Quantum Yang–Mills Theory
Arthur Jaffe and Edward Witten
1. The Physics of Gauge Theory
Since the early part of the twentieth century, it has been understood that the
description of nature at the subatomic scale requires quantum mechanics. In quan-
tum mechanics, the position and velocity of a particle are noncommuting operators
acting on a Hilbert space, and classical notions such as “the trajectory of a particle”
do not apply.
But quantum mechanics of particles is not the whole story. In nineteenth and
early twentieth century physics, many aspects of nature were described in terms
of fields—the electric and magnetic fields that enter in Maxwell’s equations, and
the gravitational field governed by Einstein’s equations. Since fields interact with
particles, it became clear by the late 1920’s that an internally coherent account of
nature must incorporate quantum concepts for fields as well as for particles.
After doing this, quantities such as the components of the electric field at dif-
ferent points in space-time become non-commuting operators. When one attempts
to construct a Hilbert space on which these operators act, one finds many surprises.
The distinction between fields and particles breaks down, since the Hilbert space
of a quantum field is constructed in terms of particle-like excitations. Conventional
particles, such as electrons, are reinterpreted as states of the quantized field. In
the process, one finds the prediction of “antimatter;” for every particle, there must
be a corresponding antiparticle, with the same mass and opposite electric charge.
Soon after P. A. M. Dirac predicted this on the basis of quantum field theory,
the “positron” or oppositely charged antiparticle of the electron was discovered in
cosmic rays.
The most important Quantum Field Theories (QFT’s) for describing elemen-
tary particle physics are gauge theories. The classical example of a gauge theory is
Maxwell’s theory of electromagnetism. For electromagnetism the gauge symmetry
group is the abelian group U(1). If A denotes the U(1) gauge connection, locally
a one-form on space-time, then the curvature or electromagnetic field tensor is the
two-form F = dA, and Maxwell’s equations in the absence of charges and currents
read 0 = dF = d?F. Here ? denotes the Hodge duality operator; indeed Hodge
introduced his celebrated theory of harmonic forms as a generalization of the solu-
tions to Maxwell’s equations. Maxwell’s equations describe large-scale electric and
magnetic fields and also—as Maxwell discovered—the propagation of light waves,
at a characteristic velocity, the speed of light.
1
2 ARTHUR JAFFE AND EDWARD WITTEN
The idea of a gauge theory evolved from the work of Hermann Weyl. One
can find in [34] an interesting discussion of the history of gauge symmetry and
the discovery of Yang-Mills theory [50], also known as “non-abelian gauge theory.”
At the classical level one replaces the gauge group U(1) of electromagnetism by a
compact gauge groupG. The definition of the curvature arising from the connection
must be modified to F = dA+A∧A, and Maxwell’s equations are replaced by the
Yang-Mills equations, 0 = dAF = dA?F, wheredA is the gauge-covariant extension
of the exterior derivative.
These classical equations can be derived as variational equations from the Yang-
Mills Lagrangian
(1) L = 14g2
integraldisplay
Tr F ∧?F,
where Tr denotes an invariant quadratic form on the Lie algebra of G. The Yang-
Mills equations are nonlinear—in contrast to the Maxwell equations. Like the
Einstein equations for the gravitational field, only a few exact solutions of the
classical equation are known. But the Yang-Mills equations have certain properties
in common with the Maxwell equations: in particular they provide the classical
description of massless waves that travel at the speed of light.
By the 1950’s, when Yang-Mills theory was discovered, it was already known
that the quantum version of Maxwell theory—known as Quantum Electrodynamics
or QED—gives an extremely accurate account of electromagnetic fields and forces.
In fact QED improved the accuracy for certain earlier quantum theory predictions
by several orders of magnitude, as well as predicting new splittings of energy levels.
So it was natural to inquire whether non-abelian gauge theory described other
forces in nature, notably the weak force (responsible among other things for certain
forms of radioactivity) and the strong or nuclear force (responsible among other
things for the binding of protons and neutrons into nuclei). The massless nature
of classical Yang-Mills waves was a serious obstacle to applying Yang-Mills theory
to the other forces, for the weak and nuclear forces are short range and many of
the particles are massive. Hence these phenomena did not appear to be associated
with long range fields describing massless particles.
In the 1960’s and 1970’s, physicists overcame these obstacles to the physical
interpretation of nonabelian gauge theory. In the case of the weak force, this was
accomplished by the Glashow-Salam-Weinberg electroweak theory [47, 40] with
gauge group H = SU(2) ×U(1). By elaborating the theory with an additional
“Higgs field,” one avoided the massless nature of classical Yang-Mills waves. The
Higgs field transforms in a two-dimensional representation of H; its non-zero and
approximately constant value in the vacuum state reduces the structure group from
H to a U(1) subgroup (diagonally embedded in SU(2) ×U(1)). This theory de-
scribes both the electromagnetic and weak forces, in a more or less unified way;
because of the reduction of the structure group to U(1), the long range fields are
those of electromagnetism only, in accord with what we see in nature.
The solution to the problem of massless Yang-Mills fields for the strong inter-
actions has a completely different nature. That solution did not come from adding
additional fields to Yang-Mills theory, but by discovering a remarkable property of
the quantum Yang-Mills theory itself, that is, of the quantum theory whose classi-
cal Lagrangian has been given in (1). This property is called “asymptotic freedom”
[21, 38]. Roughly this means that at short distances the field displays quantum
QUANTUM YANG–MILLS THEORY 3
behavior very similar to its classical behavior; yet at long distances the classical
theory is no longer a good guide to the quantum behavior of the field.
Asymptotic freedom, together with other experimental and theoretical discov-
eries made in the 1960’s and 70’s, made it possible to describe the nuclear force
by a non-abelian gauge theory in which the gauge group is G = SU(3). The ad-
ditional fields describe, at the classical level, “quarks,” which are spin 1/2 objects
somewhat analogous to the electron, but transforming in the fundamental repre-
sentation of SU(3). The non-abelian gauge theory of the strong force is called
Quantum Chromodynamics (QCD).
The use of QCD to describe the strong force was motivated by a whole series
of experimental and theoretical discoveries made in the 1960’s and 70’s, involving
the symmetries and high energy behavior of the strong interactions. But classical
nonabelian gauge theory is very different from the observed world of strong in-
teractions; for QCD to describe the strong force successfully, it must have at the
quantum level the following three properties, each of which is dramatically different
from the behavior of the classical theory:
(1) It must have a “mass gap;” namely there must be some constant ? > 0
such that every excitation of the vacuum has energy at least ?.
(2) It must have “quark confinement,” that is, even though the theory is
described in terms of elementary fields, such as the quark fields, that
transform non-trivially under SU(3), the physical particle states—such as
the proton, neutron, and pion—are SU(3)-invariant.
(3) It must have “chiral symmetry breaking,” which means that the vacuum
is potentially invariant (in the limit that the quark bare masses vanish)
only under a certain subgroup of the full symmetry group that acts on
the quark fields.
The first point is necessary to explain why the nuclear force is strong but short-
ranged; the second is needed to explain why we never see individual quarks; and
the third is needed to account for the “current algebra” theory of soft pions that
was developed in the 1960’s.
Both experiment—since QCD has numerous successes in confrontation with
experiment—and computer simulations, see for example [8], carried out since the
late 1970’s have given strong encouragement that QCD does have the properties
citedabove. These properties canbe seen, tosomeextent, intheoreticalcalculations
carried out in a variety of highly oversimplified models (like strongly coupled lattice
gauge theory, see for example [48]). But they are not fully understood theoretically;
there does not exist a convincing, even if not mathematically complete, theoretical
computation demonstrating any of the three properties in QCD, as opposed to a
severely simplified truncation of it.
2. Quest For Mathematical Understanding
In surveying the physics of gauge theories in the last section, we considered both
classical properties—such as the Higgs mechanism for the electroweak theory—
and quantum properties that do not have classical analogs—like the mass gap and
confinement for QCD. Classical properties of gauge theory are within the reach of
established mathematical methods, and indeed classical non-abelian gauge theory
has played a very important role in mathematics in the last twenty years, especially
in the study of three- and four-dimensional manifolds. On the other hand, one does
4 ARTHUR JAFFE AND EDWARD WITTEN
not yet have a mathematically complete example of a quantum gauge theory in
four-dimensional space-time, nor even a precise definition of quantum gauge theory
in four dimensions. Will this change in the twenty-first century? We hope so!
At times, mathematical structures of importance have first appeared in physics
before their mathematical importance was fully recognized. This happened with the
discovery of calculus, which was needed to develop Newtonian mechanics, with func-
tional analysis and group representation theory, topics whose importance became
clearer with quantum mechanics, and even with the study of Riemannian geometry,
whose development was greatly intensified once it became clear, through Einstein’s
invention of General Relativity to describe gravity, that this subject plays a role in
the description of nature. These areas of mathematics became generally accessible
only after a considerable time, over which the ideas were digested, simplified, and
integrated into the general mathematical culture.
Quantum Field Theory (QFT) became increasingly central in physics through-
out the twentieth century. There are reasons to believe that it may be important in
twenty-first century mathematics. Indeed, many mathematical subjects that have
been actively studied in the last few decades appear to have natural formulations—
at least at a heuristic level—in terms of QFT. New structures spanning probability,
analysis, algebra, and geometry have emerged, for which a general mathematical
framework still is in its infancy.
On the analytic side, a byproduct of the existence proofs and mathematical
construction of certain quantum field theories was the construction of new sorts
of measures, in particular non-gaussian, Euclidean-invariant measures on spaces of
generalized functionals. Dirac fields and gauge fields require measures on spaces
of functions taking values in a Grassmann algebra and on spaces of functions into
other target geometries.
Renormalization theory arises from the physics of quantum field theory and pro-
vides the basis for the mathematical investigation of local singularities (ultra-violet
regularity) and of global decay (infra-red regularity) in quantum field theories. As-
ymptotic freedom ensures the decisive regularity in the case when classical Sobolev
inequalities are borderline. Surprisingly the ideas from renormalization theory also
apply in other areas of mathematics, including in classic work on the convergence
of Fourier series, and in recent progress on classical dynamical systems.
On the algebraic side, investigations of soluble models of quantum field theory
and statistical mechanics have led to many new discoveries involving topics such as
Yang-Baxter equations, quantum groups, bose-fermi equivalence in two dimensions,
and rational conformal field theory.
Geometry abounds with new mathematical structures rooted in quantum field
theory, many of them very actively studied in the last twenty years. Examples
include Donaldson theory of four-manifolds, the Jones polynomial of knots and its
generalizations, mirror symmetry of complex manifolds, elliptic cohomology, and
SL(2,Z) symmetry in the theory of affine Kac-Moody algebras.
QFT has in certain cases suggested new perspectives on mathematical prob-
lems, while in other cases its mathematical value up to the present time is mo-
tivational. In the case of the geometric examples cited above, a mathematical
definition of the relevant QFT’s (or one in which the relevant physical techniques
can be justified) is not yet at hand. Existence theorems that put QFT’s on a solid
mathematical footing are needed to make the geometrical applications of QFT into
QUANTUM YANG–MILLS THEORY 5
a full-fledged part of mathematics. Regardless of the future role of QFT in pure
mathematics, it is a great challenge for mathematicians to understand the physical
principles that have been so important and productive throughout the twentieth
century.
Finally, QFT is the jumping off point for a quest that may prove central in
twenty-first century physics—the effort to unify gravity and quantum mechanics,
perhaps in string theory. For mathematicians to participate in this quest, or even
to understand the possible results, QFT must be developed further as a branch of
mathematics. It is important not only to understand the solution of specific prob-
lems arising from physics, but also to set such results within a new mathematical
framework. One hopes that this framework will provide a unified development of
several fields of mathematics and physics, and that it will also provide an arena for
the development of new mathematics and new physics.
For these reasons the Scientific Advisory Board of CMI has chosen a Millennium
problem about quantum gauge theories. Solution of the problem requires both
understanding one of the deep unsolved physics mysteries, the existence of a mass
gap, and also producing a mathematically complete example of quantum gauge
field theory in four dimensional space-time.
3. Quantum Fields
A quantum field, or local quantum field operator, is an operator-valued gen-
eralized function on spacetime obeying certain axioms. The properties required of
the quantum fields are described at a physical level of precision in many textbooks,
see for example [27]. G?arding and Wightman gave mathematically precise axioms
for quantum field theories on R4 with a Minkowski signature, see [45], and Haag
and Kastler introduced a related scheme for local functions of the field, see [24].
Basically one requires that the Hilbert space H of the quantum field carry
a representation of the Poincar′e group (or inhomogeneous Lorentz group). The
Hamiltonian H and momentum vectorP are the self-adjoint elements of the Lie algebra
of the group that generate translations in time and space. A vacuum vector is an
element ofHthat is invariant under the (representation of the) Poincar′e group. One
assumes that the representation has positive energy, 0 ≤ H, and a vacuum vector
? ∈ H that is unique up to a phase. Gauge-invariant functions of the quantum
fields also act as linear transformations on H and transform covariantly under the
Poincar′e group. Quantum fields in space-time regions that cannot be connected by a
light signal should be independent; G?arding and Wightman formulate indepencence
as the commuting of the field operators (anti-commuting for two fermionic fields).
One of the achievements of twentieth century axiomatic quantum field the-
ory was the discovery of how to convert a Euclidean-invariant field theory on a
Euclidean space-time to a Lorentz-invariant field theory on Minkowski space-time,
and vice-versa. Wightman used positive energy to establish analytic continuation of
expectations of Minkowski field theories to Euclidean space. Kurt Symanzik inter-
preted the Euclidean expectations as a statistical mechanical ensemble of classical
Markov fields [46], with a probability density proportional to exp(?S), whereS de-
notes the Euclidean action functional. E. Nelson reformulated Symanzik’s picture
and showed that one can also construct a Hilbert space and a quantum-mechanical
field from a Markov field [33]. Osterwalder and Schrader then discovered the ele-
mentary “reflection-positivity” condition to replace the Markov property. This gave
6 ARTHUR JAFFE AND EDWARD WITTEN
rise to a general theory establishing equivalence between Lorentzian and Euclidean
axiom schemes [35]. See also [13].
One hopes that the continued mathematical exploration of quantum field theory
will lead to refinements of the axiom sets that have been in use up to now, perhaps
to incorporate properties considered important by physicists such as the existence
of an operator product expansion or of a local stress-energy tensor.
4. The Problem
To establish existence of four-dimensional quantum gauge theory with gauge
group G, one should define a quantum field theory (in the above sense) with local
quantum field operators in correspondence with the gauge-invariant local polyno-
mials in the curvature F and its covariant derivatives, such as TrFijFkl(x).1 Cor-
relation functions of the quantum field operators should agree at short distances
with the predictions of asymptotic freedom and perturbative renormalization the-
ory, as described in textbooks. Those predictions include among other things the
existence of a stress tensor and an operator product expansion, having prescribed
local singularities predicted by asymptotic freedom.
Since the vacuum vector ? is Poincar′e invariant, it is an eigenstate with zero
energy, namely H? = 0. The positive energy axiom asserts that in any quantum
field theory, the spectrum of H is supported in the region [0,∞). A quantum field
theory has a mass gap ? ifH has no spectrum in the interval (0,?) for some ? > 0.
The supremum of such ? is the mass m, and we require m<∞.
Yang–Mills Existence and Mass Gap. Prove that for any compact simple
gauge group G, a non-trivial quantum Yang–Mills theory exists on R4 and has a
mass gap ? > 0. Existence includes establishing axiomatic properties at least as
strong as those cited in [45, 35].
5. Comments
An important consequence of the existence of a mass gap is clustering: let
vectorx∈R3 denote a point in space. We letH and vectorP denote the energy and momentum,
generators of time and space translation. For any positive constant C < ? and for
any local quantum field operator O(vectorx) = e?ivectorP·vectorxOeivectorP·vectorx such that 〈?,O?〉 = 0, one
has
(2) |〈?,O(vectorx)O(vectory)?〉|≤ exp(?C|vectorx?vectory|),
as long as |vectorx?vectory| is sufficiently large. Clustering is a locality property that, roughly
speaking, may make it possible to apply mathematical results established on R4 to
any four-manifold, as argued at a heuristic level (for a supersymmetric extension of
four-dimensional gauge theory) in [49]. Thus the mass gap not only has a physical
significance (as explained in the introduction), but it may also be important in
mathematical applications of four-dimensional quantum gauge theories to geometry.
In addition the existence of a uniform gap for finite-volume approximations may
play a fundamental role in the proof of existence of the infinite-volume limit.
1A natural 1?1 correspondence between such classical ‘differential polynomials’ and quan-
tized operators does not exist, since the correspondence has some standard subtleties involving
renormalization [27]. One expects that the space of classical differential polynomials of dimension
≤d does correspond to the space of local quantum operators of dimension ≤d.
QUANTUM YANG–MILLS THEORY 7
There are many natural extensions of the Millennium Problem. Among other
things, one would like to prove the existence of an isolated one-particle state (an
upper gap, in addition to the mass gap), to prove confinement, to prove existence of
other four-dimensional gauge theories (incorporating additional fields that preserve
asymptotic freedom), to understand dynamical questions (such as the possible mass
gap, confinement, and chiral symmetry breaking) in these more general theories,
and to extend the existence theorems from R4 to an arbitrary four-manifold.
But a solution of the existence and mass gap problem as stated above would be
a turning point in the mathematical understanding of quantum field theory, with
a chance of opening new horizons for its applications.
6. Mathematical Perspective
Wightman and others have questioned for approximately fifty years whether
mathematically well-defined examples of relativistic, non-linear quantum field the-
ories exist. We now have a partial answer: extensive results on the existence and
physical properties of non-linear QFT’s have been proved through the emergence
of the body of work known as “constructive quantum field theory” (CQFT).
The answers are partial, for in most of these field theories one replaces the
Minkowski space-time M4 by a lower-dimensional space-time M2 or M3, or by a
compact approximation such as a torus. (Equivalently in the Euclidean formulation
one replaces Euclidean space-time R4 by R2 or R3.) Some results are known for
Yang-Mills theory on a four-torus T4 approximating R4, and while the construction
is not complete, there is ample indication that known methods could be extended
to construct Yang-Mills theory on T4.
In fact, at present one does not know any non-trivial relativistic field theory
that satisfies the Wightman (or any other reasonable) axioms in four-dimensions.
So even having a detailed mathematical construction of Yang-Mills theory on a
compact space would represent a major breakthrough. Yet, even if this were ac-
complished, no present ideas point the direction to establish the existence of a mass
gap that is uniform in the volume. Nor do present methods suggest how to obtain
the existence of the infinite volume limit T4 →R4.
6.1. Methods. Since the inception of quantum field theory, two central meth-
ods have emerged to show the existence of quantum fields on non-compact config-
uration space (such as Minkowski space). These known methods are:
(i) Find an exact solution in closed form.
(ii) Solve a sequence of approximate problems, and establish convergence of
these solutions to the desired limit.
Exact solutions are possible only in very special cases, often for linear quantum
fields. But it may also apply to non-linear fields for special values of the coupling
constants yielding extra symmetries or exactly integrable models. It can sometimes
be used after a clever change of variables. Even if one could find an exact solution to
a quantum Yang-Mills theory, getting control over the detailed properties of that
field theory (such as establishing reflection positivity or Wightman-like axioms)
could present substantial difficulties.
The second method is to use mathematical approximations to show the con-
vergence of approximate solutions to exact solutions of the non-linear problems,
8 ARTHUR JAFFE AND EDWARD WITTEN
known as constructive quantum field theory, or CQFT. Two principle approaches—
studyingquantumtheoryonHilbertspace, andstudyingclassicalfunctionalintegrals—
are related by the Osterwalder-Schrader construction. Establishing uniform a priori
estimates is central to CQFT, and three schemes for establishing estimates are
(a) Correlation inequalities.
(b) Symmetries of the interaction.
(c) Convergent expansions.
The correlation inequality methods have an advantage: they are general. But
correlation inequalities rely on special properties of the interaction that often apply
only for scalar bosons or abelian gauge theories. The use of symmetry also applies
only to special values of the couplings and is generally combined with another
method to obtain analytic control. In most known examples perturbation series,
i.e. power series in the coupling constant, are divergent expansions; even Borel and
other resummation methods have limited applicability.
This led one to develop expansion methods, generally known as cluster expan-
sions. Each term in a cluster expansion sum depends on the coupling constants in a
complicated fashion; they often arise as functional integrals. One requires sufficient
quantitative knowledge of the properties of each term in an expansion to ensure
the convergence of the sum and to establish its qualitative properties. Refined
estimates yield the rate of exponential decay of Green’s functions, magnitude of
masses, the existence of symmetry breaking (or its preservation), etc.
Over the past thirty years, a panoply of expansion methods have emerged as a
central tool for establishing mathematial results in CQFT. In their various incarna-
tions, these expansions encapsulate ideas of the asymptotic nature of perturbation
theory, of space-time localization, of phase-space localization, of renormalization
theory, of semi-classical approximations (including “non-perturbative” effects), and
of symmetry breaking. One can find an introduction to many of these methods and
references in [18], and a more recent review of results in [28]. These expansion
methods can be complicated and the literature is huge, so we can only hope to in-
troduce the reader to a few ideas; we apologize in advance for important omissions.
6.2. The First Examples: Scalar Fields. Since the 1940’s the quantum
Klein-Gordon field ? provided an example of a linear, scalar, mass-m field theory
(arising from a quadratic potential). This field, and the related free spinor Dirac
field, served as models for formulating the first axiom schemes in the 1950’s [45].
Moments of a Euclidean-invariant, reflection-positive, ergodic, Borel measure
dμ on the space Sprime parenleftbigRdparenrightbig of tempered distributions may satisfy the Osterwalder-
Schrader axioms. The Gaussian measure dμ with mean zero and covariance C =
(?? +m20)?1 yields the free, mass-m0 field; but one requires non-Gaussian dμ to
obtain non-linear fields. (For the Gaussian measure, reflection positivity is equiva-
lent to positivity of the transformation ΘC, restricted to L2(Rd+) ? L2(Rd). Here
Θ : t → ?t denotes the time-reflection operator, and Rd+ = {(t,vectorx) : t ≥ 0} is the
positive-time subspace.)
The first proof that non-linear fields satisfy the Wightman axioms and the
first construction of such non-Gaussian measures only emerged in the 1970’s. The
initial examples comprised fields with small, polynomial non-linearities on R2: first
in finite volume, and then in the infinite volume limit [15, 19, 22]. These field
theories obey the Wightman axioms (as well as all other axiomatic formulations),
QUANTUM YANG–MILLS THEORY 9
the fields describe particles of a definite mass, and the fields produce multi-particle
states with non-trivial scattering [19]. The scattering matrix can be expanded as
an asymptotic series in the coupling constants, and the results agree term-by-term
with the standard description of scattering in perturbation theory that one finds in
physics texts [37].
A quartic Wightman QFT on R3 also exists, obtained by constructing a re-
markable non-Gaussian measure dμ on Sprime(R3) [16, 10]. This merits further study.
We now focus on some properties of the simplest perturbation to the action-
density of the free field, namely the even quartic polynomial
(3) λ?4 + 12σ?2 +c.
The coefficients 0 < λ and σ,c ∈ R are real parameters, all zero for the free field.
For 0 <λlessmuch 1, one can choose σ(λ),c(λ) so the field theory described by (3) exists,
is unique, and has a mass equal m such that |m?m0| is small.
Because of the local singularity of the non-linear field, one must first cut-off
the interaction. The simplest method is to truncate the Fourier expansion of the
field ? in (3) to ?κ(x) = integraltext|k|≤κ ??(k)e?ikxdk, and to compactify the spatial volume
of the perturbation to V. One obtains the desired quantum field theory as a limit
of the truncated approximations. The constants σ,c have the form σ = αλ+βλ2
and c = γλ+δλ2 +epsilon1λ3. One desires that the expectations of products of fields
have a limit as κ → ∞. One chooses α,γ (in dimension two), and one chooses
all the coefficients α,β,γ,δ,epsilon1 (in dimension three), to depend on κ in the way
that perturbation theory suggests. One then proves that the expectations converge
as κ → ∞, even though the specified constants α,..., diverge. These constants
provide the required renormalization of the interaction. In the three-dimensional
case one also needs to normalize vectors in the Fock space a constant that diverges
with κ; one calls this constant a wave-function renormalization constant.
The “mass” operator in natural units is M =
radicalbig
H2 ? vectorP2 ≥ 0, and the vacuum
vector ? is a null vector, M? = 0. Massive single particle states are eigenvectors of
an eigenvalue m> 0. If m is an isolated eigenvalue of M, then one infers from the
Wightman axioms and Haag-Ruelle scattering theory that asymptotic scattering
states of an arbitrary number of particles exist, see [24, 18].
The fundamental problem of asymptotic completeness is the question whether
these asymptotic states (including possible bound states) span H. Over the past
thirty years, several new methods emerged yielding proofs of asymptotic complete-
ness in scattering theory for non-relativistic quantum mechanics. This gives some
hope that one can now attack the open problem of asymptotic completeness for
any known example of non-linear quantum field theory.
In contrast to the existence of quantum fields with a ?4 non-linearity in di-
mension two and three, the question of extending these results to four dimensions
is problematic. It is known that self-interacting scalar fields with a quartic non-
linearity do not exist in dimension five or more [12, 1]. (The proofs apply to
field theories with a single, scalar field.) Analysis of the borderline dimension four
(between existence and non-existence) is more subtle; if one makes some reason-
able (but not entirely proved) assumptions, one also can conclude triviality for the
quartic coupling in four dimensions. Not only is this persuasive evidence, but fur-
thermore the quartic coupling does not have the property of asymptotic freedom
in four dimensions. Thus all insights from random walks, perturbation theory, and
10 ARTHUR JAFFE AND EDWARD WITTEN
renormalization analysis point toward triviality of the quartic interaction in four
dimensions.
Other polynomial interactions in four dimensions are even more troublesome:
the classical energy of the cubic interaction is unbounded from below, so it appears
an unlikely candidate for a quantum theory where positivity of the energy is an ax-
iom. And polynomial interactions of degree greater than quartic are more singular
in perturbation theory than the quartic interaction.
All these reasons complement the physical and geometric importance of Yang-
Mills theory and highlight it as the natural candidate for four-dimensional CQFT.
6.3. Large Coupling Constant. In two dimensions the field theory with
energy density (3) exists for all positive λ. For 0 ≤ λ lessmuch 1 the solution is unique
under a variety of conditions; but for λ greatermuch 1 two different solutions exist, each
characterized by its ground state or “phase.” The solution in each phase satisfies
the Osterwalder-Schrader and Wightman axioms with a non-zero mass gap and
a unique, Poincar′e-invariant vacuum state. The distinct solutions appear as a
bifurcation of a unique approximating solution with finite volume V as V →∞.
One exhibits this behavior by reordering and scaling the λ?4 interaction (3)
with λgreatermuch 1 to obtain an equivalent double-well potential of the form
(4) λ
parenleftbigg
?2 ? 1λ
parenrightbigg2
+ 12σ?2 +c.
Here λ lessmuch 1 is a new coupling constant and the renormalization constants σ and
c are somewhat different from those above. The two solutions for a given λ are
related by the broken ?→?? symmetry of the interaction (4). The proof of these
facts relies upon developing a convergent cluster expansion about each minimum
of the potential arising from (4) and proving the probability of tunneling between
the two solutions is small [20].
A critical value λc of λ in (3) provides a boundary between the uniqueness of
the solution (for λ < λc) and the existence of a phase transition λ > λc. As λ
increases to λc, the mass gap m = m(λ) decreases monotonically and continuously
to zero [23, 17, 32].
The detailed behavior of the field theory (or the mass) in the neighborhood of
λ = λc is extraordinarily difficult to analyze; practically nothing has been proved.
Physicists have a qualitative picture based on the assumed fractional power-law
behavior m(λ) ~ |λc ?λ|ν above or below the critical point, where the exponent
ν depends on the dimension. One also expects that the critical coupling λc cor-
responds to the largest physical force between particles, and that these critical
theories are close to scaling limits of Ising-type modes in statistical physics. One
expects that further understanding of these ideas will result in new computational
tools for quantum fields and for statistical physics.
There is some partial understanding of a more general multi-phase case. One
can find an arbitrary number n of phases by making a good choice of a polynomial
energy density Pn(?) with n minima. It is interesting to study the perturbation
of a fixed such polynomial Pn by polynomials Q of lower degree and with small
coefficients. Among these perturbations one can find families of polynomials Q(?)
that yield field theories with exactly nprime ≤n phases [26].
6.4. Yukawa Interactions and Abelian Gauge Theory. The existence
of boson-fermion interactions is also known in two dimensions, and partial results
QUANTUM YANG–MILLS THEORY 11
exist in three dimensions. In two dimensions Yukawa interactions of the form ψψ?
exist with appropriate renormalization, as well as their generalizations of the form
P(?)+ψψQprimeprime(?), see [42, 18]. The supersymmetric case P = |Qprime|2 requires extra
care in dealing with cancellations of divergences, see [28] for references.
A continuum two-dimensional Higgs model describes an abelian gauge field in-
teracting with a charged scalar field. Brydges, Fr¨ohlich, and Seiler constructed this
theory and showed that it satisfies the Osterwalder-Schrader axioms [7], providing
the only complete example of an interacting gauge theory satisfying the axioms. A
mass gap exists in this model [4]. Extending all these conclusions to a non-abelian
Higgs model, even in two dimensions, would represent a qualitative advance.
Partial results on the three-dimensionalψψ?interaction have been established,
see [30], as well as for other more singular interactions [14].
6.5. Yang-Mills Theory. Much of the mathematical progress reviewed above
results from understanding functional integration and using those methods to con-
struct Euclidean field theories. Functional integration for gauge theories raises new
technical problems revolving about the rich group of symmetries, especially gauge
symmetry. Both the choice of gauge and the transformation between different
choices complicate the mathematical structure; yet gauge symmetry provides the
possibility of asymptotic freedom. Certain insights and proposals in the physics lit-
erature [9, 5] have led to an extensive framework; yet the implications of these ideas
for a mathematical construction of Yang-Mills theory need further understanding.
Wilson suggested a different approach based on approximating continuum space
time by a lattice, on which he defined a gauge-invariant action [48]. With a compact
gauge group and a compactified space-time, the lattice approximation reduces the
functional integration to a finite-dimensional integral. One must then verify the
existence of limits of appropriate expectations of gauge-invariant observables as the
lattice spacing tends to zero and as the volume tends to infinity.
Reflection positivity holds for the Wilson approximation [36], a major advan-
tage; few methods exist to recover reflection positivity in case it is lost through
regularization—such as with dimensional regularization, Pauli-Villiars regulariza-
tion, and many other methods. Establishing a quantum mechanical Hilbert space
is part of the solution to this Millennium Problem.
Balabanstudiedthisprograminathree-dimensionallatticewithperiodicbound-
ary conditions, approximating a space-time torus [2]. He studied renormalization
transformations (integration of large-momentum degrees of freedom followed by
rescaling) and established many interesting properties of the effective action they
produce. These estimates are uniform in the lattice spacing, as the spacing tends
to zero. The choices of gauges are central to this work, as well as the use of Sobolev
space norms to capture an analysis of geometric effects.
One defines these gauges in phase cells: the choices vary locally in space-time,
as well as on different length scales. The choices evolve inductively as the renormal-
ization transformations proceed, from gauges suited for local regularity (ultraviolet
gauges) to those suitable for macroscopic distances (infrared gauges). This is an
important step toward establishing the existence of the continuum limit on a com-
pactified space-time. These results need to be extended to the study of expectations
of gauge-invariant functions of the fields.
While this work in three dimensions is important in its own right, a qualitative
break-through came with Balaban’s extension of this analysis to four dimensions
12 ARTHUR JAFFE AND EDWARD WITTEN
[3]. This includes an analysis of asymptotic freedom to control the renormalization
group flow, as well as obtaining quantitative estimates on effects arising from large
values of the gauge field.
Extensive work has also been done on a continuum regularization of the Yang-
Mills interaction, and it has the potential for further understanding [39, 29].
These steps toward understanding quantum Yang-Mills theory lead to the vision
that one can extend the present methods to establish a complete construction of the
Yang-Mills quantum field theory on a compact, four-dimensional space-time. One
presumably needs to revisit known results at a deep level, simplify the methods,
and extend them.
New ideas are needed to prove the existence of a mass gap that is uniform in
the volume of space-time. Such a result presumably would enable the study of the
limit as T4 →R4.2
6.6. Further Remarks. Because four-dimensional gauge theory is a theory
in which the mass gap is not visible classically, to demonstrate it may require a
non-classical change of variables or “duality transformation.” For example duality
has been used to establish a mass gap in the statistical mechanics problem of a
Coulomb gas, where the phenomenon is known as Debye screening: macroscopic
test charges in a neutral Coulomb gas experience a mutual force that decays expo-
nentially with the distance. The mathematical proof of this screening phenomenon
proceeds through the identity of the partition function of the Coulomb gas to that
of a cos(λ?) (Sine-Gordon) field theory, and the approximate parabolic potential
near a minimum of this potential, see [6].
One view of the mass gap in Yang–Mills theory suggests that it could arise from
the quartic potential (A∧A)2 in the action, where F = dA+gA∧A, see [11], and
may be tied to curvature in the space of connections, see [44]. Although the Yang-
Mills action has flat directions, certain quantum mechanics problems with potentials
involving flat directions (directions for which the potential remains bounded as
|x|→∞) do lead to bound states [43].
A prominent speculation about a duality that might shed light on dynamical
properties of four-dimensional gauge theory involves the 1/N expansion [25]. It is
suspected that four-dimensional quantum gauge theory with gauge group SU(N)
(or SO(N), or Sp(N)) may be equivalent to a string theory with 1/N as the string
coupling constant. Such a description might give a clear-cut explanation of the
mass gap and confinement, and perhaps a good starting point for a rigorous proof
(for sufficiently large N). There has been surprising progress along these lines for
certain strongly coupled four-dimensional gauge systems with matter [31], but as
of yet there is no effective approach to the gauge theory without fermions. In-
vestigations of supersymmetric theories and string theories have given a variety of
other approaches to understanding the mass gap in certain four-dimensional gauge
theories with matter fields; for example, see [41].
2We specifically exclude weak-existence (compactness) as the solution to the existence part
of the Millennium Problem, unless one also uses other techniques to establish properties of the
limit (such as the existence of a mass gap and the axioms).
Bibliography
[1] Michael Aizenman, Geometric analysis of ?4 fields and Ising models, Commun. Math.
Phys. 86 (1982), 1–48.
[2] Tadeusz Balaban, Ultraviolet stability of three-dimensional lattice pure gauge field theo-
ries, Commun. Math. Phys. 102 (1985), 255–275.
[3] Tadeusz Balaban, Renormalization group approach to lattice gauge field theories. I: gen-
eration of effective actions in a small field approximation and a coupling constant renor-
malization in 4D, Commun. Math. Phys. 109 (1987), 249–301.
[4] Tadeusz Balaban, David Brydges, John Imbrie, and Arthur Jaffe, The mass gap for Higgs
models on a unit lattice, Ann. Physics 158 (1984), 281–319.
[5] C. Becchi, A. Rouet, and R. Stora, Renormalization of gauge theories, Ann. Phys. 98
(1976), 287–321.
[6] David Brydges and Paul Federbush, Debye screening, Commun. Math. Phys. 73 (1980),
197–246.
[7] David Brydges, J¨urg Fr¨ohlich, and Erhard Seiler, On the construction of quantized gauge
fields, I. Ann. Phys. 121 (1979), 227–284, II. Commun. Math. Phys. 71 (1980), 159–205,
and III. Commun. Math. Phys. 79 (1981), 353–399.
[8] M. Creutz, Monte carlo study of quantized SU(2) gauge theory, Phys. Rev. D21 (1980),
2308–2315.
[9] L. D. Faddeev and V. N. Popov, Feynman diagrams for the Yang-Mills fields, Phys. Lett.
B25 (1967), 29–30.
[10] Joel Feldman and Konrad Osterwalder, The Wightman axioms and the mass gap for
weakly coupled ?43 quantum field theories, Ann. Physics 97 (1976), 80–135.
[11] Richard Feynman, The quantitative behavior of Yang–Mills theory in 2 + 1 dimensions,
Nucl. Phys. B188 (1981), 479–512.
[12] J¨urg Fr¨ohlich, On the triviality of λ?4d theories and the approach to the critical point,
Nucl. Phys. 200[FS4] (1982), 281–296.
[13] J¨urg Fr¨ohlich, Konrad Osterwalder, and Erhard Seiler, On virtual representations of sym-
metric spaces and theory analytic continuation, Ann. Math. 118 (1983), 461–489.
[14] Kryzstof Gawedzki, Renormalization of a non-renormalizable quantum field theory, Nucl.
Phys. B262 (1985), 33–48.
[15] James Glimm and Arthur Jaffe, The λ?42 quantum field theory without cut-offs, I. Phys.
Rev. 176 (1968), 1945–1951, II. Ann. Math. 91 (1970), 362–401, III. Acta. Math. 125
(1970), 203–267, and IV. J. Math. Phys. 13 (1972), 1568–1584.
[16] James Glimm and Arthur Jaffe, Positivity of the ?43 Hamiltonian, Fortschr. Phys. 21
(1973), 327–376.
[17] James Glimm and Arthur Jaffe, ?42 quantum field model in the single phase region: differ-
entiability of the mass and bounds on critical exponents, Phys. Rev. 10 (1974), 536–539.
[18] James Glimm and Arthur Jaffe, Quantum Physics, Second Edition, Springer Verlag, 1987,
and Selected Papers, Volumes I and II, Birkh¨auser Boston, 1985. (Volume II includes
reprints of [15, 16, 19, 20].)
[19] James Glimm, Arthur Jaffe, and Thomas Spencer, The Wightman axioms and particle
structure in the P(?)2 quantum field model, Ann. of Math. 100 (1974), 585–632.
[20] James Glimm, Arthur Jaffe, and Thomas Spencer, A convergent expansion about mean
field theory, I. Ann. Phys. 101 (1976), 610–630, and II. Ann. Phys. 101 (1976), 631–669.
[21] D. J. Gross and F. Wilczek, Ultraviolet behavior of non-abelian gauge theories, Phys.
Rev. Lett. 30 (1973), 1343–1346.
[22] F. Guerra, L. Rosen, and B. Simon, The P(?)2 Euclidean quantum field theory as classical
statistical mechanics, Ann. Math. 101 (1975), 111–259.
[23] F. Guerra, L. Rosen, and B. Simon, Correlation inequalities and the mass gap in P(?)2,
III. Mass gap for a class of strongly coupled theories with non-zero external field, Commun.
Math. Phys. 41 (1975), 19–32.
[24] R. Haag, Local Quantum Physics Springer Verlag, 1992.
13
14 ARTHUR JAFFE AND EDWARD WITTEN
[25] G. ’t Hooft, A planar diagram theory for strong interactions, Nucl. Phys. B72 (1974),
461–473.
[26] John Imbrie, Phase diagrams and cluster expansions for low temperature P(?)2 models,
Commun. Math. Phys. 82 (1981), 261–304 and 305–343.
[27] Claude Itzykson and Jean-Bernard Zuber, Quantum Field Theory, McGraw-Hill, New
York, 1980.
[28] Arthur Jaffe, Constructive quantum field theory, in Mathematical Physics, edited by T.
Kibble, World Scientific, Singapore, 2000.
[29] Jacques Magnen, Vincent Rivasseau, and Roland S′en′eor, Construction of YM4 with an
infrared cutoff, Commun. Math. Phys. 155 (1993), 325–383.
[30] Jacques Magnen and Roland S′en′eor, Yukawa quantum field theory in three dimensions,
in Third International Conference on Collective Phenomena, J. Lebowitz et. al. editors,
New York Academy of Sciences, 1980.
[31] J. Maldacena, The large N limit of superconformal field theories and supergravity, Adv.
Theor. Math. Phys. 2 (1998), 231–252.
[32] O. McBryan and J. Rosen, Existence of the critical point in ?4 field theory, Commun.
Math. Phys. 51 (1976), 97–105.
[33] Edward Nelson, Quantum fields and Markoff fields, in Proc. Sympos. Pure Math. XXIII
1971, pp. 413–420, Amer. Math. Soc., Providence, R.I., 1973.
[34] L. O’Raifeartaigh, The Dawning of Gauge Theory Princeton University Priss, 1997.
[35] Konrad Osterwalder and Robert Schrader, Axioms for Euclidean Green’s functions, Com-
mun. Math. Phys. 31 (1973), 83–112, and Commun. Math. Phys. 42 (1975), 281–305.
[36] K. Osterwalder and E. Seiler, Gauge theories on the lattice, Ann. Phys. 110 (1978),
440–471.
[37] K. Osterwalder and R. S′en′eor, A nontrivial scattering matrix for weakly coupled P(?)2
models, Helv. Phys. Acta 49 (1976), 525–535.
[38] H. D. Politzer, Reliable perturbative results for strong interactions? Phys. Rev. Lett. 30
(1973), 1346–1349.
[39] V. Rivasseau, From Perturbative to Constructive Renormalization, Princeton University
Press, 1991.
[40] A. Salam, Weak and electromagnetic interactions, pp. 367–377 in Svartholm: Elementary
Particle Theory, Proceedings of The Nobel Symposium held in 1968 at Lerum, Sweden,
Stockholm, 1968.
[41] N. Seiberg and E. Witten, Electric-magnetic duality, monopole condensation, and con-
finement in N = 2 supersymmetric Yang–Mills theory, Nucl. Phys. B426 (1994), 19–52.
[42] E. Seiler, Schwinger functions for the Yukawa model in two dimensions, Commun. Math.
Phys. 42 (1975), 163–182.
[43] Barry Simon, Some quantum operators with discrete spectrum but classically continuous
spectrum, Ann. Phys. 146 (1983), 209–220.
[44] I. M. Singer, The geometry of the orbit space for nonabelian gauge theories, Phys. Scripta
24 (1981), 817–820.
[45] Raymond Streater and Arthur Wightman, PCT, Spin and Statistics and all That,
W. A. Benjamin, New York, 1964.
[46] Kurt Symanzik, Euclidean quantum field theory, in Local Quantum Theory, pp. 152–226,
R. Jost, Editor, Academic Press, New York, 1969.
[47] S. Weinberg, A model of leptons, Phys. Rev. Lett. 19 (1967), 1264–1266.
[48] K. G. Wilson, Quarks and strings on a lattice, in New Phenomena In Subnuclear Physics,
Proceedings of the 1975 Erice School, ed. A. Zichichi, Plenum Press, New York, 1977.
[49] Edward Witten, Supersymmetric Yang–Mills theory on a four-manifold, J. Math. Phys.
35 (1994), 5101–5135.
[50] C. N. Yang and R. L. Mills, Conservation of isotopic spin and isotopic gauge invariance,
Phys. Rev. 96 (1954), 191–195.
Harvard University November 2005
Institute for Advanced Study c?Copyright by the authors
|
|