The Index Theorem and the Zero-Mode Count in Biquaternionic Form

Introduction

The Hopf fibration of the companion article The Hopf Fibration and the Biquaternion Gauge Bundle supplied the geometric home of the gauge field and two integers: the winding of a map from the three-sphere, which is the instanton number, and the Chern number of a circle bundle over the two-sphere, which is the magnetic charge. This article supplies the tool that turns those integers into a count. The Atiyah–Singer index theorem states that the difference between the number of positive-chirality and negative-chirality zero modes of a Dirac operator is a topological integral, and in the gauge-theoretic setting that integral is the same density whose integral gave the instanton charge in Instantons and Solitons in Biquaternionic Form. The charge that classifies the field and the number of fermion zero modes that the field supports are therefore the same number up to the Dynkin index of the representation: the index is $2T(R)$ times the charge, and for the defining representation of $\mathrm{SU}(2)$, where $2T=1$, the two coincide.

The framework enters through three points, two of which are its own and one of which is not.

  • The operator is the framework's. The Dirac operator in biquaternionic form is built from the biquaternionic gradient of Operator and Calculus Conventions, §The Gradient, and its coupling to the connection, and its composition with its conjugate returns the d'Alembertian of the same section. Its carrier is the algebra itself, and The Dirac Equation in Biquaternionic Form fixes the equation. The index is a property of that operator.
  • The counting is by the topological density, which the framework already has. The index density is a total derivative whose integral is the instanton number, and the abelian form of the density is the framework's invariant $I_2=\vec E\cdot\vec B$. This was established in Instantons and Solitons in Biquaternionic Form and is not repeated here.
  • The trace is the matrix trace, and the framework's biquaternion norm does not supply it. The non-abelian index density is $\mathrm{Tr}(F\wedge F)$ with the matrix trace on the gauge factor. The framework's quadratic invariant, the biquaternion norm of Algebra and Element Conventions, is a bilinear form of a single algebra element and has no second slot in which to trace two gauge-algebra-valued objects. This is the same rank-two limitation recorded for the topological density, and it controls what the framework can and cannot express about the index.

The article states the theorem as the standard result it is, derives the specialisations that the framework uses, and keeps the framework's own content apart from the standard content.

  • Companion article The Dirac Equation in Biquaternionic Form, for the operator whose index is taken.
  • Companion article The Spinor Module in Biquaternionic Form and Its Lorentz Action, for the module and its chirality.
  • Companion article Instantons and Solitons in Biquaternionic Form, for the topological background and its charge.
  • Companion article The Hopf Fibration and the Biquaternion Gauge Bundle, for the abelian bundle and its charge.
  • Companion article The ADHM Construction and Biquaternion Instanton Data, for the moduli whose dimension the adjoint index counts.
  • Companion article Clifford Modules and the Twisted Cauchy–Riemann Operator, for the theorem in its mathematical form.

The Dirac Operator and Its Index

Notation. The imaginary unit has $i^2=-1$, and it is the only imaginary unit in the article. Where two of them meet in a product they are reduced before the formula is read. A field is an element of $\mathbb{B}$, and an action on it is a multiplication, written $a\tilde{Ξ}$ when it acts from the left and $\tilde{Ξ} b$ when it acts from the right; the side is read from the position, and no letter names it. The letter $R$ labels a representation, on $\tilde D_R$, $T(R)$ and $\mathrm{Tr}_R$; the same letter as a subscript on a field is the chirality value of the section below.

Frame. Displays are read in the $\mathbb{B}$ frame of Frame and Metric Conventions, with the four complex coordinates $Q_\mu$ and the positive-definite metric that frame carries. The index is a property of an elliptic operator on a Riemannian manifold, and this is the frame whose flat slice is Euclidean.

The gradient. The derivative of the article is the biquaternionic gradient of Operator and Calculus Conventions, §The Gradient,

$$ \tilde\nabla_\Phi=\sum_{\mu\in\{t,x,y,z\}}e_\mu\frac{\partial}{\partial R_\mu} =e_t\frac{\partial}{\partial R_t}+\vec e_x\frac{\partial}{\partial R_x}+\vec e_y\frac{\partial}{\partial R_y}+\vec e_z\frac{\partial}{\partial R_z}, $$

with its quaternion conjugate and the scalar d'Alembertian

$$ (\tilde\nabla_\Phi)^{\natural}=\sum_{\mu\in\{t,x,y,z\}}\varepsilon_\mu e_\mu\frac{\partial}{\partial R_\mu}, \qquad \Box_\Phi=\mathrm{Sc}\bigl((\tilde\nabla_\Phi)^{\natural}\tilde\nabla_\Phi\bigr) =\frac{\partial^2}{\partial R_t^2}+\Delta_\Phi , \qquad \varepsilon=(1,-1,-1,-1). $$

Each term is a left multiplication by one frame unit. The symbol is the same sum with the derivative replaced by the covector component, so it is an element of the algebra and not an array.

The operator. The twisted Dirac operator in the representation $R$ is the gauge-covariant gradient

$$ \tilde D_R=\sum_{\mu}e_\mu D_\mu,\qquad D_\mu\tilde{Ξ}=\frac{\partial\tilde{Ξ}}{\partial R_\mu}+i\kappa\,\mathcal A_\mu\tilde{Ξ}, $$

with the sum over $\mu\in\{t,x,y,z\}$ and the connection $\mathcal A_\mu$ valued in the gauge algebra, multiplying the field from the left. The components are those of the gauge-covariant derivative of Algebra and Element Conventions, and

$$ [D_\mu,D_\nu]\tilde{Ξ}=i\kappa\,F_{\mu\nu}\tilde{Ξ},\qquad F_{\mu\nu}=\frac{\partial\mathcal A_\nu}{\partial R_\mu}-\frac{\partial\mathcal A_\mu}{\partial R_\nu}+i\kappa[\mathcal A_\mu,\mathcal A_\nu]. $$

The symbol, and the square. The principal symbol at a covector $\zeta$ is the left multiplication by $\sum_\mu\zeta_\mu e_\mu$. That element is invertible for every real $\zeta\neq0$, because its norm is $\sum_\mu\zeta_\mu^2$ and that is nonzero; so the operator is elliptic, and on a closed manifold its kernel and cokernel are finite-dimensional. The square of the symbol is not a multiple of the identity,

$$ \Bigl(\sum_\mu\zeta_\mu e_\mu\Bigr)^2=\zeta_t^2-\sum_\eta\zeta_k^2+2\zeta_t\sum_\eta\zeta_k\vec e_\eta , $$

because $e_t$ commutes with the three spatial units, so the mixed time–space terms add; the three spatial units anticommute, so their mixed terms vanish; and each spatial square is $-e_t$, which gives the negative sum. Ellipticity needs the invertibility of the symbol and not of its square, so the vector term changes nothing.

The square of the operator is not scalar either. What is scalar is the composition with the conjugate, $\mathrm{Sc}((\tilde\nabla_\Phi)^{\natural}\tilde\nabla_\Phi)=\Box_\Phi$, and on the material sector that reads $\Delta_\Phi-\partial^2/\partial {r'_t}^2$. The connection enters through the commutator $[D_\mu,D_\nu]$, and for the pure-gauge connection $\mathcal A_\mu=(i\kappa)^{-1}u^{-1}\partial u/\partial R_\mu$ the covariant derivative is a conjugation of the free one, $D_\mu\tilde{Ξ}=u^{-1}\partial(u\tilde{Ξ})/\partial R_\mu$.

The first order, and the mass. The free equation is first order and carries one field, $\tilde\nabla_\Phi\tilde{Ξ}=0$. The massive equation is the chiral pair

$$ \tilde\nabla_\Phi\tilde{Ξ}_{R}=k_m\tilde{Ξ}_{L}, \qquad (\tilde\nabla_\Phi)^{\natural}\tilde{Ξ}_{L}=k_m\tilde{Ξ}_{R} , $$

and it is not a one-field first-order equation: $\tilde\nabla_\Phi^2$ is not scalar, so no single first-order operator on one field has a scalar square to take a mass from. The first-order form of the d'Alembertian is the pair of Operator and Calculus Conventions, §The Two First-Order Equations and Their Symbol, and each line composed with the conjugate of the other returns $\Box_\Phi$; $\tilde\nabla_\Phi^2$ is not $\Box_\Phi$, and the article states the pair and never a one-field massive form. Here $k_m=mc/\hbar$ is the mass wave number, the inverse of the mass length.

The chirality. Chirality is the subscript, and it takes two values. A field is written $\tilde{Ξ}_{L}+\tilde{Ξ}_{R}$: $L$ is the minus value and $R$ the plus value, in the Dirac letters, and the split is unique, the two parts of the zero field being both zero (Special Element Conventions, §The Chirality Convention and §The Spinor Convention). The two values are the two eigenspaces of right multiplication by $-i\vec e_z$, the frame's chirality $\xi$; the body carries the subscript alone. A right multiplication commutes with every left multiplication, so the gradient commutes with $\xi$ and preserves each value: the frame's chirality is even on the operator, and it does not grade it.

The grading. An index compares two kernel dimensions across a grading that the operator reverses, and this grading is not the algebra's. It belongs to the spinor factor of the twisted operator and is written $G_M$, with a capital to distinguish it from the frame's $\xi$; it anticommutes with the manifold Dirac operator, while the algebra acts on the fibre and commutes with $G_M$. On the two sides of $G_M$ the twisted operator is $\tilde D_R^{+}$ on the plus side and $\tilde D_R^{-}$ on the minus side, and

$$ n_+=\dim\ker\tilde D_R^{+},\qquad n_-=\dim\ker\tilde D_R^{-}. $$

The two are adjoints of one another in the Hermitian form, because the conjugate of the gradient is the operator that form pairs with it, $\tilde\nabla_\Phi^{*}=(\tilde\nabla_\Phi)^{\natural}$ (Operator and Calculus Conventions, §The Gradient, the conjugates of the gradient); the integration by parts carries a sign, and a sign moves no kernel. Nothing forces the two dimensions to agree, and that asymmetry is what the index measures.

The two Dirac equations. The framework's equation is the chiral pair of Operator and Calculus Conventions, §The Two First-Order Equations and Their Symbol, each member acting from the left and the mass coupling the two parts; no single glyph is written for them. The mass is the pairing of the two parts and a zeroth-order term, and it does not move the index: the index of an elliptic first-order operator is fixed by its symbol, which the mass leaves untouched. The index computed below is therefore the index of the twisted operator, and the gauge-covariant gradient $\tilde D_R$ of the section above is its derivative.

The index. The index is the difference

$$ \mathrm{ind}\,\tilde D_R=n_+-n_- , $$

an integer, and locally constant in the connection and in the metric; the two facts together make it a topological invariant, unchanged by any continuous deformation. It is the property of the operator that the rest of the article computes.

The Atiyah–Singer Theorem

For a closed even-dimensional spin manifold $M$ and a twisted Dirac operator, the index is the integral of a characteristic class,

$$ \boxed{\;\mathrm{ind}\,\tilde D_R \;=\; \int_M A_{\mathrm{aro}}(TM)\wedge \mathrm{ch}(E_R)\;}, $$

where $A_{\mathrm{aro}}(TM)$ is the A-roof genus of the tangent bundle and $\mathrm{ch}(E_R)$ is the Chern character of the twisting bundle. This is the Atiyah–Singer theorem; it is standard and is transcribed here, not derived. Its power in the gauge-theoretic setting is that the right-hand side is purely topological: it depends on the bundle data and not on the shape of the connection.

On the flat four-dimensional slice that the framework uses, the tangent bundle is trivial and the A-roof genus reduces to $1$: $A_{\mathrm{aro}}=1-\frac{p_1}{24}+\cdots$, and the Pontryagin class $p_1$ vanishes when the metric is flat. The index is therefore the degree-four part of the Chern character alone. With the trace in $R$ and the standard normalisation, that part is proportional to $\mathrm{Tr}_R(F\wedge F)$, with constant $\frac{1}{8\pi^2}$; the overall sign is carried by the orientation of $M$ together with the naming of the positive chirality, and both are fixed once here, so that

$$ \boxed{\;\mathrm{ind}\,\tilde D_R \;=\; \frac{1}{8\pi^2}\int_{\mathbb R^4}\mathrm{Tr}_R\bigl(F\wedge F\bigr) \;=\; 2\,T(R)\,Ϙ\;}, $$

where $T(R)$ is the Dynkin index of the representation, normalised so that the defining representation of $\mathrm{SU}(2)$ has $T=\tfrac12$, equivalently $2\,\mathrm{Sc}(T_aT_b)=\tfrac12\delta_{ab}$ there, and $Ϙ$ is the integer topological charge of the companion article, defined with the trace in the defining representation. The two statements to read from this formula are that the index is proportional to the topological charge, and that its constant of proportionality is a representation-theoretic number.

The Dynkin index. For $\mathrm{SU}(2)$ the two representations that occur inside the framework are the defining module of dimension two and the adjoint representation of dimension three, and their indices are fixed by the Casimir relation $T(R)=\dim R\,C_2(R)/\dim G$, with $C_2(\text{defining})=\tfrac34$, $C_2(\text{adjoint})=2$ and $\dim G=3$:

$$ T(\text{defining})=\frac{2\cdot\tfrac34}{3}=\frac12 , \qquad T(\text{adjoint})=\frac{3\cdot 2}{3}=2 . $$

Both are computed from the same commutation relations $[T_a,T_b]=\varepsilon_{abc}T_c$ that define $\mathfrak{su}(2)\subset\mathbb{B}_{\Phi}$ in Non-Abelian Gauge Fields in Biquaternionic Form, with $T_a=\tfrac12 \vec e_a$, so the group theory here is the framework's group theory. The index formula then reads

$$ \begin{aligned} \mathrm{ind}\,\tilde D_{\text{defining}}&=Ϙ,\\ \mathrm{ind}\,\tilde D_{\text{adjoint}}&=4Ϙ. \end{aligned} $$

For a self-dual configuration of charge $Ϙ=k>0$ — the convention of the companion article, in which the self-dual unit instanton has $Ϙ=+1$ — all zero modes carry one chirality (the positive one in the convention fixed above): $n_+=k$, $n_-=0$ in the defining representation, and $n_+=4k$, $n_-=0$ in the adjoint. A single BPST instanton therefore supports one fermion zero mode in the doublet and four in the triplet.

The abelian specialisation. For the Hopf bundle over the two-sphere, the same theorem reduces to the two-dimensional statement

$$ \mathrm{ind}\,\tilde D_n \;=\; \frac{1}{2\pi}\int_{S^2}F \;=\; n , $$

the Chern number of the bundle, which is $1$ for the Hopf bundle. Because the curvature of a $\mathrm{U}(1)$ bundle is a definite multiple of the area form, one of the two kernels is empty and the index is the total number of zero modes: the count is $\lvert n\rvert$, with the sign of $n$ naming the chirality that carries them. The two-sphere case is the simplest instance of the same counting: index equals topological charge, and the charge is a winding number of the algebra's group.

Boundary corrections. On a manifold with boundary the index acquires a boundary contribution, and the theorem is replaced by the Atiyah–Patodi–Singer form

$$ \mathrm{ind}\,\tilde D_R \;=\; \int_M A_{\mathrm{aro}}(TM)\wedge\mathrm{ch}(E_R) - \frac{h+\eta(0)}{2}, $$

where $h$ is the dimension of the space of boundary zero modes and $\eta(0)$ is the eta invariant of the boundary Dirac operator. The correction is a boundary term, and it is the reason a background with a nontrivial boundary can support zero modes whose count is not given by the bulk integral alone. The framework's instanton boundary is the three-sphere of The Hopf Fibration and the Biquaternion Gauge Bundle, and the eta invariant is a spectral invariant of the operator restricted to that boundary; its evaluation is a standard analytic computation and is not an algebraic statement about $\mathbb B$.

The Density, the Charge, and the Total Derivative

The index is computed by the density $\mathrm{Tr}(F\wedge F)$, and the reason it is topological is that this density is a total derivative. The companion article established the chain

$$ \mathrm{Tr}\bigl(F\wedge F\bigr)=d\,\Omega_{\mathrm{CS}} , $$

with the Chern–Simons three-form

$$ \Omega_{\mathrm{CS}}=\mathrm{Tr}\Bigl(\mathcal A\wedge d\mathcal A+\tfrac23\,i\kappa\,\mathcal A\wedge\mathcal A\wedge\mathcal A\Bigr), $$

and therefore

$$ Ϙ=\frac{1}{8\pi^2}\oint_{\partial\mathbb R^4}\Omega_{\mathrm{CS}} , $$

a surface integral over the three-sphere at infinity. The coefficient $\tfrac23 i\kappa$ is the one that makes the identity exact: the derivative of the cubic term contributes $3\cdot\tfrac23 i\kappa=2i\kappa$ times $\mathrm{Tr}(d\mathcal A\wedge\mathcal A\wedge\mathcal A)$, which is the cross term of $\mathrm{Tr}(F\wedge F)$, and the quartic term $\mathrm{Tr}(\mathcal A^4)$ vanishes — the trace is cyclic and the wedge is graded antisymmetric, so moving one of the four factors past the other three multiplies the term by the sign of a four-cycle, and that sign is $-1$. The index theorem's integral and the instanton's charge are therefore the same integral, computed in two ways: as the integral of the index density over the bulk, and as the Chern–Simons flux through the boundary. They agree in the defining representation, and differ by the Dynkin factor $2T(R)$ otherwise.

In the abelian case the density collapses to the framework's second invariant. Every index in these displays is written low and the four labels are summed explicitly. The alternating symbol is the orientation density: in the Lorentzian ordering $(Q_t,Q_x,Q_y,Q_z)$ it is fixed by $\epsilon_{0123}=+1$, and in the Euclidean ordering $(x_4,x_1,x_2,x_3)$ by $\epsilon_{4123}=+1$; no index is raised, so no metric sign enters it. With the components $F_{0\eta}=iE_\eta/c$ and $F_{j\eta}=\epsilon_{j\eta l}B_l$ in the Lorentzian case, and $F_{4\eta}=E_\eta$ and $F_{j\eta}=\epsilon_{j\eta l}B_l$ in the Euclidean one, the contraction reads

$$ \frac14\sum_{\mu,\nu,\rho,\sigma}\epsilon_{\mu\nu\rho\sigma}F_{\mu\nu}F_{\rho\sigma} = \begin{cases} \dfrac{2i}{c}\,I_2 & (\text{Lorentzian}),\\[2mm] 2\,I_2 & (\text{Euclidean}), \end{cases} \qquad I_2=\vec E\cdot\vec B , $$

so the abelian index density is the framework's own invariant $I_2$ up to a constant. This is the affirmative half of the story: for the abelian sector the index and the charge are written in the invariant calculus the framework already possesses.

A Worked Case: the Abelian Index and the Charge-$n$ Zero Modes

The index formula is worth exercising in the framework's abelian sector, where every ingredient is the algebra's. Take the two-sphere with a $\mathrm{U}(1)$ gauge field whose first Chern number is $n$; the associated line bundle has $c_1=n$, and the Hopf bundle of The Hopf Fibration and the Biquaternion Gauge Bundle is the case $n=1$. The twisted Dirac index is

$$ \mathrm{ind}\,\tilde D_n=\int_{S^2}A_{\mathrm{aro}}(TS^2)\wedge\mathrm{ch}(L)=\int_{S^2}c_1(L)=n , $$

since the A-roof genus of a surface is $1$. The count of zero modes is therefore $\lvert n\rvert$, with the sign of $n$ selecting the chirality, and for $n=1$ there is exactly one zero mode. This is the standard result that a charge-$n$ background binds one fermion mode per unit charge, and it is the abelian counterpart of the $n_+=k$ count in the defining representation.

The zero modes can be exhibited explicitly. In the $\mathrm{U}(1)$ patch with $a=\tfrac12(1-\cos\theta)d\theta_{\mathrm{azm}}$ of the Hopf article the field equation $\tilde D\tilde{Ξ}=0$ separates, and for $n>0$ the $R$ solutions are the lowest monopole harmonics of the field,

$$ \tilde{Ξ}_{n}^{(m)}=f_n(\theta)\,\exp(im\theta_{\mathrm{azm}}), \qquad n_{\mathrm{mag}}=-j,\dots,j, \qquad j=\frac{n-1}{2}, $$

which number $2j+1=n$, matching the index; for $n=1$ the single mode is the constant section allowed in the patch where the connection is regular. The count is not the degeneracy of the scalar lowest Landau level in the same flux, which is $n+1$: the Dirac operator counts the harmonics of a field of spin one half, and the shift by one is the spin. What the framework supplies here is the identification of the bundle, its connection and the charged field with the Hopf data already computed; the counting itself is the standard one.

The Index and the Instanton Moduli: $4k$ Adjoint Modes

The index of the adjoint Dirac operator has a second interpretation that ties this article to the instanton data. For a self-dual background of charge $k$ in the adjoint representation,

$$ n_+=4k , \qquad n_-=0 , $$

and the framed moduli space of the background has real dimension $8k$. For $\mathrm{SU}(2)$ that space $\mathcal M_k$ is a smooth hyperkähler manifold of real dimension $8k$ and complex dimension $4k$ in each of its complex structures; the $4k$ adjoint zero modes of one chirality are counted as complex objects, so they match the complex dimension, while the $8k$ real collective coordinates are the tangent space. The index

$$ \mathrm{ind}\,\tilde D_{\text{adjoint}}=4k=\tfrac12\dim_{\mathbb R}\mathcal M_k=\dim_{\mathbb C}\mathcal M_k $$

is the precise form of this statement for $\mathrm{SU}(2)$.

Two consequences follow and are used in the later articles. First, the adjoint zero modes are the infinitesimal deformations of the background, so they are tangent to the moduli space: integrating over the moduli in the semiclassical expansion of The Semiclassical Expansion and the Instanton Gas in Biquaternionic Form is integrating over the collective coordinates, and the Grassmann integral over the zero modes fills that measure mode for mode. Second, the number of zero modes is a topological invariant and so cannot change under a deformation of the background within its charge sector. The standard bridge to the moduli counting of the ADHM construction of The ADHM Construction and Biquaternion Instanton Data is the quantum mechanics whose target is $\mathcal M_k$, whose index is the Euler characteristic of $\mathcal M_k$.

The Biquaternion Reading and the Rank-Two Limitation

The framework's contribution to the index theorem is not the theorem but the identification of its ingredients. The operator is the gauge-covariant gradient $\tilde D_R$ above, whose composition with its conjugate has scalar part $\Box_\Phi$; the carrier is the algebra $\mathbb{B}$ read as a module over itself, and the two chiral halves are its two parts $\mathbb{B}_L$ and $\mathbb{B}_R$, which is the reading of The Dirac Equation in Biquaternionic Form on the regular module. The adjoint representation is the three-dimensional real subspace $\mathfrak{su}(2)=\mathrm{span}_{\mathbb R}\{\vec e_x,\vec e_y,\vec e_z\}\subset\mathbb{B}_{\Phi}$ that the algebra carries as its own commutator algebra; and the group theory that produces the Dynkin indices is read directly from the quaternion commutators $[\vec e_a,\vec e_b]=2\varepsilon_{abc}\vec e_c$. The zero-mode counts $k$ and $4k$ are therefore a statement about the algebra's modules: the defining module gives one mode per unit charge, and the adjoint, being the algebra acting on itself, gives four.

What the framework does not supply is the invariant expression of the non-abelian index density. The density $\mathrm{Tr}(F\wedge F)$ requires the matrix trace over the gauge factor and the wedge of two algebra-valued two-forms, and the framework's biquaternion norm cannot produce it. The biquaternion norm pairs a single algebra element with its conjugate through the quaternion product and returns one scalar; it has one algebra slot. The index density needs two. This is the same rank-two limitation that the companion article found for the topological density and that the energy–momentum-tensor article found for the stress tensor: the framework's quadratic invariant is rank-two and single-slot, while the non-abelian index density is traced over a second algebra factor. The consequence is precise: the framework can write the index density in components with the matrix trace, and it can identify its integral as the topological charge, but the density is not an object of the biquaternion-norm invariant calculus. The zero-mode count is expressible; the invariant whose integral computes it, in the non-abelian case, is not one of the framework's invariants.

A second place where the count is imported is the fermion itself. The index theorem counts zero modes of a Dirac operator, and the operator presupposes a field valued in the algebra, the multiplication that acts on it and the chirality label; those are supplied by The Dirac Equation in Biquaternionic Form and Chiral Fermions in the Biquaternion Framework, and the physical identification of the counted modes with fermionic states is standard quantum field theory. The relevant anomaly statement — that the divergence of the axial current is proportional to the same density — belongs to the spin-half programme of the corpus and is not derived here; the index theorem is its Euclidean, elliptic counterpart, and the two are related by the standard argument.

Summary

The index of the biquaternionic Dirac operator is the difference $n_+-n_-$ of the numbers of positive- and negative-chirality zero modes, and the Atiyah–Singer theorem computes it as the integral of the index density $A_{\mathrm{aro}}(TM)\wedge\mathrm{ch}(E_R)$. On the flat four-dimensional slice the A-roof genus is trivial and the index is $2T(R)Ϙ$, where $Ϙ$ is the instanton charge of Instantons and Solitons in Biquaternionic Form and $T(R)$ is the Dynkin index. For the framework's two representations the indices are $T(\text{defining})=\tfrac12$ and $T(\text{adjoint})=2$, so the counts are $n_+=k$ in the defining module and $n_+=4k$ in the adjoint, one zero mode and four for the unit self-dual instanton. The abelian specialisation on the two-sphere gives index equal to the charge, which is the Chern number $1$ of the Hopf bundle.

The index density is a total derivative whose integral equals the instanton charge, and in the abelian case it is the framework's invariant $I_2=\vec E\cdot\vec B$. The operator, its gradient, the module it acts on, the adjoint representation and the group theory of the indices are all the algebra's; the theorem itself, the A-roof genus, the boundary eta invariant, the chirality of the spinor factor that grades the operator, and the identification of the counted modes with physical fermions are standard results transcribed. The one place where the framework is silent is the invariant form of the non-abelian density: it is computable in components with the matrix trace, but it is not an object of the biquaternion-norm invariant calculus, because that calculus is rank-two and single-slot while the density is traced over a second algebra factor.

Summary of Notation

Symbol Meaning
$\mathbb B=\mathbb C\otimes_{\mathbb R}\mathbb H$ Biquaternion algebra, $\cong M_2(\mathbb C)$
$e_t=1,\vec e_x,\vec e_y,\vec e_z$ Quaternion basis, $\vec e_\eta^2=-e_t$, $\vec e_x \vec e_y=\vec e_z$
$i$ Central scalar imaginary, $i^2=-1$
$\mathbb{B}_{\Psi},\mathbb{B}_{\Phi}$ Informational (Hermitian) and material (anti-Hermitian) sectors
$\mathfrak{su}(2)=\mathrm{span}_{\mathbb R}\{\vec e_x,\vec e_y,\vec e_z\}$ Compact gauge algebra inside $\mathbb{B}_{\Phi}$
$T_a=\tfrac12 \vec e_a$, $[T_a,T_b]=\varepsilon_{abc}T_c$ Normalised generators
$\Box_\Phi=\mathrm{Sc}((\tilde\nabla_\Phi)^{\natural}\tilde\nabla_\Phi)=\Delta_\Phi - \partial^2/\partial {r^{\prime}_t}^2$ d'Alembertian of the $\mathbb{B}$ frame; the scalar part of $(\tilde\nabla_\Phi)^{\natural}\tilde\nabla_\Phi$
$\Delta_\Phi=\partial^2/\partial R_x^2+\partial^2/\partial R_y^2+\partial^2/\partial R_z^2$ Laplacian in the three vector coordinates
$\tilde\nabla_\Phi=\sum_\mu e_\mu\,\partial/\partial R_\mu$ Biquaternionic gradient, acting from the left; $(\tilde\nabla_\Phi)^{\natural}=\sum_\mu\varepsilon_\mu e_\mu\,\partial/\partial R_\mu$
$\tilde D_R=\sum_\mu e_\mu D_\mu$ Twisted Dirac operator in the representation $R$; the gauge-covariant gradient
$D_\mu=\partial/\partial R_\mu+i\kappa\,\mathcal A_\mu$ Covariant derivative; $[D_\mu,D_\nu]=i\kappa F_{\mu\nu}$
$\tilde{Ξ}_{L}$, $\tilde{Ξ}_{R}$ The two values of the chirality subscript; $R$ is the plus value
$\xi$ Chirality of the frame: right multiplication by $-i\vec e_z$, whose two eigenspaces are $\mathbb B_L$ and $\mathbb B_R$
$G_M$ Chirality of the manifold's spinor factor; the grading the twisted operator reverses
$\tilde D_R^{\pm}$ The twisted operator on the $G_M=\pm1$ side; the two are adjoint
$n_\pm$ Dimensions of the kernel on the two sides of $G_M$
$\mathrm{ind}\,\tilde D_R=n_+-n_-$ Index of the Dirac operator
$A_{\mathrm{aro}}(TM)\wedge\mathrm{ch}(E_R)$ Index density; $A_{\mathrm{aro}}=1$ on the flat slice
$T(R)$ Dynkin index; $T(\text{defining})=\tfrac12$, $T(\text{adjoint})=2$
$R$ Representation label, defining (dim $2$) or adjoint (dim $3$)
$\kappa$ Coupling constant of the gauge field
$\mathcal F=d\mathcal A+i\kappa\,\mathcal A\wedge\mathcal A$ Field-strength two-form
$F_{\mu\nu}=\partial/\partial R_\mu\,\mathcal A_\nu-\partial/\partial R_\nu\,\mathcal A_\mu+i\kappa[\mathcal A_\mu,\mathcal A_\nu]$ Its components; $[D_\mu,D_\nu]=i\kappa F_{\mu\nu}$
$Ϙ=\frac{1}{8\pi^2}\int\mathrm{Tr}(F\wedge F)$ Topological (instanton) charge, $Ϙ\in\mathbb Z$; the trace in the defining representation
$\Omega_{\mathrm{CS}}$ Chern–Simons three-form; $d\Omega_{\mathrm{CS}}=\mathrm{Tr}(F\wedge F)$
$\eta(0)$, $h$ Eta invariant and boundary zero-mode count (APS)
$I_2=\vec E\cdot\vec B$ Framework's second invariant; index density in the abelian case
$\epsilon_{\mu\nu\rho\sigma}$ Alternating symbol, orientation density, written all-low; $\epsilon_{0123}=+1$ Lorentzian, $\epsilon_{4123}=+1$ Euclidean
$\langle\tilde F\rangle_{\natural}=\tilde F\tilde F^{\natural}$ Biquaternion norm; rank-two, single-slot, does not supply $\mathrm{Tr}(F\wedge F)$

Further Reading

  • Michael F. Atiyah and Isadore M. Singer, "The index of elliptic operators I", Annals of Mathematics 87 (1968) 484–530, for the index theorem and the index density.
  • Michael F. Atiyah, Vijay K. Patodi and Isadore M. Singer, "Spectral asymmetry and Riemannian geometry I", Mathematical Proceedings of the Cambridge Philosophical Society 77 (1975) 43–69, for the index theorem with boundary and the eta invariant.
  • Michael F. Atiyah, Nigel J. Hitchin and Isadore M. Singer, "Self-duality in four-dimensional Riemannian geometry", Proceedings of the Royal Society A 362 (1978) 425–461, for the instanton zero modes and the dimension formula.
  • Thomas Eguchi, Peter B. Gilkey and Andrew J. Hanson, "Gravitation, gauge theories and differential geometry", Physics Reports 66 (1980) 213–393, for the index density, the Chern character and the eta invariant in a physics setting.
  • Arthur L. Besse, Einstein Manifolds (Springer, 1987), for the A-roof genus and the differential-geometric background.
  • Michael F. Atiyah, Geometry of Yang–Mills Fields (Accademia Nazionale dei Lincei, 1979), for the index theorem in the gauge-theoretic form used here.
  • Roman Jackiw and Claudio Rebbi, "Solitons with fermion number $1/2$", Physical Review D 13 (1976) 3398–3409, for the fermion number of the instanton zero modes.
  • Gerard 't Hooft, "Symmetry breaking through Bell–Jackiw anomalies", Physical Review Letters 37 (1976) 8–11, for the anomalous divergence and the zero-mode count.
  • Mikio Nakahara, Geometry, Topology and Physics (CRC Press, 2nd ed. 2003), for the index theorem and the Chern–Weil construction in a form suited to the present notation.
  • Raoul Bott and Loring W. Tu, Differential Forms in Algebraic Topology (Springer, 1982), for the Chern character and the characteristic-class calculus used throughout.