12.18.04 · quantum / gauge-and-symmetry

Theta-vacua, the vacuum angle, and the strong-CP problem

shipped3 tiersLean: none

Anchor (Master): Weinberg, S., *The Quantum Theory of Fields, Vol. 2: Modern Applications* (Cambridge, 1996), §23.6 (the $\theta$-vacuum, instanton-induced tunnelling, the $\theta F\tilde F$ term, and the strong-CP problem); Coleman, S., *Aspects of Symmetry* (Cambridge, 1985), Ch. 7 §3-7; 't Hooft, G., *Phys. Rev. D* 14, 3432 (1976) (the instanton computation and the $U(1)_A$ resolution)

Intuition Beginner

A quantum system can sit in more than one lowest-energy configuration, and when it can quietly leak from one to another, the true resting state is a blend of all of them. The strong nuclear force, described by the theory of quarks and gluons, turns out to have not one resting state but a whole ladder of them, labelled by a whole number you can picture as a winding count. The gluon field can wind around itself once, twice, any number of times, and each winding is its own lowest-energy candidate.

These winding states are not sealed off from each other. The field can tunnel from one rung of the ladder to the next, the way a particle can leak through a wall it has too little energy to climb over. Because the tunnelling links every rung, the real vacuum is a mixture of all the winding states at once. The mixture carries one free dial, an angle written , that says how the rungs are weighted relative to one another. Different settings of the dial give genuinely different physics.

Why bother with this? Because that dial controls whether the strong force treats a process and its mirror-and-antimatter image the same way. Experiment says it does, to extraordinary accuracy. So the dial sits almost exactly at zero, and nobody built into the theory a reason why. That unexplained near-zero is the strong-CP problem.

Visual Beginner

Picture a landscape of identical valleys in a row, each valley one notch higher in winding than the last, like a tilted egg carton stretching off in both directions. A classical ball would settle in one valley and stay. A quantum field can tunnel through the ridges between valleys, so it does not belong to any single valley; it spreads across all of them as one connected state.

The way the field spreads across the valleys is fixed by one angle. Set the angle to zero and the field crosses every valley with the same sign; set it elsewhere and each valley is reached with a different twist in the quantum phase. This is the same mathematics that governs an electron in a crystal, where rows of identical atomic wells produce energy bands and the electron state carries a phase that advances valley to valley. The strong-force vacuum is the field-theory version of that crystal electron, and the angle is its band label.

Worked example Beginner

A field hops along a row of valleys, and the rule for the lowest-energy mixture says that moving from one valley to the next valley adds a fixed phase twist set by the angle . Take to be one-third of a full turn, so each hop multiplies the contribution by a phase that advances by degrees. Start by giving valley number the weight .

Step 1. Fix the rule. Each step to the next valley advances the phase by degrees.

Step 2. Walk three valleys up. Valley sits at degrees, valley at degrees, valley at degrees, which is back to degrees.

Step 3. Read off valley . After three hops the phase has wound through a full turn and returned to where valley started. So valley carries the same phase as valley .

Step 4. Check the special setting. If instead , every valley carries phase : the field reaches all valleys in step, with no twist at all. That even, untwisted blend is the setting the strong force appears to sit at.

What this tells us: the angle is exactly the per-step phase twist that distinguishes one vacuum mixture from another. At there is no twist and the physics respects the mirror-and-antimatter symmetry; switch the angle on and that symmetry is spoiled. The whole strong-CP puzzle is the question of why nature chose the untwisted setting.

Check your understanding Beginner

Formal definition Intermediate+

Throughout, spacetime is Euclidean for the tunnelling computation and Minkowski (mostly-minus) for the Hamiltonian picture, and the gauge group is a compact simple group , with the case of physical interest. The gauge field is with field strength of 03.07.05, and the dual field strength is .

The Pontryagin index (topological charge) of a finite-action Euclidean configuration is $$ \nu ;=; \frac{g^2}{32\pi^2} \int d^4x; F^a_{\mu\nu},\tilde F^{a,\mu\nu} ;\in; \mathbb{Z}, $$ integer-valued by the argument of 03.07.0703.07.08: finite action forces the gauge field to approach a pure gauge at the boundary at infinity, and the integral computes the degree of the resulting map , an element of . The BPST instanton of 03.07.07 is the minimal-action representative with .

In the temporal gauge , the residual symmetry is time-independent gauge transformations. These split into small gauge transformations, continuously connected to the identity, and large gauge transformations of winding number . Physical states must be invariant under small gauge transformations but transform by a representation under large ones. Label by the candidate vacuum built over the spatial gauge field of winding ; the large-gauge operator acts as a ladder, . Because commutes with the Hamiltonian and with all gauge-invariant operators, the physical vacua are its eigenstates.

The -vacuum is the one-parameter family $$ |\theta\rangle ;=; \sum_{n \in \mathbb{Z}} e^{in\theta},|n\rangle, \qquad T|\theta\rangle = e^{-i\theta}|\theta\rangle, \qquad \theta \in [0, 2\pi), $$ the simultaneous eigenstates of . The angle is a superselection label: no local gauge-invariant operator connects different , so each defines a distinct, internally consistent theory. Equivalently, the Euclidean path integral computing amplitudes in the sector weights each topological sector by , which is reproduced by adding to the Lagrangian the -term $$ \mathcal{L}\theta ;=; \theta,\frac{g^2}{32\pi^2}, F^a{\mu\nu},\tilde F^{a,\mu\nu}. $$

This term is a total derivative, with the Chern-Simons current, so it does not alter the classical field equations; it contributes to physics only through configurations of nonzero , the instantons. It is even under charge conjugation but odd under parity and time reversal , hence odd under . A nonzero therefore makes the strong interaction violate .

Counterexamples to common slips

  • The -term is not zero just because is a total derivative. The integral of a total derivative need not vanish when the boundary data carry nonzero winding; it equals , and counts the instanton number. A total derivative integrates to a boundary term, and here the boundary term is topological.
  • The physical angle is not alone. With massive quarks, a chiral rotation of the quark fields shifts through the axial anomaly, where is the quark mass matrix. The observable is . Setting the bare to zero is meaningless without specifying the quark-mass phase.
  • If any quark were massless, would be unobservable. A massless quark allows a chiral rotation that removes entirely at no cost, because the anomaly relates the rotation to a shift of and the massless quark has no mass term to obstruct it. The strong-CP problem exists precisely because all quark masses are nonzero.

Key derivation Intermediate+

Theorem (-vacua as Bloch waves; -dependence of the energy). The states of distinct winding number do not mix under any local gauge-invariant dynamics within a single superselection sector; the eigenstates of the large-gauge ladder operator are , with . In the dilute-instanton-gas approximation the vacuum energy density acquires the dependence , where is the one-instanton tunnelling amplitude per unit four-volume. The minimum is at , and the -dependence is physical and -periodic.

Proof. The Euclidean amplitude to propagate from a configuration of winding at to winding at is a path integral over gauge fields interpolating between them. By the integrality of the Pontryagin index, every such path has , so the amplitude depends on the winding numbers only through their difference: $$ \langle n_+ | e^{-H T_E} | n_- \rangle ;=; \mathcal{F}(n_+ - n_-). $$ Diagonalise this Toeplitz structure by Fourier transform on . Define ; then $$ \langle \theta' | e^{-H T_E} | \theta \rangle = \sum_{n_+, n_-} e^{-in_+\theta'} e^{in_-\theta},\mathcal{F}(n_+ - n_-) = 2\pi,\delta(\theta - \theta') \sum_\nu e^{i\nu\theta},\mathcal{F}(\nu). $$ The Kronecker-to-Dirac delta shows that amplitudes vanish: is conserved and superselected. Within a fixed the propagator is the number , and the weight on the -instanton sector is exactly the insertion of the -term into the path integral.

To extract the energy, evaluate in the dilute gas: a configuration of instantons and anti-instantons (net ) at well-separated locations contributes, in a four-volume , a Boltzmann-like factor from the collective-coordinate integration of 03.07.09, with the single-instanton amplitude including the functional determinant of 't Hooft 1976. Summing, $$ Z(\theta) = \sum_{\nu_+,\nu_- \ge 0} \frac{(KVT_E, e^{i\theta})^{\nu_+}}{\nu_+!},\frac{(KVT_E, e^{-i\theta})^{\nu_-}}{\nu_-!} = \exp!\big[KVT_E(e^{i\theta} + e^{-i\theta})\big] = \exp!\big[2KVT_E\cos\theta\big]. $$ Reading gives , minimised at , periodic with period . The dependence is physical: is a genuine, nonzero tunnelling rate set by the instanton action , so up to the determinant prefactor.

Bridge. This computation builds toward the strong-CP problem and appears again in the axion story below. The foundational reason the vacuum is a Bloch wave is that the winding sectors are connected by tunnelling exactly as the cells of a periodic crystal are connected, and this is the central insight that identifies the gauge-theory vacuum with the crystal-electron problem: the angle is dual to the crystal momentum, and the large-gauge operator is dual to the lattice-translation operator. Putting these together, the weighting on instanton sectors is exactly the Lagrangian -term, so the Hamiltonian Bloch-wave picture and the Euclidean path-integral picture are one phenomenon in two languages. The dilute-gas energy generalises to the full vacuum energy whose curvature at is the topological susceptibility , the quantity that the same instanton physics ties to the mass in the resolution of the problem.

Exercises Intermediate+

Advanced results Master

The dilute-gas energy is the leading term of a structure whose curvature at the origin, the topological susceptibility, ties the vacuum angle to the spectrum of light hadrons and underlies both the resolution and the strong-CP problem. The following results assemble the topological, anomalous, and phenomenological content.

Proposition (topological susceptibility and its sign). The vacuum energy density of a confining gauge theory is even, -periodic, and minimised at (Vafa-Witten), with curvature , the topological susceptibility. In pure Yang-Mills , the variance of the topological charge per unit four-volume.

The non-negativity and the location of the minimum at are the Vafa-Witten theorem: in a vector-like theory with positive Euclidean measure, and are not spontaneously broken, and the free energy is extremised at the -conserving point. The susceptibility is directly measurable on the lattice and enters the Witten-Veneziano formula below.

Proposition (Witten-Veneziano and the problem). The mass of the meson, the would-be ninth Goldstone boson of the spontaneously broken flavour symmetry, is fixed by the pure-glue topological susceptibility: $$ m_{\eta'}^2 + m_\eta^2 - 2m_K^2 ;=; \frac{2N_f}{f_\pi^2},\chi_t^{\text{(YM)}}, $$ where is the number of light flavours and the pion decay constant. The is heavy ( MeV, not light like the pion) precisely because instantons explicitly break the anomalous axial .

This is 't Hooft's resolution of the problem. Naively the chiral symmetry of QCD with light flavours is , whose spontaneous breaking would yield Goldstone bosons; experiment shows only light pseudoscalars and one heavy . The axial is not a symmetry of the quantum theory — its current is anomalous, — and instantons make the anomaly dynamically effective, lifting the mass. The same that defines the -term resolves the problem.

Theorem (instanton-induced effective vertex). In a background of one instanton, the fermionic functional integral has exact zero modes (one per light Weyl flavour of each chirality, by the Atiyah-Singer index theorem applied to the Dirac operator in the instanton background). The integral over these zero modes produces a -fermion effective vertex, the 't Hooft vertex, which violates the axial charge by units while conserving the non-anomalous vector and charges.

The index theorem of 03.07.32 gives the zero-mode count: for topological charge and a Dirac fermion in the fundamental, the index is , so a single instanton supports one left-handed zero mode per flavour. A nonzero amplitude requires saturating every zero mode with an external fermion line, which is why the instanton generates a vertex with one in-and-out leg per light flavour. For three light flavours this is a six-fermion operator, the microscopic origin of the breaking and a contributor to the mass.

Proposition (Peccei-Quinn resolution and the axion). If the theory possesses a spontaneously broken global that is itself anomalous under the colour group, the effective angle is promoted to a dynamical field (the axion), whose potential is the instanton-induced . The field relaxes to the minimum at , dynamically enforcing conservation. The axion is the pseudo-Goldstone boson of the broken , with mass , , and is a leading cold-dark-matter candidate.

Synthesis. The foundational reason the strong interaction has a free -violating parameter is that its vacuum is a Bloch wave over topological sectors, and this is the central insight that identifies the gauge-theory vacuum angle with the crystal momentum of a periodic system and the large-gauge operator with lattice translation. Putting these together, the single quantity does three jobs at once: integrated, it is the integer winding number of 03.07.07; weighted by , it is the superselection phase that builds and the -odd -term; and as the anomaly , it is dual to the axial-current non-conservation that lifts the and resolves the problem. This is exactly the topology that appears again in the chiral-anomaly index computation of 03.07.32, where the zero-mode count that gives the 't Hooft vertex is the same index that the descent equations compute. The strong-CP problem is then sharp: is physical, the neutron-dipole bound forces , and the Peccei-Quinn mechanism builds toward a dynamical that the instanton potential relaxes to zero, predicting the axion.

Full proof set Master

Proposition (topological susceptibility and its sign), proof. Write the Euclidean partition function in the -sector as with the (positive, by reflection positivity of the vector-like measure) partition function at fixed topological charge . Then and $$ \chi_t = \left.\frac{\partial^2\mathcal{E}}{\partial\theta^2}\right|{0} = \frac{1}{V}\Big(\langle\nu^2\rangle - \langle\nu\rangle^2\Big)\Big|{\theta=0} = \frac{\langle\nu^2\rangle}{V}, $$ using at by the symmetry of the -invariant measure (instanton/anti-instanton symmetry). A variance is non-negative, so . For the location of the minimum, the Vafa-Witten argument observes that since each ; hence , so is the global minimum and , are unbroken there.

Proposition (Witten-Veneziano), proof sketch with the controlling identity. Expand to second order, . In full QCD with light quarks, the -dependence is screened: a chiral rotation absorbs into the quark mass phases, so the physical curvature involves the quark masses and the couples to the topological charge density. The large- counting of Witten relates the pure-glue susceptibility (order ) to the anomalous part of the mass through the two-point function of the topological charge density . Saturating the correlator by the pole on the quark side and by the pure-glue susceptibility on the glue side gives $$ \chi_t^{\text{(YM)}} = \frac{f_\pi^2}{2N_f}\big(m_{\eta'}^2 + m_\eta^2 - 2m_K^2\big), $$ the Gell-Mann-Okubo-corrected combination isolating the anomaly-induced (singlet) mass. The empirical left side from the lattice, MeV, reproduces MeV, confirming that the mass is topological in origin. The full proof requires the large- expansion and is carried out in Witten 1979 and Veneziano 1979.

Theorem (instanton-induced effective vertex), proof. In a one-instanton background, the massless Dirac operator has, by the Atiyah-Singer index theorem applied to the twisted Dirac operator on the compactified of 03.07.08, index , where count zero modes of each chirality. For and a fundamental fermion there is exactly one left-handed zero mode and no right-handed one (a vanishing theorem rules out the opposite chirality in the self-dual background). The Grassmann integral vanishes unless every zero mode is saturated, because but for the Grassmann zero-mode coefficients. With flavours there are such modes (one per flavour), so the leading nonvanishing amplitude carries one factor of and one of per flavour: an effective operator $$ \mathcal{L}{\text{'t Hooft}} ;\sim; K, e^{i\theta} \prod{f=1}^{N_f} \big(\bar q_{f,L}, q_{f,R}\big) + \text{h.c.}, $$ which changes the axial charge by (each bilinear carries axial charge summed over chiralities normalised per flavour) while leaving the vector charge and the non-singlet axial charges intact. The prefactor exhibits the -dependence; the vertex is the microscopic source of violation.

Proposition (Peccei-Quinn resolution), proof. Promote to a field by coupling an axion through , as forced by the anomaly of a spontaneously broken at scale . The instanton-induced potential for the combination is, from the dilute-gas result with light-quark dressing, , with by the proposition above. The stationary condition has its stable solution at , since there. The vacuum dynamically sets the effective angle to zero, removing strong violation. Expanding about the minimum, the axion mass-squared is ; inserting the QCD value of in terms of gives . The axion is the pseudo-Goldstone boson whose potential is the instanton vacuum energy itself.

Connections Master

  • BPST instanton and the Bogomolny bound 03.07.07. The integer that labels the winding sectors is exactly the Pontryagin index of 03.07.07, and the BPST instanton is the minimal-action Euclidean path that carries the field from sector to sector . The single tunnelling amplitude in the -vacuum energy is governed by the instanton action that the Bogomolny bound of 03.07.07 saturates. The -vacuum is the vacuum-structure physics built on the instanton geometry of 03.07.07.

  • Conformal compactification and finite-action instantons 03.07.08. The integrality of the topological charge, and the index theorem giving the fermionic zero-mode count behind the 't Hooft vertex, both rely on the compactification of Euclidean to of 03.07.08: finite action forces pure-gauge behaviour at infinity, the boundary , and the winding of that boundary map is the integer of . Without the finite-action compactification of 03.07.08 there would be no discrete sectors to label with and no .

  • Anomalies via descent equations and the index theorem 03.07.32. The axial anomaly that makes non-conserved is the same that defines the -term, and the index theorem of 03.07.32 that the descent equations encode is exactly the count of instanton zero modes producing the 't Hooft vertex. The resolution of the problem and the existence of the -term are two faces of the single anomaly computed in 03.07.32; the chiral rotation that shifts is the anomalous transformation of 03.07.32.

  • The Higgs mechanism: spontaneously broken gauge symmetry 12.18.01. Both units concern the vacuum structure of a non-Abelian gauge theory, but in complementary regimes: 12.18.01 treats the weakly coupled, spontaneously broken (Higgs) phase where perturbation theory about a single vacuum suffices, whereas this unit treats the strongly coupled confining phase where the vacuum is an irreducibly non-perturbative superposition over topological sectors. The Peccei-Quinn axion is itself a (pseudo-)Goldstone boson of a broken global symmetry, connecting the symmetry-breaking machinery of 12.18.01 to the strong-CP resolution.

Historical & philosophical context Master

The topological vacuum was discovered in 1976 by two groups working independently and almost simultaneously. Jackiw and Rebbi (Phys. Rev. Lett. 37, 172, 1976) [source pending] and Callan, Dashen, and Gross (Phys. Lett. B 63, 334, 1976) recognised that the gauge-theory vacuum is periodic in a topological winding coordinate and that the physical vacua are the Bloch-wave superpositions , one for each value of a continuous angle that labels a superselection sector. The instanton — the finite-action Euclidean solution of Belavin, Polyakov, Schwartz, and Tyupkin (1975), anchored in 03.07.07 — was identified as the tunnelling path between adjacent winding sectors. 't Hooft's detailed functional-determinant computation (Phys. Rev. D 14, 3432, 1976; with an erratum in Phys. Rev. D 18, 2199, 1978) [source pending] supplied the tunnelling amplitude and, through the fermionic zero modes, the effective multi-fermion vertex that resolved the long-standing problem: the ninth would-be Goldstone boson, the , is heavy because the axial singlet symmetry is anomalous and instantons make the anomaly dynamically effective.

The price of this resolution was a new puzzle. The -term is -violating, and the physical angle is in principle order one, yet the neutron electric dipole moment bound, currently (Abel et al., Phys. Rev. Lett. 124, 081803, 2020), forces . Peccei and Quinn (Phys. Rev. Lett. 38, 1440, 1977) proposed a dynamical resolution: an additional anomalous global symmetry whose spontaneous breaking promotes to a field that relaxes to zero. Weinberg (Phys. Rev. Lett. 40, 223, 1978) and Wilczek (Phys. Rev. Lett. 40, 279, 1978) immediately observed that this symmetry breaking implies a light pseudo-Goldstone boson, the axion, which remains an actively searched-for dark-matter candidate.

Bibliography Master

@article{JackiwRebbi1976,
  author  = {Jackiw, Roman and Rebbi, Claudio},
  title   = {Vacuum Periodicity in a Yang-Mills Quantum Theory},
  journal = {Phys. Rev. Lett.},
  volume  = {37},
  year    = {1976},
  pages   = {172--175}
}

@article{CallanDashenGross1976,
  author  = {Callan, Curtis G. and Dashen, Roger F. and Gross, David J.},
  title   = {The Structure of the Gauge Theory Vacuum},
  journal = {Phys. Lett. B},
  volume  = {63},
  year    = {1976},
  pages   = {334--340}
}

@article{tHooft1976PRD,
  author  = {'t Hooft, Gerard},
  title   = {Computation of the Quantum Effects Due to a Four-Dimensional Pseudoparticle},
  journal = {Phys. Rev. D},
  volume  = {14},
  year    = {1976},
  pages   = {3432--3450},
  note    = {Erratum: Phys. Rev. D 18, 2199 (1978)}
}

@article{tHooft1976PRL,
  author  = {'t Hooft, Gerard},
  title   = {Symmetry Breaking through Bell-Jackiw Anomalies},
  journal = {Phys. Rev. Lett.},
  volume  = {37},
  year    = {1976},
  pages   = {8--11}
}

@article{PecceiQuinn1977PRL,
  author  = {Peccei, Roberto D. and Quinn, Helen R.},
  title   = {CP Conservation in the Presence of Pseudoparticles},
  journal = {Phys. Rev. Lett.},
  volume  = {38},
  year    = {1977},
  pages   = {1440--1443}
}

@article{Weinberg1978Axion,
  author  = {Weinberg, Steven},
  title   = {A New Light Boson?},
  journal = {Phys. Rev. Lett.},
  volume  = {40},
  year    = {1978},
  pages   = {223--226}
}

@article{Wilczek1978Axion,
  author  = {Wilczek, Frank},
  title   = {Problem of Strong $P$ and $T$ Invariance in the Presence of Instantons},
  journal = {Phys. Rev. Lett.},
  volume  = {40},
  year    = {1978},
  pages   = {279--282}
}

@article{Witten1979,
  author  = {Witten, Edward},
  title   = {Current Algebra Theorems for the $U(1)$ Goldstone Boson},
  journal = {Nucl. Phys. B},
  volume  = {156},
  year    = {1979},
  pages   = {269--283}
}

@article{Veneziano1979,
  author  = {Veneziano, Gabriele},
  title   = {$U(1)$ without Instantons},
  journal = {Nucl. Phys. B},
  volume  = {159},
  year    = {1979},
  pages   = {213--224}
}

@article{VafaWitten1984,
  author  = {Vafa, Cumrun and Witten, Edward},
  title   = {Parity Conservation in Quantum Chromodynamics},
  journal = {Phys. Rev. Lett.},
  volume  = {53},
  year    = {1984},
  pages   = {535--536}
}

@article{Abel2020nEDM,
  author  = {Abel, C. and others},
  title   = {Measurement of the Permanent Electric Dipole Moment of the Neutron},
  journal = {Phys. Rev. Lett.},
  volume  = {124},
  year    = {2020},
  pages   = {081803}
}

@book{ColemanAspects,
  author    = {Coleman, Sidney},
  title     = {Aspects of Symmetry: Selected Erice Lectures},
  publisher = {Cambridge University Press},
  year      = {1985}
}

@book{WeinbergQTFVol2,
  author    = {Weinberg, Steven},
  title     = {The Quantum Theory of Fields, Vol. 2: Modern Applications},
  publisher = {Cambridge University Press},
  year      = {1996}
}