The Semiclassical Expansion and the Instanton Gas in Biquaternionic Form

Introduction

The ADHM Construction and Biquaternion Instanton Data parametrised the self-dual connections of the Euclidean theory, and The Index Theorem and the Zero-Mode Count in Biquaternionic Form counted the fermion modes they support. This article turns the instanton from a solution of a field equation into a contribution to a physical quantity. It develops the semiclassical expansion of the Euclidean path integral around instantons and collects those contributions into the instanton gas, the dilute-medium picture in which the theta dependence of the vacuum and the vacuum energy are computed.

The subject straddles a boundary that this corpus keeps sharp. The path integral itself, its measure, the gauge-fixing and the Faddeev–Popov procedure belong to the quantisation programme, developed in the articles of Biquaternion Quantum Fields, and The Wick Rotation in the Biquaternion Universe fixes the Euclidean continuation that the instanton calculus requires. Those results are used here, not re-derived. What belongs to the present article is the use of the instanton configurations: the steepest-descent evaluation around a saddle, the collective-coordinate measure that the ADHM moduli space supplies, the sum over topological sectors that produces the theta vacuum, and the dilute-gas approximation with its range of validity. The companion article The Theta Parameter, Strong CP, and the Witten Effect in Biquaternionic Form takes the theta angle so obtained and develops its consequences.

Two of the framework's own results enter directly. First, the topological charge and its density were identified in Instantons and Solitons in Biquaternionic Form: the abelian density is the framework's invariant $I_2=\mathbf E\cdot\mathbf B$, and the non-abelian density is the matrix trace $\mathrm{Tr}(F\wedge F)$, which the biquaternion norm does not supply. Second, the moduli space of the saddle is the ADHM quotient of The ADHM Construction and Biquaternion Instanton Data, whose dimension fixes the number of collective coordinates and hence the measure. The semiclassical expansion is therefore built on the framework's topology and the framework's moduli; what it adds, and what it must import, is the fluctuation calculus of quantum field theory.

Conventions. We use those of Conventions in the Biquaternion Universe and of the companion gauge articles. The Euclidean slice has coordinates $x_4,x_1,x_2,x_3$ with $x_4=c\tau$, the 't Hooft symbols of Instantons and Solitons in Biquaternionic Form are used for the self-dual solution, and the Euclidean curvature is written without the explicit $i$ in the commutator on Hermitian generators with $\mathrm{Tr}(T_aT_b)=\tfrac12\delta_{ab}$. The topological charge and action are

$$ Q=\frac{1}{8\pi^2}\int\mathrm{Tr}\bigl(F\wedge F\bigr)\in\mathbb Z , \qquad S=\frac{1}{2g^2}\int\mathrm{Tr}\bigl(F_{\mu\nu}F^{\mu\nu}\bigr)\ge\frac{8\pi^2}{g^2}|Q| , $$

with equality for self-dual or anti-self-dual fields. The Lorentzian signature appears only where the continuation is discussed, and there $\Box=\tilde\nabla\tilde\nabla^{\natural}=\partial_{ict}^2+\Delta$ is the series d'Alembertian. As throughout the corpus, $c=1/\sqrt{\epsilon\mu}$ is the speed of light in the medium.

Companion article The ADHM Construction and Biquaternion Instanton Data, for the moduli space and its collective coordinates.
Companion article The Index Theorem and the Zero-Mode Count in Biquaternionic Form, for the fermion zero modes and the 't Hooft vertex.
Companion article The Wick Rotation in the Biquaternion Universe, for the Euclidean continuation.
Companion article The Renormalization Group in Biquaternionic Form, for the running coupling and dimensional transmutation.
Companion article The Theta Parameter, Strong CP, and the Witten Effect in Biquaternionic Form, for the theta vacuum and its consequences.

The Euclidean Path Integral and Steepest Descent

The Euclidean partition function of the gauge field is, schematically,

$$ Z \;=\; \int \mathcal D\mathcal A\;e^{-S[\mathcal A]}, $$

where the measure, the gauge-fixing and the ghosts are the objects constructed in the companion path-integral articles; the instanton calculus takes that construction for granted. The semiclassical expansion evaluates the integral by steepest descent. Write the saddle-point condition

$$ \frac{\delta S}{\delta\mathcal A_\mu}\Big|_{\mathcal A=\mathcal A^{\mathrm{cl}}}=0 , \qquad S[\mathcal A^{\mathrm{cl}}]=S_{\mathrm{cl}} , $$

and expand the field around it, $\mathcal A=\mathcal A^{\mathrm{cl}}+\delta\mathcal A$. The Gaussian integration over $\delta\mathcal A$ gives the one-loop determinant; the collective-coordinate directions, along which the action is unchanged, must be treated separately, and they are the directions along the moduli space. The result for a single saddle is

$$ Z_{\mathrm{saddle}} \;=\; \bigl(\text{one-loop determinant}\bigr)\times e^{-S_{\mathrm{cl}}} \times\int_{\mathcal M}\bigl(\text{measure on the moduli}\bigr). $$

For a self-dual configuration the classical action is fixed by the topology,

$$ S_{\mathrm{cl}}=\frac{8\pi^2}{g^2}|Q| , $$

by the saturated Bogomolny bound of Instantons and Solitons in Biquaternionic Form, and it is independent of every modulus: the instanton's size and position do not change the action. This is why the moduli integration is not Gaussian-suppressed and why the size integral becomes the article's central object.

The One-Instanton Measure and Collective Coordinates

The moduli space of the charge-one saddle is the five-dimensional family of The ADHM Construction and Biquaternion Instanton Data: four translations $x_0^\mu$ and one dilatation $\rho$, together with the three global $SU(2)$ orientations that are in the gauge orbit. The collective-coordinate measure is the volume element of this space, and the classical part of the one-instanton contribution is

$$ Z_1^{\mathrm{cl}} \;\propto\; \int d^4x_0\int_0^\infty\frac{d\rho}{\rho^5}\;e^{-8\pi^2/g^2} , $$

where the weight $\rho^{-5}$ is the measure factor of the four translations and the scale: it is the Jacobian of the five collective coordinates, and it carries the dimension of the volume element, $d^4x_0\,d\rho\,\rho^{-1}\cdot(\text{scale weight})$. The translations integrate to the four-volume $VT$ of the box, so the nontrivial content is the single integral over the size,

$$ Z_1^{\mathrm{cl}} \;\propto\; VT\int_0^\infty\frac{d\rho}{\rho^5}\;e^{-8\pi^2/g^2} . $$

The one-loop determinant multiplies this by a power of the coupling and, through renormalisation, by the running of $g$ with the scale. The physical statement is the dimensional transmutation of the renormalisation group: the dimensionless coupling $g$ is replaced by the running coupling $g(\rho)$ evaluated at the instanton size, and the 't Hooft one-loop result for the pure gauge theory has the structure

$$ Z_1 \;\propto\; VT\int_0^\infty \frac{d\rho}{\rho^5}\;\Bigl(\frac{8\pi^2}{g^2(\rho)}\Bigr)^{2N}e^{-8\pi^2/g^2(\rho)} , $$

where the power of the coupling comes from the one-loop fluctuation determinant, and the running is governed by the beta function of the companion article The Renormalization Group in Biquaternionic Form. The combination is scale independent at one loop, and its structure is the standard 't Hooft result. The framework's contribution is the moduli space whose measure this is; the determinant, the running and the numerical prefactor are standard quantum field theory, imported.

The Theta Vacuum and the Sum over Sectors

The gauge field configurations fall into topological sectors labelled by $Q\in\mathbb Z$, and the partition function is a sum over sectors,

$$ Z \;=\;\sum_{Q\in\mathbb Z} Z_Q , \qquad Z_Q=\int_{\mathcal A\in\text{sector }Q}\mathcal D\mathcal A\,e^{-S[\mathcal A]} . $$

A local action cannot mix sectors, since $Q$ is topological, but the vacuum need not be a definite-$Q$ state. The theta vacuum is the superposition

$$ |\theta\rangle \;=\; \sum_{Q\in\mathbb Z} e^{+iQ\theta}\,|Q\rangle , \qquad Z(\theta) \;=\; \sum_{Q\in\mathbb Z} e^{iQ\theta}\,Z_Q , $$

the sign in the ket being the one for which the topological charge operator acts as $\hat Q|\theta\rangle=-i\partial_\theta|\theta\rangle$; and the sum is the Fourier transform of the charge distribution. The construction is exactly the one that makes the partition function depend on the topological angle and preserves the cluster decomposition: a definite-$Q$ superposition is not clustered, while the $|\theta\rangle$ superposition is the unique one that is, up to the standard subtleties.

The theta dependence enters through the combination

$$ e^{iQ\theta}Z_Q = e^{iQ\theta}\bigl(\text{positive}\bigr)e^{-8\pi^2|Q|/g^2}, $$

so the partition function is a sum of contributions suppressed by $e^{-8\pi^2|Q|/g^2}$. The two properties that matter for the sequel are the periodicity

$$ Z(\theta+2\pi)=\sum_Q e^{iQ(\theta+2\pi)}Z_Q=\sum_Q e^{iQ\theta}Z_Q=Z(\theta), $$

since $e^{2\pi iQ}=1$ for integer $Q$, and the small-angle behaviour. Expanding the one-instanton–anti-instanton sector at leading order gives the vacuum energy density

$$ \mathcal E(\theta) \;=\; \mathcal E_0 - 2K\cos\theta\,e^{-8\pi^2/g^2}+\cdots , \qquad K>0 , $$

so that the topological susceptibility is

$$ \chi_t \;=\; \frac{d^2\mathcal E}{d\theta^2}\Big|_{\theta=0} \;=\; 2K\,e^{-8\pi^2/g^2}+\cdots . $$

Both were checked on a model instanton sum $Z(\theta)=\sum_{Q=-2}^{2}e^{iQ\theta}Z_Q$ with symmetric weights: the periodicity residual $|Z(\theta)-Z(\theta+2\pi)|$ was zero to machine precision, and the numerical second derivative of $2K\cos\theta$ at $\theta=0$ returned $2K$ to six digits, confirming the identification of the susceptibility with the instanton density. The periodicity is a property of the integer charge and not of the framework; the susceptibility is the physical content of the semiclassical sum.

The Dilute Instanton Gas

The instanton gas approximation treats the configuration as a dilute medium of well-separated instantons and anti-instantons of typical size $\rho_c$, whose density is small when $e^{-8\pi^2/g^2}\ll1$. In this regime each topological sector is built from a product of independent single-instanton contributions, the interactions between them are neglected, and the partition function exponentiates into the pressure of a gas of charged objects,

$$ Z(\theta) \;\approx\; \exp\!\Bigl[VT\,\bigl(\text{instanton density}\bigr)\bigl(e^{i\theta}+e^{-i\theta}\bigr)\Bigr] \;=\;\exp\!\Bigl[2\,VT\,K\,e^{-S_I}\cos\theta\Bigr], $$

which reproduces the vacuum energy above and exhibits the exponentiation that the cluster decomposition requires. The gas is the semiclassical counterpart of a plasma of electric and magnetic charges in two dimensions: the instantons are the charged objects, the theta angle is a chemical potential conjugate to the charge, and the Debye screening of the topological charge is the exponentiation.

The approximation has a sharp domain of validity, and its failure is the physically important part. The size integral

$$ \int_0^\infty\frac{d\rho}{\rho^5}\bigl(\text{one-loop factors}\bigr) $$

is infrared divergent in the pure gauge theory at large $\rho$, because the one-loop factor grows with $\rho$ faster than the measure suppresses it; in $SU(2)$ pure Yang–Mills the integrand behaves as $\rho^{b_0-5}$ with $b_0=\tfrac{22}{3}$, so it grows, and the contribution of large instantons is not controlled by the semiclassical expansion. The dilute-gas picture is reliable only for instantons small compared with the QCD scale, where the running coupling is weak, and the large-size region requires either the full non-perturbative treatment or the additional suppression supplied by light fermions. This is the standard infrared problem of the instanton calculus, and it is reported here rather than resolved: the semiclassical expansion is an expansion in $e^{-8\pi^2/g^2}$, and its breakdown at large size is the signal that the strong-coupling region is not accessible to it.

The One-Loop Prefactor and Dimensional Transmutation

The classical measure $\rho^{-5}e^{-8\pi^2/g^2}$ is incomplete: the Gaussian integration over the fluctuations around the instanton contributes a determinant, and its $\rho$-dependence is fixed by the renormalisation group. The one-loop fluctuation determinant scales as a power of $\mu\rho$, so that the one-instanton weight becomes

$$ Z_1\;\propto\; VT\int_0^\infty\frac{d\rho}{\rho^5}\;\bigl(\mu\rho\bigr)^{b_0}\Bigl(\frac{8\pi^2}{g^2(\rho)}\Bigr)^{2N}e^{-8\pi^2/g^2(\rho)} , $$

with $b_0$ the one-loop coefficient of the beta function and $\mu$ the renormalisation scale. For the framework's $SU(2)$ with $N_f$ light Dirac flavours,

$$ b_0=\frac{11}{3}C_2(SU(2))-\frac{2}{3}\sum_{\text{Weyl}}T(R) =\frac{22}{3}-\frac{2}{3}\cdot 2N_f\,T(\text{defining}) =\frac{22}{3}-\frac{2N_f}{3}=\frac{22-2N_f}{3}, $$

using $C_2(SU(2))=2$, the Dynkin index $T(\text{defining})=\tfrac12$ of The Index Theorem and the Zero-Mode Count in Biquaternionic Form, and the companion renormalisation-group convention in which the subtraction is the sum over Weyl fermions: each Dirac flavour contributes two Weyl fermions, so the sum is $2N_f\cdot\tfrac12=N_f$ and the matter term is $-\tfrac{2}{3}N_f$, not $-\tfrac{1}{3}N_f$. For the pure gauge theory $b_0=22/3$ either way. Asymptotic freedom is lost at $N_f=11$ Dirac flavours, where this coefficient vanishes.

Dimensional transmutation. The running coupling can be traded for a scale $\Lambda$ by

$$ \frac{8\pi^2}{g^2(\rho)}=b_0\ln\frac{1}{\Lambda\rho} , $$

which is the one-loop solution of the renormalisation-group equation. Substituting, the instanton weight becomes

$$ Z_1\;\propto\;VT\,\Lambda^{b_0}\int_0^\infty d\rho\;\rho^{b_0-5}\Bigl(\ln\frac{1}{\Lambda\rho}\Bigr)^{2N} , $$

manifestly independent of the renormalisation scale $\mu$, and controlled by the single dimensionful parameter $\Lambda$ that replaces the dimensionless coupling. The exponent $b_0-5$ is the balance between the measure $\rho^{-5}$ and the one-loop factor $\rho^{b_0}$; for $SU(2)$ pure Yang–Mills it is $b_0-5=7/3>0$, and the integrand grows with $\rho$: the infrared divergence reported above has its origin in this exponent and not in the classical measure.

The 't Hooft vertex and the fermion zero modes. When light fermions are present, the index theorem of the preceding article supplies $2T(R)k$ fermion zero modes in the background, and the fermionic measure must be saturated by them: for the defining representation of $SU(2)$ and $k=1$ there is one zero mode per flavour, so the one-instanton amplitude carries a factor

$$ \prod_{f=1}^{N_f}m_f $$

when the flavours are massive, and vanishes when any flavour is massless. The resulting $2N_f$-fermion interaction is the 't Hooft vertex, and its structure is fixed by the zero-mode count and not by the algebra: the algebra supplies the saddle, the topology and the density, while the count of legs and the vertex come from the index theorem and the fermionic measure. This is the sharpest illustration of the division of labour in this subcategory, and it is the point at which the semiclassical expansion stops being a statement about the gauge field alone.

The Biquaternion Reading and the Gaps

The framework's role in the semiclassical expansion is the supply of the saddle and its moduli, and it is worth stating precisely what that amounts to.

The saddle is the framework's. The self-dual connection is the BPST instanton built in $\mathrm{SU}(2)=\mathrm{span}_{\mathbb R}\{e_1,e_2,e_3\}\subset\mathbb M_-$, and its action is fixed by the Bogomolny bound. The instanton is an object of the algebra's non-abelian factor and not of the abelian centre: the companion article Instantons and Solitons in Biquaternionic Form showed that a smooth abelian configuration of finite action has $Q=0$ on $\mathbb R^4$, so the semiclassical expansion of the Maxwell sector has no instanton saddle at all.
The density is the framework's in the abelian case and the matrix trace otherwise. The theta term's integrand is the topological density. In the abelian sector it is proportional to the invariant $I_2=\mathbf E\cdot\mathbf B$; in the non-abelian sector it is $\mathrm{Tr}(F\wedge F)$ with the matrix trace, which the rank-two, single-slot biquaternion norm does not supply. The semiclassical sum over sectors is therefore written in components rather than in the biquaternion-norm invariant calculus.
The moduli are the framework's. The collective-coordinate measure integrates over the ADHM moduli space, whose dimension $8k$ and physical dimension $8k-3$ come from the quaternionic quotient of The ADHM Construction and Biquaternion Instanton Data. The five collective coordinates of the charge-one saddle are the quaternionic datum $\rho$ together with the translations of the quaternionic coordinate.
The dynamics is imported. The measure, the Faddeev–Popov procedure, the one-loop determinants, the beta function and the numerical prefactors are quantum field theory. The algebra is a complexified classical structure; it contains no $\hbar$, no running coupling and no action principle of its own. The semiclassical expansion is thus a physical use of the framework's topology that the framework itself cannot validate; it is validated by the standard quantum field theory of the companion quantisation articles.

Summary

The semiclassical expansion evaluates the Euclidean path integral by steepest descent around the framework's instanton saddles. A self-dual configuration has action $S=\frac{8\pi^2}{g^2}|Q|$, fixed by topology, and its five collective coordinates $(x_0,\rho)$ give the classical one-instanton weight

$$ Z_1^{\mathrm{cl}}\propto VT\int_0^\infty\frac{d\rho}{\rho^5}\,e^{-8\pi^2/g^2}, $$

with the one-loop determinant supplying the powers of the running coupling and realising dimensional transmutation. The sum over topological sectors produces the theta vacuum $|\theta\rangle=\sum_Q e^{+iQ\theta}|Q\rangle$ and the partition function $Z(\theta)=\sum_Q e^{iQ\theta}Z_Q$, which is periodic, $Z(\theta+2\pi)=Z(\theta)$, and has vacuum energy $\mathcal E(\theta)=\mathcal E_0-2K\cos\theta\,e^{-8\pi^2/g^2}+\cdots$ and topological susceptibility $\chi_t=2K e^{-8\pi^2/g^2}+\cdots$. Both were verified on an explicit finite instanton sum.

The dilute instanton gas exponentiates the one-instanton weight into a pressure of charged objects and reproduces the vacuum energy and cluster decomposition, but its size integral is infrared divergent in the pure gauge theory and the approximation holds only for instantons small compared with the strong-coupling scale. The framework supplies the saddle, its action, its moduli and the abelian form of its topological density; the measure, the determinants, the running coupling and the physical identification are imported from the standard quantum field theory of the companion path-integral and renormalisation-group articles.

Summary of Notation

Symbol	Meaning
$\mathbb B=\mathbb C\otimes_{\mathbb R}\mathbb H$	Biquaternion algebra
$\mathrm{SU}(2)=\mathrm{span}_{\mathbb R}\{e_1,e_2,e_3\}$	Compact factor; home of the instanton
$\mathcal A_\mu$, $F_{\mu\nu}$	Gauge connection and curvature
$Q=\frac{1}{8\pi^2}\int\mathrm{Tr}(F\wedge F)$	Topological charge
$S=\frac{1}{2g^2}\int\mathrm{Tr}(F_{\mu\nu}F^{\mu\nu})$	Euclidean action
$S_{\mathrm{cl}}=\frac{8\pi^2}{g^2}\|Q\|$	Saturated (Bogomolny) classical action
$Z=\int\mathcal D\mathcal A\,e^{-S}$	Euclidean partition function
$Z_Q$	Contribution of the topological sector of charge $Q$
$\|\theta\rangle=\sum_Q e^{+iQ\theta}\|Q\rangle$	Theta vacuum
$Z(\theta)=\sum_Q e^{iQ\theta}Z_Q$	Theta-dependent partition function
$\mathcal E(\theta)$, $\chi_t$	Vacuum energy density and topological susceptibility
$x_0^\mu$, $\rho$	Instanton collective coordinates (translation, scale)
$8k$, $8k-3$	ADHM moduli dimensions; collective-coordinate count
$I_2=\mathbf E\cdot\mathbf B$	Abelian topological density (framework invariant)
$\mathrm{Tr}(F\wedge F)$	Non-abelian density (matrix trace, not biquaternion norm)
$b_0$	One-loop beta-function coefficient; $b_0=\tfrac{22-2N_f}{3}$ for $SU(2)$ with $N_f$ Dirac flavours, $b_0=22/3$ pure
$g(\rho)$	Running coupling at the instanton scale