Hilbert Space Fundamentalism

• Hilbert Space Fundamentalism is defined as the claim that a Hilbert space, Hamiltonian, and state vector completely describe physical reality, making emergent structures secondary.
• Researchers employ tensor-product decompositions, decoherence, and entanglement measures to recover subsystem structure, locality, and geometric relations from minimalist quantum data.
• Critics argue that unique emergent classical structures cannot be derived solely from abstract Hilbert space data, a debate that also informs studies of Hilbert space fragmentation.

Hilbert Space Fundamentalism is the thesis that the most basic ontology of physics is exhausted by a Hilbert space, a Hamiltonian, and a state vector, with all further structure—subsystems, space, locality, fields, and classical descriptions—arising as higher-level emergent structure rather than entering the theory primitively (Carroll, 2021). In current research usage, however, the acronym HSF also denotes Hilbert space fragmentation, a distinct many-body mechanism of nonergodicity; the two meanings are unrelated except for the shared phrase “Hilbert space” (Honda et al., 5 Feb 2025).

1. Definition and formal content

In its strongest formulation, Hilbert Space Fundamentalism identifies the complete physical world with the abstract quantum data

$(\mathcal H,\hat H,|\psi\rangle),$

where $\mathcal H$ is a Hilbert space, $\hat H$ a Hermitian Hamiltonian, and $|\psi(t)\rangle$ a unit vector evolving by

$$|\psi(t)\rangle=e^{-i\hat H t/\hbar}|\psi(0)\rangle.$$

On this view, the basis-independent state vector, rather than any particular wavefunction representation, is fundamental (Stoica, 23 Feb 2026).

A central distinction in the literature is between a state vector and a wavefunction. A wavefunction such as

$\psi(x,t)=\langle x|\psi(t)\rangle$

already presupposes a configuration space, a preferred basis, or position observables. Strong HSF therefore treats $|\psi\rangle$ as fundamental but not $\psi(x,t)$, since the latter requires additional structure that HSF aims to derive rather than assume (Stoica, 2023).

Cristi Stoica formulates the doctrine as the claim that the triple $(\mathcal H,\hat H,|\psi\rangle)$ gives a complete unambiguous description of physical reality, with equivalence defined up to unitary isomorphism: $$\hat H' = U\hat H U^{-1}, \qquad |\psi'\rangle = U|\psi\rangle.$$ This is stronger than the claim that Hilbert space is merely useful or central. It is a completeness thesis: abstract Hilbert-space data are supposed to fix the physical world without primitive observables, primitive subsystem decomposition, or primitive spacetime structure (Stoica, 23 Feb 2026).

Sean Carroll’s version is especially austere. It treats the world as “completely and exactly represented by a vector in an abstract Hilbert space, evolving in time according to unitary Schrödinger dynamics,” and argues that “the laws of physics are determined solely by the energy eigenspectrum of the Hamiltonian” in the finite-dimensional nondegenerate setting (Carroll, 2021).

2. Emergence program from the bare quantum structure

The constructive side of HSF is an emergence program. Its aim is not simply to assert that space, fields, and subsystems are unreal, but to show how they can arise from the minimalist quantum structure itself. Carroll’s paper is the clearest positive statement of this program: the state vector and Hamiltonian are fundamental, while particles, fields, space, and classical objects are emergent descriptions (Carroll, 2021).

A first step is the recovery of subsystem structure through tensor-product decomposition,

$\mathcal H$0

or, more schematically,

$\mathcal H$1

Given such a factorization, the Hamiltonian may be written in the familiar form

$\mathcal H$2

and decoherence can then be used to identify pointer observables and quasi-classical states. Carroll proposes that the preferred factorization should be recovered by searching for decompositions in which subsystem dynamics are quasi-classical, interaction terms support decoherence rather than generic scrambling, and localized states remain approximately localized (Carroll, 2021).

A second step is the recovery of locality and space. In the intended factorization,

$\mathcal H$3

the Hamiltonian should admit a quasi-local expansion in self, two-point, three-point, and higher operators. Carroll appeals to “locality from the spectrum” as evidence that when such a local factorization exists, it can be essentially unique and recoverable from abstract quantum structure (Carroll, 2021). Stoica reconstructs this ambition in similar terms: HSF hopes that subsystem decomposition, tensor-product structure, locality, and space can emerge from the bare pair $\mathcal H$4 rather than being posited (Stoica, 23 Feb 2026).

A third step is geometry from entanglement. Carroll uses reduced density matrices and mutual information,

$\mathcal H$5

with

$\mathcal H$6

to motivate an emergent notion of spatial distance. The suggestion is that if mutual information falls with effective distance, then geometry can be reconstructed from entanglement structure, and under additional assumptions even gravitational dynamics may be related to entanglement variations (Carroll, 2021).

The literature presents this as a research program rather than a finished derivation. Carroll explicitly says that much work remains, especially for the emergence of local quantum fields, full spacetime rather than space alone, and realistic low-energy physics (Carroll, 2021).

3. Major objections and no-go arguments

The main criticisms target the uniqueness and completeness claims of strong HSF. Stoica’s general theorem states that if a candidate emergent structure is physically relevant, then it is not unique when it is supposed to arise from $\mathcal H$7 alone (Stoica, 2021). In the simplified presentation, candidate structures are represented by families of Hermitian operators or more general tensorial data, and the core argument uses unitaries $\mathcal H$8 commuting with $\mathcal H$9. If $\hat H$0 is physically distinct from $\hat H$1, then a structure transported from $\hat H$2 back to $\hat H$3 yields another admissible structure of the same kind. Hence a preferred basis, preferred factorization, or emergent 3D space cannot be both physically relevant and uniquely determined by the minimalist quantum structure (Stoica, 2021).

This argument is applied directly to several targets. For preferred bases, a basis can be encoded by rank-1 projectors $\hat H$4 satisfying

$\hat H$5

Because these defining conditions are unitary invariant, commuting unitaries generate alternative candidate preferred bases. Stoica’s conclusion is that a physically relevant preferred basis cannot emerge uniquely from $\hat H$6 (Stoica, 2021).

The same style of argument is extended to tensor-product structure and space. Stoica’s 2021 paper argues that 3D space and preferred factorization cannot uniquely emerge from the quantum structure alone, and that HSF therefore fails as a doctrine of unique emergence (Stoica, 2021). The “prince and pauper” paper sharpens the point by constructing measurement scenarios in which the same abstract basic quantum structure can correspond to macroscopically distinct worlds—one in which Edward becomes rich and one in which he becomes poor—even when a tensor-product structure is held fixed (Stoica, 2023).

A distinct but related criticism concerns time. Stoica’s “No change in Hilbert space fundamentalism” argues that if HSF identifies unitarily isomorphic triples as the same physical reality, then Schrödinger evolution itself becomes an automorphism of the triple: $\hat H$7 The triple at time $\hat H$8 is therefore isomorphic, in HSF’s own sense, to the triple at time $\hat H$9. Stoica’s conclusion is that strong HSF cannot account for temporal change, because it identifies the entire Schrödinger orbit with one physical reality (Stoica, 23 Feb 2026).

These objections do not show that Hilbert-space language is dispensable. Their target is the stronger thesis that the unitary-isomorphism class of $|\psi(t)\rangle$0 by itself yields a complete and unique description of the world. Several of the papers explicitly leave room for weaker views that add preferred observables, preferred factorization, or other extra structure (Stoica, 2021).

4. Reconstruction from non-Hilbert primitives

A different line of work is relevant because it undercuts the claim that Hilbert space must be primitive at all. In “Hilbert Spaces from Path Integrals,” a Hilbert space is reconstructed from a histories-based framework in which the primitive data are a sample space $|\psi(t)\rangle$1 of histories, an event algebra $|\psi(t)\rangle$2, and a decoherence functional

$|\psi(t)\rangle$3

satisfying Hermiticity, additivity on disjoint unions, normalization, and strong positivity (Dowker et al., 2010).

The construction begins with the free vector space $|\psi(t)\rangle$4 of finitely supported complex-valued functions on $|\psi(t)\rangle$5, and defines the sesquilinear form

$|\psi(t)\rangle$6

Because this form can be degenerate, one quotients by null vectors and completes, yielding the History Hilbert space $|\psi(t)\rangle$7 (Dowker et al., 2010). Philosophically, this matters because the construction uses only event/decoherence structure, not a primitive Hilbert space of states.

The same paper proves that for finite configuration-space systems and for a non-relativistic particle in $|\psi(t)\rangle$8, the reconstructed History Hilbert space is naturally isomorphic to the standard state space under explicit assumptions. In the particle case, the event algebra is built from homogeneous events

$|\psi(t)\rangle$9

the restricted-evolution vectors $$|\psi(t)\rangle=e^{-i\hat H t/\hbar}|\psi(0)\rangle.$$0 are defined by propagator integrals, and the decoherence functional is

$$|\psi(t)\rangle=e^{-i\hat H t/\hbar}|\psi(0)\rangle.$$1

Under continuity and nonvanishing assumptions on the propagator, the induced map from the History Hilbert space onto $$|\psi(t)\rangle=e^{-i\hat H t/\hbar}|\psi(0)\rangle.$$2 is onto, so $$|\psi(t)\rangle=e^{-i\hat H t/\hbar}|\psi(0)\rangle.$$3 (Dowker et al., 2010).

The philosophical consequence is double-edged. On one reading, Hilbert space is derivative rather than fundamental: histories and decoherence are enough. On another reading, the result shows the structural robustness of Hilbert space: once one has a strongly positive decoherence functional, Hilbert-space structure “automatically arises” in standard cases (Dowker et al., 2010). The paper therefore weakens primitive HSF without eliminating the possibility that Hilbert structure is a canonical representational form.

5. Structural enrichment and higher generalizations

Some work relevant to HSF does not defend the doctrine directly but argues that Hilbert space carries richer intrinsic structure than the textbook “vector space of states” picture suggests. Sunko’s “Fundamental invariants of many-body Hilbert space” treats many-body Hilbert space as a graded algebra under ordinary multiplication of wavefunctions, with a finite generating set of shapes. For $$|\psi(t)\rangle=e^{-i\hat H t/\hbar}|\psi(0)\rangle.$$4 identical particles in $$|\psi(t)\rangle=e^{-i\hat H t/\hbar}|\psi(0)\rangle.$$5 dimensions, the Hilbert series

$$|\psi(t)\rangle=e^{-i\hat H t/\hbar}|\psi(0)\rangle.$$6

leads to the counting result

$$|\psi(t)\rangle=e^{-i\hat H t/\hbar}|\psi(0)\rangle.$$7

Every state has the decomposition

$$|\psi(t)\rangle=e^{-i\hat H t/\hbar}|\psi(0)\rangle.$$8

where the $$|\psi(t)\rangle=e^{-i\hat H t/\hbar}|\psi(0)\rangle.$$9 are shapes and the $\psi(x,t)=\langle x|\psi(t)\rangle$0 are symmetric-polynomial coefficients interpreted as bosonic excitations (Sunko, 2017).

This does not establish HSF as a metaphysical doctrine. What it does show is that many-body Hilbert space admits an internal algebraic architecture—grading, generators, ideals, vacua, and excitation structure—that is at least partly Hamiltonian-independent at the level of state-space organization (Sunko, 2017). For HSF-adjacent programs, that supports the narrower claim that physically significant structure may be encoded in Hilbert space itself.

Recent higher-categorical work further complicates any naive appeal to “the” Hilbert structure. “The many faces of higher Hilbert spaces” argues that once one categorifies quantum linear algebra there is not a single canonical higher analogue of Hilbert space. Instead, several inequivalent notions are organized by subgroup choices

$\psi(x,t)=\langle x|\psi(t)\rangle$1

with different $\psi(x,t)=\langle x|\psi(t)\rangle$2-fixed-point data yielding $\psi(x,t)=\langle x|\psi(t)\rangle$3-Hilbert, $\psi(x,t)=\langle x|\psi(t)\rangle$4-, $\psi(x,t)=\langle x|\psi(t)\rangle$5-, and $\psi(x,t)=\langle x|\psi(t)\rangle$6-type structures (Ferrer et al., 9 Jun 2026). The paper defines $\psi(x,t)=\langle x|\psi(t)\rangle$7-Hermitian $\psi(x,t)=\langle x|\psi(t)\rangle$8-vector spaces as fixed points for the induced $\psi(x,t)=\langle x|\psi(t)\rangle$9-action on $|\psi\rangle$0, and treats positivity as a categorified generalization of the passage from Hermitian vector spaces to Hilbert spaces.

For foundational debates, the implication is not that ordinary Hilbert spaces are obsolete. Rather, it is that “Hilbertian” structure itself has multiple mathematically natural higher-categorical faces (Ferrer et al., 9 Jun 2026). Any generalized HSF must therefore specify which Hilbert-like structure is supposed to be fundamental.

6. Terminological divergence: HSF as Hilbert space fragmentation

In many-body physics, HSF very often means Hilbert space fragmentation, not Hilbert Space Fundamentalism. In that literature, HSF is a mechanism of nonergodicity in which the many-body Hilbert space decomposes into many dynamically disconnected or nearly disconnected sectors. The 2025 Bose–Hubbard experiment states this definition explicitly: “the Hamiltonian matrix splits into an exponentially large number of sectors due to the presence of nontrivial conserved quantities” (Honda et al., 5 Feb 2025).

This usage is conceptually distinct from the philosophical doctrine. In the strongly interacting one-dimensional Bose–Hubbard chain, the relevant observation was slow relaxation from a doublon charge-density wave together with approximate conservation of singlon and doublon numbers. The work interprets this as HSF caused by strong interactions rather than disorder or tilt, and presents it as experimental confirmation of the conserved quantities responsible for fragmentation in that setting (Honda et al., 5 Feb 2025).

Related papers extend the fragmentation vocabulary in several directions. In Floquet-engineered Stark lattices, the drive frequency selects which density-dependent hopping channels are resonant, producing three distinct strong-HSF regimes in a tilted chain of interacting spinless fermions (Zhang et al., 2023). In a resonantly driven tilted fermion chain with emergent chiral symmetry, “pseudo HSF” refers to approximate block structure generated by a hierarchy of matrix elements rather than exact kinetic constraints; its interplay with chiral symmetry yields a localized zero-energy many-body scar (Zhang et al., 2024). In a driven-dephasing Rydberg atom array, HSF appears in an open-system Liouvillian setting, with fragmented sectors labeled by the consecutive double excitation addressing operator and a number of sectors that grows according to the modified Fibonacci recurrence

$|\psi\rangle$1

(Yan et al., 28 Nov 2025).

Because this many-body usage is now standard, any discussion of HSF must disambiguate whether the subject is the philosophical thesis about ontology or the dynamical fragmentation phenomenon in constrained quantum systems. The two literatures intersect only loosely: both treat Hilbert space as structurally significant, but one is a metaphysical thesis about what is fundamental, while the other is a concrete claim about sectorization and nonergodic dynamics.

Hilbert Space Fundamentalism therefore remains a contested foundational thesis rather than an established consequence of quantum theory. Its strongest advocates argue that a basis-independent state vector and Hamiltonian are enough, in principle, to recover the observed world (Carroll, 2021). Its sharpest critics argue that such a program cannot uniquely recover physically relevant structure and cannot, in its pure form, account for temporal change (Stoica, 2021). At the same time, reconstruction results from histories-based quantum mechanics show that Hilbert space can be derived from non-Hilbert primitives (Dowker et al., 2010), while algebraic and higher-categorical work shows that Hilbert structure itself is richer and less monolithic than the traditional formulation suggests (Sunko, 2017). The current state of the subject is accordingly plural: Hilbert space remains central, but whether it is fundamental, derivative, or one member of a broader family of Hilbert-like structures is still an open question.