
Not true if stationary states are particle-antiparticle combinations
1 1
KL ≈ √ (K 0 + K 0 ) , KS ≈ √ (K 0 − K 0 ) ,
2 2
MKL 6= MKS , ΓKL 6= ΓKS

I Particle and antiparticle must have equal and opposite mag. moment
I Hydrogen and antihydrogen must have identical spectra ....
I T violation necessarily means CP violation, like EDM

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Consequences of CPT conservation
I Particle and antiparticle must have same mass and opposite electric
charge
I Particle and antiparticle, if unstable, must have same decay width
Not true if stationary states are particle-antiparticle combinations
1 1
KL ≈ √ (K 0 + K 0 ) , KS ≈ √ (K 0 − K 0 ) ,
2 2
MKL 6= MKS , ΓKL 6= ΓKS

I Particle and antiparticle must have equal and opposite mag. moment
I Hydrogen and antihydrogen must have identical spectra ....
I T violation necessarily means CP violation, like EDM

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

In a local QFT with well-defined asymptotic states, CPT violation needs
Lorentz violation
[Greenberg, PRL 2002]

The reverse is not true.

Motivation 2: Strings are extended objects, so nonlocal. Critical

dimensionality d > 4, Higher dimensional breaking of Lorentz covariance
incorporated in a 4-d world?
Lorentz symmetry can be broken in noncommutative FT too.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

In a local QFT with well-defined asymptotic states, CPT violation needs
Lorentz violation
[Greenberg, PRL 2002]

The reverse is not true.

Motivation 2: Strings are extended objects, so nonlocal. Critical

dimensionality d > 4, Higher dimensional breaking of Lorentz covariance
incorporated in a 4-d world?
Lorentz symmetry can be broken in noncommutative FT too.

Motivation 3: Asymptotic states for q, q̄ are not well-defined

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

In a local QFT with well-defined asymptotic states, CPT violation needs
Lorentz violation
[Greenberg, PRL 2002]

The reverse is not true.

Motivation 2: Strings are extended objects, so nonlocal. Critical

dimensionality d > 4, Higher dimensional breaking of Lorentz covariance
incorporated in a 4-d world?
Lorentz symmetry can be broken in noncommutative FT too.

Motivation 3: Asymptotic states for q, q̄ are not well-defined

Motivation 4: CPT might not be valid near extreme gravity regions, like
near the black holes

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

In a local QFT with well-defined asymptotic states, CPT violation needs
Lorentz violation
[Greenberg, PRL 2002]

The reverse is not true.

Motivation 2: Strings are extended objects, so nonlocal. Critical

dimensionality d > 4, Higher dimensional breaking of Lorentz covariance
incorporated in a 4-d world?
Lorentz symmetry can be broken in noncommutative FT too.

Motivation 3: Asymptotic states for q, q̄ are not well-defined

Motivation 4: CPT might not be valid near extreme gravity regions, like
near the black holes

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Lorentz violation
Lorentz breaking may be spontaneous (like SSB, but VEV to, say, a
vector field) or explicit
The low-energy effective theory (SME) contains operators whose
coefficients are Lorentz breaking

L = −(aL )µij L̄i γ µ Lj − (aR )µij R̄i γ µ Rj

Physics depends on direction ! [Colladay and Kostelecky, PRD 1998]

I Lorentz transformations on the frame (observer) [passive] or on the

fields (particle) [active]
I Should be inverse of each other if Lorentz symmetry is respected
I With LV, they are no longer so. Particle transformations are
physically important.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Lorentz violation
Lorentz breaking may be spontaneous (like SSB, but VEV to, say, a
vector field) or explicit
The low-energy effective theory (SME) contains operators whose
coefficients are Lorentz breaking

L = −(aL )µij L̄i γ µ Lj − (aR )µij R̄i γ µ Rj

Physics depends on direction ! [Colladay and Kostelecky, PRD 1998]

I Lorentz transformations on the frame (observer) [passive] or on the

fields (particle) [active]
I Should be inverse of each other if Lorentz symmetry is respected
I With LV, they are no longer so. Particle transformations are
physically important.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Observables that depend on sidereal time are smoking gun signals of
Lorentz violation

LV operators might be CPT-even or CPT-odd

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

For each Dirac fermions, there are 44 possible observable LV terms in
nonrelativistic limit. 20 among them are CPT-odd.
[Kostelecky and Russell, RMP 2008, 0801.0287]

L = ψ̄i (iΓµ D µ − Mij )ψj

ij 1 ij
Γµ = γµ δij + cµν γ ν + dµνij
γ µ γ5 + eµij + ifµij γ5 + gµκρ σ κρ
2
1 ij µν
Mij = mij + im5ij γ5 + aµij γ µ + bµij γ µ γ5 + hµν σ
2
Blue terms are CPT-odd. Red terms are LV but CPT-even
CPT (and LV) tests have been carried out in gravity, photon, charged
lepton, neutrino, proton, neutron, and meson sectors.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

CPT and LV tests
I Astrophysics
— Pulsar rates
— CMB polarization
— Birefringence
I Atomic physics
— K/He magnetometer
— H maser
— QED tests with Penning trap

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

CPT and LV tests
I Astrophysics
— Pulsar rates
— CMB polarization
— Birefringence
I Atomic physics
— K/He magnetometer
— H maser
— QED tests with Penning trap
I Optics
— Optical and microwave resonators
— Atomic clocks
— Lunar laser ranging

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

CPT and LV tests
I Astrophysics
— Pulsar rates
— CMB polarization
— Birefringence
I Atomic physics
— K/He magnetometer
— H maser
— QED tests with Penning trap
I Optics
— Optical and microwave resonators
— Atomic clocks
— Lunar laser ranging

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Several particle physics experiments too.
1. Neutrino oscillation
2. (g − 2)e,µ
3. e + e − annihilation
4. Particle and antiparicle mass measurement
5. Neutral meson oscillation
6. and others ...

Sidereal time variation for observables =⇒ LV

Difference between particle and antiparticle =⇒ LV + CPTV

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Not all measurements are equally precise.

Define p = [Obs(particle) − Obs(antiparticle)]/Obs(average)

p(mW ) −0.002 ± 0.007 p(mπ ) (2 ± 5) × 10−4

p(mp ) < 2 × 10−9 p(mn ) (9 ± 6) × 10−5
p(ge + ) (−0.5 ± 2.1) × 10−12 p(gµ+ ) (−0.11 ± 0.12) × 10−8
mt − mt̄ (−1.4 ± 2.0) GeV

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Take, as an example, the charge equality of electron and positron

e+ + e− → γ + γ

Direct PDG: (Qe + + Qe − )/|Qe − | < 4 × 10−8

Much better bound assuming charge conservation

(Qe + + Qe − )/|Qe − | ∼ Qγ /|Qe − | < 10−33

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Muons are stored in the storage ring — their spin precession frequency ω
can be measured very precisely

I Is ω+ = ω− ?
I Is there a sidereal variation?

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Muons are stored in the storage ring — their spin precession frequency ω
can be measured very precisely

I Is ω+ = ω− ?
I Is there a sidereal variation?
Both the answers are consistent with zero — one of the most precise
measurements. [Muon g − 2 Collab., PRL 2008]

However, this does not say anything, for example, about the CPT
violating parameters in the τ sector — CPT violation can be a flavour
dependent thing.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Muons are stored in the storage ring — their spin precession frequency ω
can be measured very precisely

I Is ω+ = ω− ?
I Is there a sidereal variation?
Both the answers are consistent with zero — one of the most precise
measurements. [Muon g − 2 Collab., PRL 2008]

However, this does not say anything, for example, about the CPT
violating parameters in the τ sector — CPT violation can be a flavour
dependent thing.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Part II: CPT violation in the K and B systems

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

K 0 − K 0 mixing and CPT violation

Beam of neutral Kaon in its rest frame

|K (t)i = a1 (t)|K 0 i + a2 (t)|K 0 i , hK 0 |K 0 i = 0

b W s
b c s s
b
c
u,c,t u,c,t c
s W b s c b
s
b

The evolution is given by

M11 − 2i Γ11 M12 − 2i Γ12

    
∂ a1 (t) a1 (t)
i = ∗
∂t a2 (t) M12 − 2i Γ∗12 M22 − 2i Γ22 a2 (t)

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

K 0 − K 0 mixing and CPT violation

Beam of neutral Kaon in its rest frame

|K (t)i = a1 (t)|K 0 i + a2 (t)|K 0 i , hK 0 |K 0 i = 0

b W s
b c s s
b
c
u,c,t u,c,t c
s W b s c b
s
b

The evolution is given by

M11 − 2i Γ11 M12 − 2i Γ12

    
∂ a1 (t) a1 (t)
i = ∗
∂t a2 (t) M12 − 2i Γ∗12 M22 − 2i Γ22 a2 (t)

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

The Hamiltonian matrix H can be parametrized by 7 parameters as
relative phase between H12 and H21 is meaningless.
Diagonalize with eigenvalues λ1 and λ2 :
i
λ1 − λ2 = ∆M + ∆Γ , ∆M = M1 − M2 , ∆Γ = Γ2 − Γ1 .
2

M = (M11 + M22 )/2 , Γ = (Γ11 + Γ22 )/2 , ∆M , ∆Γ

H22 − H11 |H12 |2 − |H21 |2

θ= χ=
∆M − 2i ∆Γ |H12 |2 + |H21 |2

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

The Hamiltonian matrix H can be parametrized by 7 parameters as
relative phase between H12 and H21 is meaningless.
Diagonalize with eigenvalues λ1 and λ2 :
i
λ1 − λ2 = ∆M + ∆Γ , ∆M = M1 − M2 , ∆Γ = Γ2 − Γ1 .
2

M = (M11 + M22 )/2 , Γ = (Γ11 + Γ22 )/2 , ∆M , ∆Γ

H22 − H11 |H12 |2 − |H21 |2

θ= χ=
∆M − 2i ∆Γ |H12 |2 + |H21 |2

θ is CPT violating =⇒ CPT conservation: M11 = M22 , Γ11 = Γ22

χ is CPT conserving but T violating.
Mass eigenstates: KL and KS

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

The Hamiltonian matrix H can be parametrized by 7 parameters as
relative phase between H12 and H21 is meaningless.
Diagonalize with eigenvalues λ1 and λ2 :
i
λ1 − λ2 = ∆M + ∆Γ , ∆M = M1 − M2 , ∆Γ = Γ2 − Γ1 .
2

M = (M11 + M22 )/2 , Γ = (Γ11 + Γ22 )/2 , ∆M , ∆Γ

H22 − H11 |H12 |2 − |H21 |2

θ= χ=
∆M − 2i ∆Γ |H12 |2 + |H21 |2

θ is CPT violating =⇒ CPT conservation: M11 = M22 , Γ11 = Γ22

χ is CPT conserving but T violating.
Mass eigenstates: KL and KS

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

The three important parameters for the Kaon sector:

A(KS → 2π)I =2 1
ω = |∆I | =
6
A(KS → 2π)I =0 2
A(KL → 2π)I =0
 = CP
/
A(KS → 2π)I =0
A(KL → 2π)I =2 A(KS → 2π)I =0 − A(KL → 2π)I =0 A(KS → 2π)I =2
0 = √
2[A(KS → 2π)I =0 ]2

Related parameters:

A(KL → π + π − ) 0
η+− = =  + √ ≈  + 0
A(KS → π + π − ) 1 + ω/ 2
A(KL → π 0 π 0 ) 20
η00 = 0 0
=− √ ≈  − 20
A(KS → π π ) 1 − 2ω

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

The three important parameters for the Kaon sector:

A(KS → 2π)I =2 1
ω = |∆I | =
6
A(KS → 2π)I =0 2
A(KL → 2π)I =0
 = CP
/
A(KS → 2π)I =0
A(KL → 2π)I =2 A(KS → 2π)I =0 − A(KL → 2π)I =0 A(KS → 2π)I =2
0 = √
2[A(KS → 2π)I =0 ]2

Related parameters:

A(KL → π + π − ) 0
η+− = =  + √ ≈  + 0
A(KS → π + π − ) 1 + ω/ 2
A(KL → π 0 π 0 ) 20
η00 = 0 0
=− √ ≈  − 20
A(KS → π π ) 1 − 2ω

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

 1
(1 + S,L )K 0 ± (1 − S,L )K 0


KS,L = p
2
2(1 + |S,L | )
iImM12 − 12 ImΓ12 ± 1
MK̄ 0 − MK 0 − 2i (ΓK̄ 0 − ΓK 0 )


2
S,L =
∆M + 2i ∆Γ
=  ± δ̄

If CPT is conserved, δ̄ = 0. The reverse is not true!

CPT violating parameters enter into the definition of the

states and hence affect the observables.

Re(δ̄) = (2.3 ± 2.7) × 10−4 , (2.51 ± 2.25) × 10−4

Im(δ̄) = (0.4 ± 2.1) × 10−5 , (−1.5 ± 1.6) × 10−5
[KLOE 2006, KTeV 2011]

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

 1
(1 + S,L )K 0 ± (1 − S,L )K 0


KS,L = p
2
2(1 + |S,L | )
iImM12 − 12 ImΓ12 ± 1
MK̄ 0 − MK 0 − 2i (ΓK̄ 0 − ΓK 0 )


2
S,L =
∆M + 2i ∆Γ
=  ± δ̄

If CPT is conserved, δ̄ = 0. The reverse is not true!

CPT violating parameters enter into the definition of the

states and hence affect the observables.

Re(δ̄) = (2.3 ± 2.7) × 10−4 , (2.51 ± 2.25) × 10−4

Im(δ̄) = (0.4 ± 2.1) × 10−5 , (−1.5 ± 1.6) × 10−5
[KLOE 2006, KTeV 2011]

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

B 0 − B 0 mixing and CPT violation

Formalism is almost identical [Lavoura, Ann. Phys. (1991)]

The constraints can be completely different — CPT violation may be
flavour-dependent. Also, ∆M large but ∆Γ small.

I Lifetime difference can be significant

I CPT violation may affect direct CP-violating asymmetries, including
semileptonic and dileptonic
I For semileptonic decays B, B̄ → `± X ∓ f , the time-ordering of
leptonic and hadronic decays may change due to CPT violation
[Datta, Paschos, Singh (PLB 2002), Balaji, Horn, Paschos (PRD 2003),
Xing (PRD 1994, PLB 1999)]

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

B 0 − B 0 mixing and CPT violation

Formalism is almost identical [Lavoura, Ann. Phys. (1991)]

The constraints can be completely different — CPT violation may be
flavour-dependent. Also, ∆M large but ∆Γ small.

I Lifetime difference can be significant

I CPT violation may affect direct CP-violating asymmetries, including
semileptonic and dileptonic
I For semileptonic decays B, B̄ → `± X ∓ f , the time-ordering of
leptonic and hadronic decays may change due to CPT violation
[Datta, Paschos, Singh (PLB 2002), Balaji, Horn, Paschos (PRD 2003),
Xing (PRD 1994, PLB 1999)]

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

CPT violation in Mixing

Introduce CPT violation in the Hamiltonian matrix through the

parameter δ, can be complex:
H22 − H11
δ= √ ,
H12 H21
Solutions:
     
δ δ
λ = H11 + H12 α y + , H22 − H12 α y + ,
2 2
q
2
p
where y = 1 + δ4 and α = H21 /H12 .

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Eigenstates

|BH i = p1 |B 0 i + q1 |B 0 i ,
|BL i = p2 |B 0 i − q2 |B 0 i .

Normalisation
|p1 |2 + |q1 |2 = |p2 |2 + |q2 |2 = 1 .
Define
   
q1 δ q2 δ η1
η1 = = y+ α; η2 = = y− α; ω= .
p1 2 p2 2 η2

δ and hence y are CPT violating. If |δ|  1, y ≈ 1.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

 ~
∆aµ = r1 aµ1 − r2 aµ2 , β µ = (1, β)

1 β µ ∆aµ
δ=−
2 ∆M − i∆Γ/2

Numerator varies with time as β~ rotates with ∆~a.

BaBar (0711.2713) got δ consistent with zero (first two spectral powers)
from OS dilepton events
Belle(1203.0930):

Re(δd ) = (−3.8 ± 9, 9) × 10−2

Im(δd ) = (1.14 ± 0.93) × 10−2

Similar results in K (KTeV) and D (FOCUS) systems. Time for Bs .

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Consider decay to a CP eigenstate f .

Af = hf |H|Bq i , Āf = hf |H|Bq i .

Āf Āf
ξf1 = η1 , ξf2 = η2 .
Af Af

In the SM, both are equal and ξf1 = ξf2 = ξf . For single-channel
processes, |ξf | = 1.

The untagged rate ΓU [f , t] = Γ(Bq (t) → f ) + Γ(B̄q (t) → f )

1 ∞
Z
Br [f ] = dt Γ[f , t] .
2 0

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Consider decay to a CP eigenstate f .

Af = hf |H|Bq i , Āf = hf |H|Bq i .

Āf Āf
ξf1 = η1 , ξf2 = η2 .
Af Af

In the SM, both are equal and ξf1 = ξf2 = ξf . For single-channel
processes, |ξf | = 1.

The untagged rate ΓU [f , t] = Γ(Bq (t) → f ) + Γ(B̄q (t) → f )

1 ∞
Z
Br [f ] = dt Γ[f , t] .
2 0

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Assume |δ| to be small, make Taylor expansion up to δ n , n ≤ 2

"    
2 −Γq t ∆Γq t ∆Γq t
ΓU [f , t] = |Af | e (...) cosh + (...) sinh
2 2
#
+(...) cos (∆Mq t) + (...) sin (∆Mq t)

Simplification:
For Bd system, ∆Γd  1, cosh → 1, sinh → 0, easier fit to decay profile
For |δ|  1, keep only the linear terms. For Bs , keep Γs too

"
|Af |2 1
Br [f ] = {2 − Im(δ)Im(ξf )}
2 Γs
#
Γs ∆Γs
+ Im(δ)Im(ξf ) + Re(ξf ) .
(∆m)2 + (Γs )2 (Γs )2

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Assume |δ| to be small, make Taylor expansion up to δ n , n ≤ 2

"    
2 −Γq t ∆Γq t ∆Γq t
ΓU [f , t] = |Af | e (...) cosh + (...) sinh
2 2
#
+(...) cos (∆Mq t) + (...) sin (∆Mq t)

Simplification:
For Bd system, ∆Γd  1, cosh → 1, sinh → 0, easier fit to decay profile
For |δ|  1, keep only the linear terms. For Bs , keep Γs too

"
|Af |2 1
Br [f ] = {2 − Im(δ)Im(ξf )}
2 Γs
#
Γs ∆Γs
+ Im(δ)Im(ξf ) + Re(ξf ) .
(∆m)2 + (Γs )2 (Γs )2

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

The B mesons can be tagged:
ΓT [f , t] = Γ(Bq (t) → f ) − Γ(B̄q (t) → f )
I Fit both untagged and tagged profiles
I Re(δ) from cos and sinh terms, Im(δ) from sin and cos terms
ΓT [f , t] Γ(Bq (t) → f ) − Γ(B̄q (t) → f )
ACPT (f , t) = = ,
ΓU [f , t] Γ(Bq (t) → f ) + Γ(B̄q (t) → f )
Goes to usual CP asymmetry ACP if δ = 0.
No change in semileptonic CP asymmetry if only new physics is CPT
violation.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

The B mesons can be tagged:
ΓT [f , t] = Γ(Bq (t) → f ) − Γ(B̄q (t) → f )
I Fit both untagged and tagged profiles
I Re(δ) from cos and sinh terms, Im(δ) from sin and cos terms
ΓT [f , t] Γ(Bq (t) → f ) − Γ(B̄q (t) → f )
ACPT (f , t) = = ,
ΓU [f , t] Γ(Bq (t) → f ) + Γ(B̄q (t) → f )
Goes to usual CP asymmetry ACP if δ = 0.
No change in semileptonic CP asymmetry if only new physics is CPT
violation.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

 R∞ R∞
0
dt ΓT [f , t] dt [Γ(Bq (t) → f ) − Γ(B̄q (t) → f )]
ACPT (f ) = R ∞ = R0∞ .
0
dt ΓU [f , t] 0
dt [Γ(Bq (t) → f ) + Γ(B̄q (t) → f )]

0.37 0.37

0.04
0.36 0.36

0.02 0.35 0.35

ACPT
0.34 0.34
ACPT
ACPT

0.00
0.33 0.33
!0.02
0.32 0.32

!0.04 0.31 0.31

!0.10 !0.05 0.00 0.05 0.10 !0.10 !0.05 0.00 0.05 0.10 !0.10 !0.05 0.00 0.05 0.10
Re!∆" Re!∆" Im!∆"

Top to bottom: Im/Re(δ) = −0.1, 0.0.1

[AK, Nandi, Patra, PRD 2010]

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

 I Consider a specific example
Bs , Bs → Ds± K ∓
I Can proceed through
∗
b → c ūs [∝ Vcb Vus ] and
b → uc̄s [∝ Vub Vcs∗ = exp(−iγ)].
I Only tree-level in SM and ∝ λ3 . Comparable rates.

Br(Bs → Ds K ) = (1.90 ± 0.23) × 10−4 [LHCb 2012]

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

 I Consider a specific example
Bs , Bs → Ds± K ∓
I Can proceed through
∗
b → c ūs [∝ Vcb Vus ] and
b → uc̄s [∝ Vub Vcs∗ = exp(−iγ)].
I Only tree-level in SM and ∝ λ3 . Comparable rates.

Br(Bs → Ds K ) = (1.90 ± 0.23) × 10−4 [LHCb 2012]

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Γ(Bs (t) → f ) − Γ(Bs (t) → f ) = [P1 sinh(∆Γs t/2) + Q1 cosh(∆Γs t/2)
f ≡ Ds+ K − , Tagged +R1 cos(∆Ms t) + S1 sin(∆Ms t)]
×e −Γs t |Af |2 ,
Γ(Bs (t) → f ) + Γ(Bs (t) → f ) = [P2 sinh(∆Γs t/2) + Q2 cosh(∆Γs t/2)
Untagged +R2 cos(∆Ms t) + S2 sin(∆Ms t)]
×e −Γs t |Af |2 ,

Absence of CPT violation means P1 = Q1 = R2 = S2 = 0

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Similar observables P̄1 − S̄2 for Bs → f¯(= Ds− K + )

R1 + R̄1 Re(δ) Q2 − Q̄2 Im(δ)

= , = .
P2 + P̄2 2 S1 − S̄1 2

Hadronic uncertainties and BSM effects in mixing cancel out in the ratio!

One can refine the analysis. LHCb with 200 fb−1 can reach up to
Re(δ) ∼ 0.1
[AK, Nandi, Patra, Soni, PRD 2013]

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

CP violation can be present in decay only and not mixing. Parametrize by
some complex parameter yf .

A(Bs → Ds+ K − ) = T1 e iγ (1 − yf )
A(Bs → Ds− K + ) = T2 (1 + yf∗ )
A(Bs → Ds+ K − ) = T2 (1 − yf )
A(Bs → Ds− K + ) = T1 e −iγ (1 + yf∗ )

hBr(Bs → Ds+ K − )i − hBr(Bs → Ds− K + )i

ACPT
br =
hBr(Bs → Ds+ K − )i + hBr(Bs → Ds− K + )i
Re(yf )
= −2 ≈ −2 Re(yf )
1 + |yf |2

LHCb at 200 fb−1 : Re(yf ) ∼ 0.003

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

CP violation can be present in decay only and not mixing. Parametrize by
some complex parameter yf .

A(Bs → Ds+ K − ) = T1 e iγ (1 − yf )
A(Bs → Ds− K + ) = T2 (1 + yf∗ )
A(Bs → Ds+ K − ) = T2 (1 − yf )
A(Bs → Ds− K + ) = T1 e −iγ (1 + yf∗ )

hBr(Bs → Ds+ K − )i − hBr(Bs → Ds− K + )i

ACPT
br =
hBr(Bs → Ds+ K − )i + hBr(Bs → Ds− K + )i
Re(yf )
= −2 ≈ −2 Re(yf )
1 + |yf |2

LHCb at 200 fb−1 : Re(yf ) ∼ 0.003

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Triple-product asymmetries

I Consider B → V1 V2

B(p) → V1 (k1 , ε1 ) + V2 (k2 , ε2 )

I Construct α ≡ ~k1 .(~ε1 × ~ε2 )

I The asymmetry

Γ(α > 0) − Γ(α < 0)

Γ(α > 0) + Γ(α < 0)

is odd under the time-reversal operator T. If CPT holds, this is a

signal of CP violation.
I TP asymmetries should be observables in other systems too.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

I A lot of TPs are zero in SM but nonzero in BSM with a second
amplitude. TPs can be nonzero even if the strong phase difference is
zero.
I One can also relate the s,p,d wave amplitudes with the so-called
transversity amplitudes A0 , A|| , A⊥
I Final state decay distributions probe the interference terms of these
amplitudes — probe for T violation.
I Some asymmetries are zero in SM and CPT conserving BSM but
become nonzero in SM + CPTV.
[AK and Patra 2013]

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Conclusions

I CPT is supposed to be a good symmetry in any local

Lorentz-invariant QFT.
I However, CPT may be violated if LI is broken. LI can be broken by
string interactions, noncommutative coordinates, strong gravity ....
I CPTV needs LV, the reverse is not true.
I Various particle physics tests for CPTV: neutrinos, (g − 2)e,µ , etc.
I Strong bounds in K and Bd systems.
I Time to look in Bs , LHC can uncover such signals. We should stay
tuned.
Thank you. Bon appetit.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation

Conclusions

I CPT is supposed to be a good symmetry in any local

Lorentz-invariant QFT.
I However, CPT may be violated if LI is broken. LI can be broken by
string interactions, noncommutative coordinates, strong gravity ....
I CPTV needs LV, the reverse is not true.
I Various particle physics tests for CPTV: neutrinos, (g − 2)e,µ , etc.
I Strong bounds in K and Bd systems.
I Time to look in Bs , LHC can uncover such signals. We should stay
tuned.
Thank you. Bon appetit.

Anirban Kundu Going beyond Mahabaleshwar: Search for CPT Violation