Hawking radiation is a prediction of quantum field theory in curved spacetime, not of classical collapse alone. Consider a free massless Klein-Gordon field in the collapse vacuum for Hawking radiation: initially no incoming particles from past null infinity, with the short-distance state regular for freely falling observers through the forming event horizon. At late times use the approximately fixed exterior Schwarzschild metric; neglect backreaction during the observation interval. First set $G=c=\hslash =k_B=1$.

The exterior metric has $f(r)=1-2M/r$. Introduce the tortoise coordinate and outgoing and ingoing null coordinates,The surface gravity is $\kappa =1/(4M)$. The logarithm in $r_{\ast }$ makes a late escaping ray spend a long exterior coordinate time close to the event horizon. One of the regular outgoing Kruskal–Szekeres coordinates is proportional to $-e^{-\kappa u}$. Smooth propagation backwards through the collapsing body relates this regular coordinate to the incoming coordinate on past null infinity. If $v_H$ labels the last ray able to escape, the late-time Hawking exponential ray map is consequentlyHere the symbol $u(v)$ is the outgoing retarded time evaluated on the ray; the first equation is the leading asymptotic relation, with subleading corrections that vanish at late times. The constant $C$ depends on collapse history, whereas the exponential rate is determined by the final surface gravity. The relation follows from a nondegenerate Killing horizon; a linear ray map would not have the same effect.

An outgoing positive-frequency mode at future null infinity behaves as $p_{\omega }\sim e^{-i\omega u}/\sqrt{4\pi \omega }$, where $\omega >0$ is measured with respect to the asymptotic stationary time. In the near-horizon geometric optics approximation its backward-propagated form on past null infinity isThis is not purely positive frequency in $v$. Its decomposition into incoming positive-frequency modes $f_{{\omega }^{\prime }}\sim e^{-i{\omega }^{\prime }v}/\sqrt{4\pi {\omega }^{\prime }}$ and their complex conjugates is a Bogoliubov transformation:The coefficients follow from the Klein-Gordon inner product, or equivalently from the Fourier transform with the indicated flux normalization. Writing $y=v_H-v$ and $a=\omega /\kappa$, the two Fourier integrals, after a convergence regulator, areThe equality is the defining integral of the Gamma function, continued to these arguments from the right half-plane. Up to phases of unit modulus, $\alpha$ is $\sqrt{{\omega }^{\prime }/\omega }{I}_{+}/(2\pi )$ and $\beta$ is $\sqrt{{\omega }^{\prime }/\omega }{I}_{-}/(2\pi )$. Taking $ϵ\downarrow 0$ on the principal branches, the arguments of $ϵ\pm i{\omega }^{\prime }$ tend to $\pm \pi /2$. Thus $|I_{+}|\propto e^{\pi a/2}$ and $|I_{-}|\propto e^{-\pi a/2}$, giving the thermal ratio of Hawking Bogoliubov coefficientsIn particular, using $|\Gamma (1+ia){|}^2=\pi a/\sinh (\pi a)$,The collapse-dependent phases and $C$ disappear from the ratio. The canonical identities for a bosonic Bogoliubov transformation state that the positive-frequency norm minus the negative-frequency norm is one for a normalized outgoing mode. Combining that identity with the thermal ratio yields the late-time mean occupationThese are the Bose-Einstein distribution and the Hawking temperature. Strictly, monochromatic modes have delta-function normalization, and $\int d{\omega }^{\prime }|{\beta }_{\omega {\omega }^{\prime }}{|}^2$ contains the infinite-duration divergence $\int d{\omega }^{\prime }/{\omega }^{\prime }$. Normalized wave packets with a finite frequency band and a finite retarded-time window remove this artifact. For sufficiently late packets the occupation is the frequency average of the displayed thermal factor, tending to it for a narrow band. Fermionic fields instead give the corresponding denominator $e^{2\pi \omega /\kappa }+1$ because their canonical normalization uses anticommutators.

Four-dimensional propagation adds an important qualification to the word thermal. Separation of the massless Klein-Gordon equation into spherical harmonics gives an exterior radial scattering potentialOnly a fraction ${\Gamma }_{\ell }(\omega )$ of the outgoing mode is transmitted to future null infinity; this is the greybody factor. The scalar particle flux there isThe factor $2\ell +1$ counts the angular modes, and multiplication by $\omega$ gives the spectral energy flux. An observer at infinity therefore receives an outgoing thermal population filtered by curvature scattering, rather than an exact unfiltered blackbody spectrum. In physical units, for mass $M_{\mathrm{phys}}$,There is no incoming thermal bath in the collapse vacuum for Hawking radiation. The outgoing flux has positive Killing energy; its counterpart inside the event horizon can carry negative Killing energy. Including the expectation value of the stress-energy tensor in the gravitational evolution allows the mass to decrease. The fixed-background derivation describes the slowly evolving semiclassical regime and does not determine the endpoint of evaporation. Quantum expectation values need not satisfy the classical null energy condition, so this loss of area is not a contradiction of the classical second law of black-hole mechanics. The essential late-time universality is the combination of a regular initial quantum state and the exponential redshift at the newly formed event horizon.

A Lagrangian submanifold of a $2n$-dimensional symplectic manifold $(M,\omega )$ is an $n$-dimensional submanifold $L$ such that the pullback of $\omega$ to $L$ is zero. Equivalently, each tangent space $T_{x}L$ is its own symplectic orthogonal. The dimension condition makes it maximally isotropic: in a nondegenerate alternating vector space no isotropic subspace has dimension greater than $n$.

In geometric quantization, a line bundle with a connection on a vector bundle whose curvature is proportional to $\omega /\hslash$ provides the initial quantum-state space. A real polarization foliates phase space locally by Lagrangian leaves; requiring sections to be covariantly constant along these leaves reduces their independent variables by half. The curvature restricts to zero on a Lagrangian leaf, allowing local flat sections. Globally its holonomy can impose the familiar Bohr-Sommerfeld quantization condition. Thus Lagrangian submanifolds describe choices of quantum configuration data and the support of semiclassical states, rather than requiring simultaneous sharp values of all position and momentum coordinates.

For the thermodynamic application, regard $(S,V,N)$ as base coordinates $q^i$ and $(T,-P,\mu )$ as fiber coordinates $p_i$ on $T^{\ast }{\mathbb{R}}^3$. The canonical one-form on a cotangent bundle and the position-first symplectic form areThe negative sign in the pressure coordinate is essential. A smooth potential $M(S,V,N)$ determines the embeddingwhere the last three coordinates are ordered as $(T,-P,\mu )$: explicitly $T=M_S$, $P=-M_V$, and $\mu =M_N$, so the second fiber coordinate is $-P=M_V$. Accordingly, in the actual $p_i$ coordinates the graph isProjection to the base is the identity, so this is an embedded three-dimensional submanifold, even if the Hessian of $M$ is degenerate. The first law of thermodynamics gives ${\iota }_M^{\ast }\theta =dM$. HenceIt has half the dimension of the ambient cotangent bundle and is therefore Lagrangian. This proves the thermodynamic potential as a Lagrangian graph construction for every smooth choice of $M$, locally on any domain on which the potential is defined. No invertibility, extensivity or stability assumption is needed for this geometric conclusion.

In components the same vanishing follows from equality of mixed derivatives. The thermodynamic conjugate variables satisfyThese cancel the three coefficients of the pulled-back two-form. They are the integrability relations behind the local thermodynamic potential, and demonstrate directly why the first-law graph is Lagrangian.

A Poisson manifold is a smooth manifold with a smooth antisymmetric bivector $\pi =\frac{1}{2}{\pi }^{ij}{\partial }_i\wedge {\partial }_j$ whose function bracket $\{f,g\}={\pi }^{ij}({\partial }_{i}f)({\partial }_{j}g)$ satisfies the Jacobi identity. The bracket is automatically a derivation in each argument. A symplectic manifold has a nondegenerate differential two-form $\omega$ satisfying $d\omega =0$, that is, a closed differential form; nondegeneracy forces its dimension to be even. It determines a nondegenerate Poisson structure. We use matrix convention $\pi =-{\omega }^{-1}$, so that ${\iota }_{X_H}\omega =dH$ corresponds to $X_H(f)=\{f,H\}$.

Examples of genuinely nonsymplectic Poisson geometries include the zero bracket on ${\mathbb{R}}^3$ and the Lie-Poisson bracket $\{x_i,x_j\}={ϵ}_{ijk}{x}_k$ on ${\mathbb{R}}^3$. Their three-dimensional underlying manifold cannot carry a nondegenerate two-form. In the latter case the Poisson rank is two away from the origin and zero at the origin, with spheres and the origin as symplectic leaves. Even in even dimension a given Poisson structure need not come from a symplectic form: $\pi =x{\partial }_x\wedge {\partial }_y$ on ${\mathbb{R}}^2$ satisfies Jacobi but vanishes along $x=0$. The statement concerns this degenerate structure, not a claim that ${\mathbb{R}}^2$ admits no other symplectic structure.

For any candidate bivector, applying the Jacobi identity to coordinate functions gives the necessary conditionIt is sufficient as well: expanding $\{f,\{g,h\}\}$ and its cyclic companions cancels all terms with second derivatives by antisymmetry of $\pi$. What remains is $J^{ijk}({\partial }_{i}f)({\partial }_{j}g)({\partial }_{k}h)$. This is the coordinate Jacobi condition for a Poisson bivector.

For a symplectic form, differentiate $\omega \pi =-I$ to obtain ${\partial }_{\ell }\pi =\pi ({\partial }_{\ell }\omega )\pi$. Substituting this into the cyclic expression givesbecause $d\omega =0$. Thus the inverse symplectic structure always satisfies the Jacobi identity. Conversely, for a nondegenerate Poisson structure, multiplying by three inverse matrices proves closure of its associated two-form.

In momentum space, choose Hamiltonian $H(\mathbf{p})=E(\mathbf{p})$ and the constant Poisson bivectorIts Poisson bracket is $\{f,g\}=e\mathbf{B}\cdot (\nabla f\times \nabla g)$. The Hamiltonian vector field therefore givesexactly the prescribed motion. The sign here uses the signed charge $e$ as printed; an electron's physical charge may be negative. Since the magnetic field is uniform, every derivative of ${\pi }^{ij}$ vanishes, so every term in $J^{ijk}$ is zero. Jacobi holds independently of the dispersion function $E$. This is the momentum-space Poisson structure for a uniform magnetic field.

For $e\mathbf{B}\ne 0$ this bivector has rank two and $C(\mathbf{p})=\mathbf{B}\cdot \mathbf{p}$ is a Casimir function of a Poisson manifold: $\{C,f\}=0$ for every $f$. Its symplectic leaves are the planes $C=c$. For axes with $\mathbf{B}=B_0{\mathbf{e}}_3$, the leaf form is $\omega =(e{B}_0{)}^{-1}d{p}_1\wedge d{p}_2$ in our inverse-sign convention. Also $\dot{E}=\{E,E\}=0$, so trajectories on a Fermi surface lie on its intersections with these planes, typically curves. The energy surface is invariant for this Hamiltonian, but need not be a Poisson submanifold; the bracket has been defined on the whole momentum space. If $e\mathbf{B}=0$, the Poisson structure and motion vanish.

Identify the ambient real four-dimensional vector space with $2\times 2$ real matrices byFor $A,B\in SL(2,\mathbb{R})$, the linear action $X\mapsto AX{B}^{-1}$ preserves this quadratic form. It gives a group homomorphism to the special orthogonal group of signature $(2,2)$; its determinant on matrix space is $(\det A{)}^2(\det B{)}^{-2}=1$. The group $SL(2,\mathbb{R})$ is connected: an oriented Gram–Schmidt factorization writes every element as a rotation times an upper triangular matrix with positive diagonal and determinant one, and each factor varies in a connected space. Thus the image lies in the identity component $S{O}_0(2,2)$.

If $AX{B}^{-1}=X$ for every $X$, taking $X=I$ gives $A=B$. A matrix commuting with every real $2\times 2$ matrix is scalar, and its determinant-one condition then gives $A=B=\pm I$. The kernel is therefore exactly $\{(I,I),(-I,-I)\}$. Infinitesimally the action is $X\mapsto aX-Xb$, with $a,b$ traceless. If this vanishes for every $X$, then $a=b$ and it is scalar and traceless, hence zero. The differential is injective; source and target Lie algebras each have dimension six, so it is an isomorphism. The image is therefore an open subgroup of the connected target, which must be the entire target. ConsequentlyThe finite quotient is by the diagonal central subgroup, not by a sign in just one factor. The connected-component subscript is a genuine qualification of the source's displayed formula: the full $SO(2,2)$ is disconnected and cannot be isomorphic to a quotient of the connected product.

Let the anti-de Sitter radius be $\ell >0$. Its quadric is $-u^2-v^2+x^2+y^2=-{\ell }^2$, or $\det X={\ell }^2$. Thus the map $X\mapsto g=X/\ell$ identifies this quadric smoothly and globally with $SL(2,\mathbb{R})$. This is the group model of three-dimensional anti-de Sitter spacetime. The induced Lorentzian metric is bi-invariant, since the two matrix multiplication actions preserve the ambient quadratic form. At the identity, tangent matrices $a,b$ are traceless andIndeed, for a traceless matrix $a$, $\det a=-\frac{1}{2}\mathrm{Tr}(a^2)$, and polarizing $-\det (\ell a)$ gives this metric. For ${\mathfrak{sl}}_2(\mathbb{R})$ the Killing form is $K(a,b)=4\mathrm{Tr}(ab)$: in the basis $H=\mathrm{diag}(1,-1)$, $E=\left(\begin{array}{cc}0& 1\\ 0& 0\end{array}\right)$ and $F=\left(\begin{array}{cc}0& 0\\ 1& 0\end{array}\right)$, the brackets are $[H,E]=2E$, $[H,F]=-2F$, $[E,F]=H$, giving $K(H,H)=8$, $K(E,F)=4$ and the remaining independent pairings zero. Therefore $g_{\ell }={\ell }^{2}K/8$.

The canonical torsion-free connection on a Lie group is the Levi-Civita connection of this bi-invariant pseudo-Riemannian metric. Its Ricci tensor, already calculated from the double bracket, isSubstitution into the vacuum Einstein field equations givesThus the group metric solves the three-dimensional vacuum equations with negative cosmological constant.

The antipodal map $X\mapsto -X$ is $g\mapsto -g$, multiplication by the central element $-I$. Hence the antipodal quotient is $PSL(2,\mathbb{R})=SL(2,\mathbb{R})/\{\pm I\}$, with the same local curvature and cosmological constant. To see the global time issue, useThe induced metric isOn the quadric $t$ has period $2\pi$; curves with fixed $\rho ,\phi$ are closed timelike curves. The antipodal identification is $(t,\phi )\sim (t+\pi ,\phi +\pi )$, and does not eliminate this problem. At $\rho =0$, even the half-period time shift closes after the antipodal quotient.

If global time is not identified, take the universal cover, with $t\in \mathbb{R}$. The topology of $SL(2,\mathbb{R})$ is $S^1\times {\mathbb{R}}^2$, as follows from the same rotation–triangular decomposition, so the universal cover is diffeomorphic to ${\mathbb{R}}^3$ and carries the lifted group structure $\widetilde{SL(2,\mathbb{R})}$. The metric and Einstein equation are unchanged locally. Here $t$ is a global time function with timelike gradient, excluding closed causal curves. Unwrapped anti-de Sitter spacetime is the universal covering group, rather than the original embedded quadric or its antipodal quotient; retaining the antipodal time-shift identification would wrap time again.

The given equation is a Maurer-Cartan equation, with the sign appropriate to the chosen invariant coframe. Write $\lambda =({\lambda }_{ij})$ as an antisymmetric matrix of one-forms, so $d\lambda =\lambda \wedge \lambda$. Using the graded product rule for the exterior derivative,The two sums coincide after exchanging their dummy indices $k,\ell$. This is also associativity of matrix-valued wedge products of differential forms.

To make the relation to the Jacobi identity explicit, let ${\theta }^a$ be any independent coframe and write $d{\theta }^a=-\frac{1}{2}{c}^a{}_{bc}{\theta }^b\wedge {\theta }^c$. Its dual invariant vector fields satisfy $[E_b,E_c]=c^a{}_{bc}{E}_a$, by the exterior derivative of a one-form evaluated on vector fields. Evaluating $d^2{\theta }^a=0$ on three basis fields gives the cyclic coefficient identityThese are exactly the components of $[[E_b,E_c],E_d]+[[E_c,E_d],E_b]+[[E_d,E_b],E_c]=0$. Thus the structure constants obey the Jacobi identity.

For the real four-dimensional case, abbreviate $a={\lambda }_{12}$, $b={\lambda }_{23}$, $c={\lambda }_{31}$, $d={\lambda }_{43}$, $e={\lambda }_{41}$, $f={\lambda }_{42}$; here $d$ without an argument is a one-form, whereas $d(\cdot )$ denotes the exterior derivative. Directly expanding the given equations givesDefine ${\alpha }_1^{\pm }=a\pm d$, ${\alpha }_2^{\pm }=b\pm e$, and ${\alpha }_3^{\pm }=c\pm f$. For example,The minus combination and the other two cyclic combinations giveThe six forms are independent, since the original pairs are their half-sums and half-differences. Their dual vector fields $E_i^{+},E_i^{-}$ consequently obeyEach triple is a three-dimensional Lie algebra ideal with the SO(3) Lie algebra brackets; the two ideals commute and span the original six-dimensional algebra. This proves the real four-dimensional rotation algebra splittingThis is a Lie-algebra statement. Globally $SO(4)$ is a central quotient of $SU(2)\times SU(2)$; a direct product of the corresponding $SO(3)$ groups is not asserted.

A fiber bundle with base $B$, fiber $F$ and structure group $G$ acting on $F$ from the left consists of a projection $\pi :E\to B$ and local trivializations ${\varphi }_i:{\pi }^{-1}(U_i)\to U_i\times F$. On overlaps they satisfyThe smooth transition functions are valued in $G$. A principal bundle is the case $F=G$ with the left multiplication action on itself; its transition functions multiply the group coordinate from the left.

In each principal chart define $(x,h)\cdot a=(x,ha)$. This defines a global right action because $g_{ij}(x)(ha)=(g_{ij}(x)h)a$: coordinate changes commute with right multiplication. It is free, and acts transitively on each fiber. Thus the claimed right action follows from the left transition convention rather than being an additional choice.

If $s:B\to P$ is a global section of a fiber bundle, define $\Phi :B\times G\to P$ by $\Phi (x,a)=s(x)a$. Freeness and transitivity give exactly one $a$ for every point in the fiber, so $\Phi$ is bijective and equivariant. In a local trivialization where $s(x)$ has group coordinate $h_i(x)$, the inverse group coordinate is $h_i(x{)}^{-1}h$, proving smoothness of the inverse. Hence $\Phi$ is a principal-bundle trivialization. Conversely, the identity coordinate in a product trivialization gives $s(x)=\Phi (x,e)$. ThereforeThis is the principal bundle trivialization by a global section criterion. It is special to principal bundles; a section alone need not trivialize an arbitrary fiber bundle.

For the quotient examples, assume the Lie subgroups are closed when smooth homogeneous-space statements are intended. The map $G\to G/H_1$ is a principal bundle with structure group $H_1$ and right action $g\cdot h=gh$; $G/H_1$ is a homogeneous space for left multiplication by $G$. Similarly, $G\to H_2\backslash G$ has free transitive left $H_2$ action on its fibers. With the usual principal right-action convention one writes $g\cdot h=h^{-1}g$. The quotient $H_2\backslash G$ is a homogeneous space for right multiplication by $G$. These quotients are groups only when the respective subgroups are normal. Nonclosed immersed Lie subgroups can produce non-Hausdorff quotients, so an arbitrary Lie subgroup does not automatically give a smooth Hausdorff base.

Left $H_2$ and right $H_1$ multiplication commute, so taking their orbit equivalence relations in either order gives the same double coset setThis equality does not assert that the double quotient is a manifold or that either stage is a principal bundle. The $H_2\times H_1$ action $(a,b):g\mapsto ag{b}^{-1}$ can have stabilizers: they correspond to $H_2\cap g{H}_1{g}^{-1}$. If the induced action is free and proper, its quotient is a smooth principal-bundle quotient; without those hypotheses orbit dimensions can change. For example, with $G=SO(3)$ and both subgroups the rotations fixing the third axis, double cosets are parametrized by $e_3\cdot g{e}_3\in [-1,1]$. Interior and endpoint stabilizers differ, so this is not a uniform free principal-bundle construction.

For a principal bundle $P$ and a left $G$ action on $F$, the associated bundle iswith projection $[p,f]\mapsto \pi (p)$. A local principal section gives coordinates $[s_i(x),f]\leftrightarrow (x,f)$, and its changes give precisely the left action of the principal transition functions on $F$. A linear action on a vector space gives an associated vector bundle.

For a metric of signature $(r,s)$ on an $n$-manifold, an orthonormal frame at $x$ is an isometry $u:{\mathbb{R}}^{r,s}\to T_{x}M$. The orthonormal frame bundle $O(M,g)\to M$ has structure group $O(r,s)$, acting on the right by $u\cdot a=u\circ a$. Its associated bundle for the standard representation is the tangent bundle, sinceis well defined: $(ua)(v)=u(av)$. Every tangent vector has exactly one such associated equivalence class, and local frame coordinates show that this is a vector-bundle isomorphism. The full frame bundle works similarly with $GL(n,\mathbb{R})$.

Finally, choose a basis of the Lie algebra of a Lie group $G$. Its left translates are a global frame, yielding the explicit tangent bundle trivializationThis proves parallelization of a Lie group by left translations and triviality of the full frame bundle. Choosing an inner product at the identity and translating it gives a left-invariant Riemannian metric; an orthonormal basis translates to a global orthonormal frame. The global section criterion therefore trivializes the corresponding orthonormal frame bundle as well. For any other positive-definite metric on $G$, smoothly applying the Gram-Schmidt process to this global frame gives the same conclusion. Thus the tangent and frame bundles of every Lie group are trivial; the orthonormal statement is with respect to a chosen metric as specified above.

For a Lie group $G$, Left translation on a Lie group and right translation on a Lie group are the diffeomorphisms $L_g(h)=gh$ and $R_g(h)=hg$. Their differentials identify the tangent space at the identity with every tangent space. Translating a basis $e_a$ of the Lie algebra gives global left-invariant vector fields $E_a(g)=(d{L}_g{)}_e{e}_a$ and right-invariant vector fields $F_a(g)=(d{R}_g{)}_e{e}_a$. Left and right translations commute. If $[e_a,e_b]=c^c{}_{ab}{e}_c$, thenThe opposite sign for the right-invariant frame is essential.

Define an affine connection ${\nabla }^{-}$ by declaring the $E_a$ parallel. Explicitly, ${\nabla }_X^{-}(f^a{E}_a)=X(f^a)E_a$. Its curvature form of a connection vanishes, since successive derivatives of the component functions satisfy the vector-field commutator identity. Its torsion tensor isSimilarly, declaring the $F_a$ parallel gives another flat affine connection ${\nabla }^{+}$, with $T^{+}(F_a,F_b)=-[F_a,F_b]$. In a left-invariant frame the latter connection obeysTo see this, express $E_a=F_{{\mathrm{Ad}}_g{e}_a}$. Along $g\exp (t{e}_b)$, differentiation of ${\mathrm{Ad}}_{g\exp (t{e}_b)}{e}_a$ gives ${\mathrm{Ad}}_g[e_b,e_a]$, producing the displayed bracket. Consequently $T^{+}(X,Y)=[X,Y]$ for left-invariant $X,Y$. These are the flat translation connections on a Lie group. They have opposite torsion; if $G$ is Abelian, both torsion tensors vanish and the two connections coincide. The phrase “with torsion” does not mean nonzero torsion for Abelian groups.

The average ${\nabla }^0=({\nabla }^{-}+{\nabla }^{+})/2$ is again an affine connection and satisfies ${\nabla }_X^{0}Y=\frac{1}{2}[X,Y]$ for left-invariant fields. Its torsion tensor is zero. With the curvature convention $R(X,Y)Z={\nabla }_X{\nabla }_{Y}Z-{\nabla }_Y{\nabla }_{X}Z-{\nabla }_{[X,Y]}Z$, the Jacobi identity givesDefine the Ricci tensor by $\mathrm{Ric}(X,Y)=\mathrm{Tr}(Z\mapsto R(Z,X)Y)$. Since $[[Z,X],Y]={\mathrm{ad}}_Y{\mathrm{ad}}_{X}Z$, this giveswhere $K(X,Y)=\mathrm{Tr}({\mathrm{ad}}_X{\mathrm{ad}}_Y)$ is the Killing form. Thus the canonical torsion-free connection on a Lie group is generally curved. Its exact flatness criterion is $[[\mathfrak{g},\mathfrak{g}],\mathfrak{g}]=0$: Abelian and two-step nilpotent groups are exceptions. A vanishing Ricci tensor alone does not prove flatness.

For a semisimple Lie algebra, the Killing form is nondegenerate. Its adjoint invariance gives $K([X,Y],Z)+K(Y,[X,Z])=0$, so translating it defines a bi-invariant pseudo-Riemannian metric $g_K$. For left-invariant fields,A metric-compatible torsion-free connection is uniquely the Levi-Civita connection; therefore ${\nabla }^0$ is its Levi-Civita connection. Moreover,The special feature is a canonical nondegenerate metric and an Einstein manifold structure. The metric need not be positive definite: noncompact semisimple groups generally require indefinite signature. For compact semisimple groups, using $-K$ gives a bi-invariant Riemannian metric with $\mathrm{Ric}=\frac{1}{4}{g}_{-K}$ instead.

The exterior algebra of a vector space $V$ is ${\Lambda }^{\ast }{V}^{\ast }={⨁}_{r=0}^{\dim V}{\Lambda }^r{V}^{\ast }$, whose degree-$r$ elements are alternating $r$-linear forms. Its exterior product satisfies $\alpha \wedge \beta =(-1{)}^{rs}\beta \wedge \alpha$ for degrees $r,s$. For example, one-forms anticommute, so $d{x}^i\wedge d{x}^i=0$. A differential form of degree $r$ on a smooth manifold is a smooth section of ${\Lambda }^r{T}^{\ast }M$; in coordinates it is an alternating sum of products of $r$ coordinate one-forms. Its exterior derivative increases degree by one, satisfies $d^2=0$, and obeysThe coordinate definition and pullback of a differential form are compatible, making differential forms suitable for coordinate-independent integration.

An orientation and a nondegenerate metric tensor define the Hodge star operator by $\alpha \wedge \ast \beta =⟨\alpha ,\beta ⟩{\mathrm{vol}}_g$ for equal-degree forms. It maps degree $r$ to degree $d-r$. If the metric has $s$ negative directions, its square is ${\ast }^2=(-1{)}^{r(d-r)+s}$. Thus Hodge duality turns antisymmetric tensors into complementary-degree tensors, using the metric and orientation rather than merely the underlying vector space. The Generalized Stokes theorem says ${\int }_{M}d\eta ={\int }_{\partial M}\eta$, with the induced boundary orientation, for appropriate smooth forms and compactness or support assumptions. It generalizes the fundamental theorem of calculus, circulation version of Stokes theorem, and divergence theorem.

For the field calculation, choose signature $(-,+,+)$ and a fixed orientation on Minkowski spacetime. Write $a=\delta A$, so $\delta F=da$. Symmetry of the metric pairing and the Hodge star operator gives $\delta (F\wedge \ast F)=2da\wedge \ast F$. Applying the exterior derivative product rule,Since $F\wedge a=a\wedge F$, variation of the action isFix the boundary data or use compactly supported variations. Arbitrary interior one-forms $a$ then imply the bulk field equationWith $B=\ast F$, the Lorentzian Hodge star operator squares to $-1$ on both one- and two-forms, hence $F=-\ast B$. Thus the Maxwell–Chern–Simons one-form equation becomesThese two displayed equations are equivalent. Also $dF=0$ gives $d\ast B=0$. For $k\ne 0$ the latter follows from applying $d$ to the field equation; for $k=0$ it must be retained as the identity inherited from $F=dA$. Using the opposite metric-sign convention changes the corresponding ${\ast }^2$ sign; the signature-independent equation is $d\ast F+kF=0$.

Under the Abelian gauge transformation $A\mapsto A+d\Lambda$, the curvature $F=dA$ and the one-form $B$ are unchanged because $d^2\Lambda =0$. The field equation is therefore exactly gauge-invariant. The kinetic term is unchanged, while the other term changes byThe last equality uses $dF=0$ and the Generalized Stokes theorem. Thus the action density changes by a total derivative and the integrated action can change on a boundary. The source's statement that the action is not invariant needs this qualification: on a boundaryless spacetime, or when the boundary integral vanishes, the integrated action is invariant too. No such boundary restriction is needed for invariance of the bulk field equation.