---
title: Generally-Covariant Quantum Mechanics
url: https://www.emergentmind.com/topics/generally-covariant-quantum-mechanics
type: topic
---

# Generally-Covariant Quantum Mechanics

to=arxiv_search.search  彩神争霸下载  天天中彩票不ిథింగ code?
to=arxiv_search.search 山大发cript code
Generally-covariant quantum mechanics denotes a family of formulations in which quantum laws are stated without reliance on a preferred coordinate system, foliation, clock choice, or fixed geometric frame. In this literature, covariance is realized in several distinct but related ways: as diffeomorphism-covariance of tensorial structures on manifolds of states, manifest Lorentz covariance of many-time wave functions and spacetime probabilities, local covariance of algebraic quantum field theory on curved spacetimes, splitting-independent dynamics of measures on spaces of worldlines, covariance of multi-time quantum states under \(C^*\)-algebra isomorphisms, and covariance under quantum-group changes of frame in settings where geometric reference structures are themselves quantized [1302.0194][1205.1992][1904.06944][2004.07015][2311.00162][2412.05997]. The common theme is that the physical content is encoded in relational or intrinsic structures, while coordinate descriptions, slicings, or frame choices serve only as representations.

## 1. Concepts and transformation classes

The expression “general covariance” is not used uniformly across this field. In geometric formulations on state space, it means that the equations of motion and observable relations are written as tensorial statements whose form is preserved under arbitrary diffeomorphisms of the relevant manifold. In relativistic many-time and Bohmian approaches, it means manifest Lorentz covariance together with foliation independence or reparametrization invariance. In locally covariant quantum field theory, it means naturality under spacetime embeddings and causal localization. In discrete-time operator formulations, it means covariance under per-time-step \(*\)-isomorphisms of observable algebras. In quantum-reference-frame models, it means invariance under quantum-group transformations acting jointly on system and geometry [1302.0194][1205.1992][1904.06944][2311.00162][2412.05997].

| Formulation | Primary object | Covariance notion |
|---|---|---|
| Tensorial state-space mechanics | \((\Lambda,G)\), \(g_{FS}\), \(\omega_{FS}\), \(J_{FS}\) | arbitrary diffeomorphisms of the state manifold |
| Many-time relativistic mechanics | \(\psi(x_1,\ldots,x_n)\), currents, worldlines | Lorentz covariance and reparametrization invariance |
| Locally covariant AQFT measurement | \(\Theta\), \(\varepsilon_\sigma\), \(\mathsf J_\sigma\) | covariance under spacetime embeddings and causal order |
| \(N\)-particle measure dynamics | \(\mu_t\), causal couplings, \(\mathbb P\) on worldlines | splitting independence |
| Quantum states over time | canonical multi-time operator \((E_1,\ldots,E_n)\star\rho\) | local \(*\)-isomorphisms at each time step |
| Covariant causal sets | path amplitudes, \(D_k\), \(\mu_k\) | intrinsic order-geometric covariance |
| Doubly quantum mechanics | geometry states, \(P(\uparrow)_\sigma\) | \(SU_q(2)\)-covariance |

A recurring distinction is that between covariance and invariance. In the tensorial state-space formulation, covariance means that pulled-back or pushed-forward tensors satisfy the same structural identities after any diffeomorphism, whereas invariance requires the diffeomorphism to leave the tensors themselves unchanged. The same distinction reappears elsewhere: a formulation may be observer-independent without treating every change of representation as a physical symmetry. This suggests that generally-covariant quantum mechanics is best understood as a class of representation-independent reformulations rather than a single doctrine of symmetry [1302.0194][2004.07015].

## 2. Geometric mechanics on spaces of states and phase-spacetime

One major program replaces operator algebra by differential-geometric structures on a manifold of states. In the tensorial formulation, pure states form the complex projective Hilbert manifold \(P(\mathcal H)\), equipped with the Fubini–Study Kähler triple \((g_{FS},\omega_{FS},J_{FS})\), while mixed states form the convex body \(S\) of density operators embedded in \(u^*(\mathcal H)\). Observables are represented by expectation-value functions
\[
e_A(p)=\mathrm{Tr}(pA), \qquad A=A^\dagger .
\]
The commutator and Jordan product are encoded by contravariant tensor fields \(\Lambda\) and \(G\) through
\[
\Lambda(de_A,de_B)(p)=\mathrm{Tr}\!\big(p[A,B]\big),
\qquad
G(de_A,de_B)(p)=\mathrm{Tr}\!\big(p(AB+BA)\big).
\]
On the pure-state submanifold, \(\Lambda\) is the inverse of the Fubini–Study symplectic form and \(G\) induces the Fubini–Study metric; on all of \(S\), which is stratified by rank, the same tensorial encoding persists. Heisenberg evolution becomes Hamiltonian flow on the state manifold:
\[
\frac{d}{dt}e_A=\{e_A,e_H\}=\Lambda(de_A,de_H)=e_{\frac{1}{i\hbar}[A,H]} .
\]
A central intrinsic criterion states that a function \(F\) represents a physical observable if and only if its Hamiltonian vector field preserves \(G\),
\[
\mathcal L_{X_F}G=0.
\]
The spin-\(\tfrac12\) example makes the construction explicit: the state space is the Bloch ball, pure states form the sphere \(P(\mathbb C^2)\cong \mathbb CP^1\cong S^2\), and the Poisson tensor reproduces angular-momentum brackets \(\{Y_k,Y_\ell\}=\epsilon_{k\ell m}Y_m\) [1302.0194].

In this framework, general covariance means precisely that the fundamental equations are tensorial. Under a diffeomorphism \(\phi\), covariant tensors pull back, contravariant tensors push forward, and identities such as
\[
\omega(X,Y)=g(X,JY), \qquad \{F,G\}=\Lambda(dF,dG), \qquad \iota_{X_F}\omega=dF
\]
retain their form. Not every diffeomorphism is a physical symmetry; the physical content is tied to the specific tensors \(\Lambda\), \(G\), and, on \(P(\mathcal H)\), the Kähler triple. The gain is a coordinate-free formulation in which expectation values, Born probabilities, transition probabilities, and Heisenberg derivations are all encoded geometrically [1302.0194].

A distinct geometric reformulation uses contact geometry rather than symplectic geometry. Here the basic arena is an odd-dimensional “phase-spacetime” \((Z,\alpha)\), a strict contact manifold with \(\alpha\wedge(d\alpha)^{\wedge n}\neq0\). Classical dynamics is governed by the action
\[
S[\gamma]=\int_\gamma \alpha,
\]
whose extremals are Reeb trajectories. Quantization is recast as the flatness of an operator-valued connection
\[
\nabla=d-\frac{i}{\hbar}\widehat\Omega(z,\hat s),
\qquad
\nabla^2=d\widehat\Omega-\frac{i}{\hbar}\widehat\Omega\wedge\widehat\Omega=0 .
\]
Physical quantum states are covariantly constant sections satisfying \(\nabla\Psi=0\). For a standard time-dependent Hamiltonian contact form \(\alpha=p_A\,dq^A-H(q,p,t)\,dt\), the covariant equation reduces to the usual Schrödinger equation after choosing a clock, but the underlying formulation treats time on the same footing as the other coordinates of \(Z\). The same formalism yields a covariant generalization of Wigner functions as overlaps of parallel-transported sections [1709.04557].

Taken together, these geometric approaches replace external coordinatizations by intrinsic structures: tensor fields on state space in one case, and a flat Hilbert-bundle connection over phase-spacetime in the other. A plausible implication is that “quantization” itself is being reframed from an operator prescription into a statement about geometric compatibility conditions [1302.0194][1709.04557].

## 3. Relativistic covariance, many-time wave functions, and multi-particle causality

A second line of development formulates quantum theory directly on spacetime rather than on a configuration space equipped with an external time parameter. In the many-time formalism, the basic object for \(n\) particles is a wave function
\[
\psi(x_1,\ldots,x_n),
\]
with \(x_a\in\mathbb R^{1,3}\), and the fundamental probabilistic postulate is the spacetime measure
\[
dP=|\psi(x_1,\ldots,x_n)|^2\,d^4x_1\cdots d^4x_n .
\]
For spinless particles, \(\psi\) satisfies \(n\) Klein–Gordon equations, one for each argument, and the conserved currents
\[
j_a^\mu=\frac{i}{2}\psi^*\!\overleftrightarrow{\partial_a^\mu}\!\psi
\]
obey \(\partial_{a\mu}j_a^\mu=0\). For Dirac particles, the formalism uses \(n\) Dirac equations together with conserved multi-currents; for spin-1 fields, the construction proceeds analogously. In Bohmian versions, particle worldlines satisfy
\[
\frac{dX_a^\mu}{ds}=V_a^\mu=\frac{j_a^\mu}{|\psi|^2},
\]
where \(s\) is an unobservable scalar parameter. Because the laws are written in terms of Lorentz vectors and scalar probability densities, the formulation is manifestly relativistic-covariant and does not require a preferred foliation. The same framework extends to quantum field theory by representing states as wave functions of infinitely many spacetime coordinates and interpreting creation and annihilation as effective phenomena produced by entanglement with detectors [1205.1992].

This relativistic program is paired, in a different work, with a measure-theoretic theory of \(N\)-particle dynamics on a globally hyperbolic spacetime \(M\). Given a Geroch–Bernal–Sánchez splitting \(M\simeq\mathbb R\times\Sigma\), one defines the \(N\)-particle configuration spacetime \(\mathcal M_{(N)}=\mathbb R\times\Sigma^N\) and an \(N\)-particle causal order by requiring each component worldline to be causal in \(M\). Causality is then lifted from points to Borel probability measures: \(\mu\preceq\nu\) when there exists a coupling \(\eta\) supported on the \(N\)-particle causal relation \(J^+_{(N)}\). Several equivalent characterizations are available, including inequalities for causal futures and bounded time functions. The central theorem states that a time-indexed family of measures \(t\mapsto \mu_t\), each supported on \(\Sigma_t^N\), is causal if and only if there exists a probability measure \(\mathbb P\) on the Polish space \(\Gamma_{(N)}^I\) of \(N\)-particle causal curves such that
\[
(\mathrm{ev}_t)_\#\mathbb P=\mu_t
\]
for all \(t\in I\). For \(I=\mathbb R\), \(\mathbb P\) becomes a splitting-independent measure on \(N\) inextendible worldlines [2004.07015].

The same measure-theoretic framework accommodates indistinguishable particles through \(S_N\)-invariant measures and proves that multi-photon and multi-fermion Schrödinger evolutions in Minkowski spacetime are compatible with relativistic causality whenever the induced multi-velocity fields are subluminal. In that sense, standard configuration-space dynamics can be “forgotten” in favor of an observer-independent measure on worldlines. This does not eliminate time from the construction, but it relocates the invariant content from a chosen slicing to the path-space measure itself [2004.07015].

## 4. Measurement and state update without preferred time slices

Generally-covariant quantum mechanics also includes a measurement theory for quantum fields on curved spacetimes that dispenses with instantaneous collapse on a global time slice. In the locally covariant AQFT formulation, a “system” theory \(\mathcal A(M)\) and a “probe” theory \(\mathcal P(M)\) are coupled only within a compact spacetime region \(K\). The uncoupled combined theory is \(\mathcal C(M)=\mathcal A(M)\otimes\mathcal P(M)\), and the coupled theory \(\mathcal D(M)\) is required to agree with \(\mathcal C(M)\) outside the causal hull
\[
\mathrm{ch}(K)=J^+(K)\cap J^-(K).
\]
Using in/out regions \(M^\pm=M\setminus J^\mp(K)\), one defines advanced and retarded response maps and hence the scattering morphism
\[
\Theta=(\tau^-)^{-1}\circ \tau^+:\mathcal C(M)\to\mathcal C(M).
\]

If the probe is prepared in a state \(\sigma\), partial evaluation
\[
\eta_\sigma(A\otimes B)=A\,\sigma(B)
\]
maps probe observables to induced system observables
\[
\varepsilon_\sigma(B)=\eta_\sigma\!\left(\Theta(\mathbf 1\otimes B)\right)\in\mathcal A(M).
\]
The corresponding Heisenberg-picture pre-instrument is
\[
\mathsf J_\sigma(B)(A)=\eta_\sigma\!\left(\Theta(A\otimes B)\right),
\]
and the normalized post-selected state is
\[
\omega_{|B}(A)=\frac{\omega\!\big(\mathsf J_\sigma(B)(A)\big)}{\omega\!\big(\varepsilon_\sigma(B)\big)}.
\]
The induced observable \(\varepsilon_\sigma(B)\) is localized in any connected open causally convex region containing \(K\), hence in particular within \(\mathrm{ch}(K)\). If \(B\) is localized in \(K^\perp\), then \(\varepsilon_\sigma(B)=\sigma(B)\mathbf 1\), so no information about the system can be obtained at spacelike separation. Even when \(B\) is sharp in the probe, \(\varepsilon_\sigma(B)\) is generally an effect rather than a projection, and the actual coupled-theory measurement is less sharp still [1904.06944].

The formalism includes a precise notion of causal composition. If two coupling regions \(K_1\) and \(K_2\) satisfy the stated causal factorization condition, then the combined scattering morphism factorizes, and the instruments compose as
\[
\mathsf J_{\sigma_2}(B_2)\circ \mathsf J_{\sigma_1}(B_1)
=
\mathsf J_{\sigma_1\otimes\sigma_2}(B_1\otimes B_2).
\]
If \(K_1\) and \(K_2\) are spacelike separated, the composite is independent of the admissible causal order. Selective updates can modify expectations of observables localized in \(K^\perp\), but only through pre-existing correlations; the framework explicitly states that this does not permit superluminal signaling. The scalar system–probe example further shows that induced observables and measurement statistics can be computed concretely from \(\Theta\), with probe noise appearing as an additional contribution to the variance [1904.06944].

In this approach, the temporal location of a measurement is replaced by causal localization. “When” the interaction occurs is encoded by the causal relations of \(K\), \(M^\pm\), and \(\mathrm{ch}(K)\), not by a preferred foliation. This is a specifically generally-covariant answer to the measurement problem in QFT on curved backgrounds [1904.06944].

## 5. Discrete-time, histories, and intrinsic discrete geometry

A different formulation treats quantum dynamics through “states over time,” namely a single multi-time operator encoding temporal correlations of a system evolving under a chain of CPTP maps. For finite-dimensional \(C^*\)-algebras \(A_k\), the key primitive is the canonical broadcasting map
\[
\mathfrak B_A=\frac12(\mu_A^*+\widetilde\mu_A^*):A\to A\otimes A,
\]
which in matrix algebras becomes
\[
\mathfrak B_A(X)=\frac12\{X\otimes I,\mathrm{SWAP}\}.
\]
Given a trace-preserving map \(E:A\to B\), its “bloom” is
\[
(E)=(\mathrm{id}_A\otimes E)\circ\mathfrak B_A,
\]
and a chain \((E_1,\ldots,E_n)\) produces the canonical state over time
\[
(E_1,\ldots,E_n)\star\rho=(E_1,\ldots,E_n)(\rho)\in A_0\otimes\cdots\otimes A_n.
\]
This operator is Hermitian, trace one, and has the correct single-time marginals, but it need not be positive. The central covariance theorem states that for \(*\)-isomorphisms \(\phi_k:A_k\to A_k'\),
\[
(\phi_0\otimes\cdots\otimes\phi_n)\big[(E_1,\ldots,E_n)\star\rho\big]
=
(E_1',\ldots,E_n')\star\rho',
\]
where \(E_k'=\phi_k\circ E_k\circ\phi_{k-1}^{-1}\) and \(\rho'=\phi_0(\rho)\). The same framework proves the covariance of the dynamical quantum Bayes rule
\[
E\star\rho=\gamma(\widetilde E\star E(\rho)),
\]
with \(\gamma\) the lexicographic swap. Here, “general covariance” means natural transformation under local representation changes at each time step rather than diffeomorphism invariance on a continuous manifold [2311.00162].

A more radically discrete program is supplied by covariant causal sets, or \(c\)-causets. A \(c\)-causet is a finite partially ordered set with a unique labeling, equivalently one whose shell sequence \((s_0(x),s_1(x),\ldots,s_k(x))\) uniquely determines the causal set. This class supports intrinsic notions of path length,
\[
C(w)=\left(\sum_{j=1}^{n-1}(a_{j+1}-a_j)^2\right)^{1/2},
\]
distance \(d(a,b)\), geodesics, and curvature
\[
K(a)=\big(\text{number of geodesics from }1\text{ to }a\big)-1.
\]
Quantum mechanics is defined by a covariant difference operator \(V_w\) along a path \(w\) and a free wave equation
\[
i\,V_w v(a,b)=m\,d_w(a,b)\,v(a,b),
\]
where \(d_w(a,b)=d(a,b)-L_a(w)\) measures deviation from geodesicity. The framework then builds a path Hilbert space \(L^2(\mathcal M_k(x))\), a decoherence functional \(D_k\), a grade-2 additive quantum propensity \(\mu_k\), and position probabilities on shells. In the small worked examples, higher-curvature vertices acquire larger probabilities as the mass grows because non-geodesic contributions are suppressed by the \(m\,d_w\) term [1507.04810].

These discrete formulations interpret covariance in two very different ways. In the states-over-time framework, covariance is categorical and representation-theoretic. In the \(c\)-causet framework, covariance is built into intrinsic order geometry and the absence of embedding-dependent data. Their common feature is that temporal description is reconstructed from compositional or path-based structures rather than from an a priori classical time background [2311.00162][1507.04810].

## 6. Quantum geometry of frames, misconceptions, and open directions

The most explicit attempt to quantize the background geometry of reference frames is “Doubly Quantum Mechanics.” In this formalism, the classical rotation group \(SU(2)\) is replaced by the compact quantum group \(SU_q(2)\), and the geometric data specifying spin directions, apparatus orientations, and relative frame transformations are encoded in geometry states \(|\Phi\rangle\in H_{\mathrm{geo}}\). Spin-\(\tfrac12\) pre-measurement states take the form
\[
|\psi\rangle=U_s^q|\uparrow\rangle=|\uparrow\rangle x+|\downarrow\rangle y,
\]
with operator-valued coefficients in \(C(SU_q(2))\), and a generic q-deformed Pauli observable is
\[
\sigma^q=U_a^q\,\sigma_z^q\,(U_a^q)^\dagger .
\]
Probabilities become positive self-adjoint operators on geometry,
\[
P(\uparrow)_\sigma=\langle\psi|\Pi^\uparrow_\sigma|\psi\rangle,
\]
so that “probability is an observable.” For two observers related by a quantum-group frame transformation \(U_g^q\), the formalism proves invariance of probabilities and expectation values,
\[
P_B(\uparrow)_\sigma=P_A(\uparrow)_\sigma,
\qquad
\langle\sigma^q\rangle_B^B=\langle\sigma^q\rangle_A^A .
\]
Semi-classical geometry states reproduce the standard \(SU(2)\) expressions up to \(O(1-q)\), but the reference-frame alignment protocol exhibits a residual uncertainty
\[
\Delta_\infty\langle\sigma^q\rangle=(q+q^{-1})\,\Delta_f^{1/2}
\]
that persists as the number of exchanged qubits \(N\to\infty\), unlike the standard \(SU(2)\) case [2412.05997].

Several recurring misconceptions are explicitly addressed across this literature. First, general covariance is not the same as unrestricted physical symmetry: in the tensorial state-space approach, arbitrary diffeomorphisms preserve the form of the equations, but only special transformations, such as unitary-induced Kähler isometries, preserve the tensors themselves [1302.0194]. Second, generally-covariant formulations do not automatically remove all quantum-state-update subtleties: the AQFT measurement scheme replaces collapse on a time slice by local instruments, but selective updates remain nontrivial and depend on correlations [1904.06944]. Third, manifest covariance does not necessarily exclude superluminal or nonlocal structures at the level of hidden trajectories: the many-time Bohmian formalism allows superluminal segments and nonlocal dependence while maintaining relativistic covariance of the laws and standard measurable statistics [1205.1992]. Fourth, discrete covariance does not by itself solve continuum problems: the \(c\)-causet program still leaves open the continuum limit, Lorentz invariance, and the extension beyond uniquely labeled causets [1507.04810].

The open problems are correspondingly diverse. The tensorial state-space approach is cleanest in finite dimension and faces domain issues for unbounded operators in Hilbert-manifold settings [1302.0194]. The contact-geometric program emphasizes local Darboux-patch structure and leaves global existence of the flat connection, global polarization, and obstruction theory as open questions [1709.04557]. The \(N\)-particle measure-theoretic construction currently relies on global hyperbolicity and on curved-spacetime extensions of superposition techniques that are not yet fully developed [2004.07015]. The \(c\)-causet framework highlights the need for efficient large-causet algorithms, interactions beyond the free wave equation, and a clarified continuum limit [1507.04810]. Doubly Quantum Mechanics remains kinematical in the rotational sector and explicitly lists extensions to boosts, translations, constraints, and relational time as open directions [2412.05997].

A broad conclusion nevertheless emerges. Generally-covariant quantum mechanics is not a single replacement for ordinary quantum theory, but a research program with several precise realizations. In each realization, the invariant content is shifted away from fixed coordinates, external clocks, or classical frames and toward tensors on state manifolds, causal couplings of measures, local scattering morphisms, multi-time operators, intrinsic discrete order, or quantum geometry states. This suggests that the deepest unifying idea is not a specific formalism, but the demand that quantum kinematics, dynamics, and measurement be formulated in terms of structures whose physical meaning does not depend on how observers coordinatize them [1302.0194][1904.06944][2004.07015][2311.00162][2412.05997].

Source: https://www.emergentmind.com/topics/generally-covariant-quantum-mechanics