pith. sign in

arxiv: 2502.08494 · v3 · pith:T5EEVN6Fnew · submitted 2025-02-12 · 🪐 quant-ph · math-ph· math.MP

All Hilbert spaces are the same: consequences for generalized coordinates and momenta

S.J. van Enk , Daniel A. Steck This is my paper

Pith reviewed 2026-05-23 03:21 UTC · model grok-4.3

classification 🪐 quant-ph math-phmath.MP
keywords Hilbert space isomorphismgeneralized coordinatesconjugate momentaself-adjoint extensionsNeumark extensionsquantum measurementsPOVM
0
0 comments X

The pith

All separable Hilbert spaces of the same dimension being isomorphic means there are only six basic ways to define generalized coordinate operators in quantum mechanics.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows that the isomorphism of all separable complex Hilbert spaces of a given dimension restricts the possible definitions of generalized coordinate operators to exactly six basic types. For each type a canonically conjugate momentum operator exists, although it is not always self-adjoint on the original space. Self-adjointness can always be recovered either by extending the momentum operator itself or by a Neumark dilation of the Hilbert space, and in one case both routes are available, producing seven distinct pairs. The same isomorphism argument also explains why coordinate and momentum measurements admit more distinct realizations than the operators themselves, with simultaneous measurement occupying a distinguished position.

Core claim

Making use of the simple fact that all separable complex Hilbert spaces of given dimension are isomorphic, there are just six basic ways to define generalized coordinate operators in Quantum Mechanics. In each case a canonically conjugate generalized momentum operator can be defined, but it may not be self-adjoint. Even in those cases there is always either a self-adjoint extension of the operator or a Neumark extension of the Hilbert space that produces a self-adjoint momentum operator. In one of the six cases both extensions work, thus leading to seven basic pairs of coordinate and momentum operators. There are more ways of defining basic coordinate and momentum measurements, with a特殊 role

What carries the argument

The isomorphism of separable complex Hilbert spaces of given dimension, together with the algebraic commutation relations, domains, and possible extensions of the resulting operators.

If this is right

  • Exactly six basic definitions of generalized coordinate operators exist.
  • Each definition admits a conjugate momentum, yielding seven basic pairs once extensions are included.
  • Self-adjointness of the momentum is guaranteed either by operator extension or by Neumark dilation of the space.
  • Coordinate and momentum measurements can be realized in strictly more ways than the underlying operators.
  • Simultaneous measurement of a coordinate-momentum pair occupies a special status among possible measurements.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Physical systems that appear to use different coordinates may still reduce to one of the six types once the Hilbert-space isomorphism is taken into account.
  • The result supplies a systematic way to enumerate all possible pairs of observables that can serve as generalized position and momentum.
  • The multiplicity of measurement realizations suggests that experimental protocols can distinguish more than the six operator pairs.

Load-bearing premise

The classification assumes that the only structure that distinguishes possible coordinate definitions is the isomorphism class of the Hilbert space together with the algebraic properties of the operators and their domains; any additional physical or contextual structure that could differentiate definitions is set aside.

What would settle it

Exhibiting a definition of a generalized coordinate operator whose algebraic relations and domain properties cannot be mapped onto any of the six types by an isomorphism of the underlying Hilbert space would falsify the classification.

read the original abstract

Making use of the simple fact that all separable complex Hilbert spaces of given dimension are isomorphic, we show that there are just six basic ways to define generalized coordinate operators in Quantum Mechanics. In each case a canonically conjugate generalized momentum operator can be defined, but it may not be self-adjoint. Even in those cases we show there is always either a self-adjoint extension of the operator or a Neumark extension of the Hilbert space that produces a self-adjoint momentum operator. In one of the six cases both extensions work, thus leading to seven basic pairs of coordinate and momentum operators. We also show why there are more ways of defining basic coordinate and momentum measurements. A special role is reserved for measurements that simultaneously measure both.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript uses the fact that all separable complex Hilbert spaces of fixed dimension are unitarily isomorphic to classify generalized coordinate operators in quantum mechanics. It concludes there are exactly six basic definitions, each admitting a canonically conjugate momentum operator (which may fail to be self-adjoint), and shows that self-adjoint or Naimark extensions always exist, yielding seven basic coordinate-momentum pairs. The paper further argues that measurements admit more possibilities, with simultaneous measurements playing a distinguished role.

Significance. If the case analysis is exhaustive and the extension claims are rigorously verified, the classification would provide a compact, isomorphism-based taxonomy of coordinate-momentum pairs grounded in standard functional-analysis facts. The explicit appeal to unitary equivalence and the handling of essential self-adjointness and Naimark extensions are positive features that could clarify operator-domain issues in QM foundations.

major comments (2)
  1. [Abstract / classification section] Abstract and main classification section: the central claim of 'just six basic ways' (leading to seven pairs) rests on an enumeration whose completeness cannot be checked because the manuscript supplies neither the explicit operator constructions for each case nor the full case-by-case analysis of algebraic properties, domains, and extension existence. Without these derivations the count remains unverified and the load-bearing assertion cannot be assessed.
  2. [Extensions discussion] Section on extensions: the statement that 'in all cases' either a self-adjoint extension or a Naimark extension produces a self-adjoint momentum is asserted without a concrete verification or counter-example check for each of the six classes; a single worked example (e.g., the case where both extensions work) is needed to substantiate the 'seven pairs' total.
minor comments (2)
  1. [Introduction] Notation for 'generalized coordinate' versus standard position operator should be introduced with a brief definition or reference to avoid ambiguity in the opening paragraphs.
  2. [Measurements section] The paper mentions 'more ways of defining basic coordinate and momentum measurements' but does not quantify or classify them; a short table or enumerated list would improve clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive feedback. The comments highlight areas where additional explicit detail will improve verifiability. We address each point below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract / classification section] Abstract and main classification section: the central claim of 'just six basic ways' (leading to seven pairs) rests on an enumeration whose completeness cannot be checked because the manuscript supplies neither the explicit operator constructions for each case nor the full case-by-case analysis of algebraic properties, domains, and extension existence. Without these derivations the count remains unverified and the load-bearing assertion cannot be assessed.

    Authors: We agree that the current presentation would be strengthened by explicit constructions. The classification follows directly from the possible inequivalent ways to realize a self-adjoint operator (or symmetric operator) on a separable Hilbert space up to unitary isomorphism, but the manuscript presents this at a summary level. In revision we will add the explicit operator definitions for each of the six classes together with the corresponding domain specifications, commutation relations, and verification that the listed cases exhaust the isomorphism classes. revision: yes

  2. Referee: [Extensions discussion] Section on extensions: the statement that 'in all cases' either a self-adjoint extension or a Naimark extension produces a self-adjoint momentum is asserted without a concrete verification or counter-example check for each of the six classes; a single worked example (e.g., the case where both extensions work) is needed to substantiate the 'seven pairs' total.

    Authors: We accept that a single worked example and brief checks for the remaining classes are needed to make the extension claim fully transparent. The manuscript already notes that one class admits both extensions, but does not display the concrete operators. In the revision we will supply an explicit worked example for that class (showing both the self-adjoint extension on the original space and the Naimark extension) and short arguments confirming extension existence for the other five classes. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation rests on external isomorphism fact plus case analysis of operator properties

full rationale

The paper begins from the standard, externally verifiable fact that all separable complex Hilbert spaces of fixed dimension are unitarily isomorphic and then performs an exhaustive case analysis on possible definitions of generalized coordinate operators, distinguished solely by algebraic relations, self-adjointness, essential self-adjointness, domains, and the existence of self-adjoint or Naimark extensions. No fitted parameters are renamed as predictions, no self-citations supply load-bearing uniqueness theorems, and the enumeration into six (or seven) classes is not equivalent by construction to the input isomorphism statement; it follows from applying known results in unbounded operator theory. The scope limitation that only isomorphism class plus algebraic/domain properties are considered is stated explicitly, rendering the argument self-contained within standard functional analysis rather than circular.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The paper rests on one standard mathematical fact and two standard extension theorems; no free parameters or invented entities appear.

axioms (2)
  • standard math All separable complex Hilbert spaces of given dimension are isomorphic.
    Invoked in the first sentence of the abstract as the starting point for the classification.
  • standard math Every symmetric operator on a Hilbert space admits a self-adjoint extension or a Neumark extension to a larger space.
    Used to guarantee that a momentum operator can always be made self-adjoint.

pith-pipeline@v0.9.0 · 5654 in / 1313 out tokens · 29946 ms · 2026-05-23T03:21:54.114245+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

51 extracted references · 51 canonical work pages · 6 internal anchors

  1. [1]

    That this is so, follows, in physicists’ notation, from ⟨p′|p⟩ = δ(p′ − p)

    Case 1 When the domain of x is the entire real axis, d µ p = |p⟩ ⟨p|dp is a projection-valued measure. That this is so, follows, in physicists’ notation, from ⟨p′|p⟩ = δ(p′ − p). This implies that P1 is self-adjoint. In this case, we do simply have P1 = −id/dx . The commutator [ X1,P 1] = i11H1 is thus the standard one. Arguably the more funda- mental rel...

    work page

  2. [2]

    doubled” the total Hilbert space: in the first case by extending the semi-infinite real axis to the whole real axis, in the second case by using an extra (“ancilla

    Case 2 For the semi-infinite axis, P2 is not self-adjoint, as the measure dµ p = |p⟩ ⟨p|dp is not projection-operator valued. That is, there are disjoint subsets E and E′ of R such that ( ∫ E |p⟩ ⟨p|dp ) ( ∫ E′ |p⟩ ⟨p|dp ) ̸= 0. (7) This follows, again in physicists’ notation, from the pres- ence of the second term here: ⟨p′|p⟩ = 1 2δ(p′ − p) + i 2π (p − p...

    work page

  3. [3]

    wrap around

    Case 3 For the finite interval case, just as in Case 2, the mea- sure dµ p = |p⟩ ⟨p|dp is not projection-operator valued, although it is a positive-operator valued measure. The operator P3 is, therefore, not self-adjoint. In this case, both types of extensions work. That is, we can define a self-adjoint extension of P3 on H3 and we can define a Neumark exten...

    work page

  4. [4]

    delta-function normalized

    Case 4 Case 4, with x unbounded in two directions, leads to the most straightforward definition of momentum (just like it did for the continuous case). We may choose the eigenvalues of the operatorX4 to be xm =mL0 +λ 0, with L0 a constant with units of length and 0 ≤ λ 0 < L0. We may, for example, choose L0 to equal the Planck length. In this way, we can b...

    work page

  5. [5]

    This case nat- urally describes a quantum system moving on the semi- infinite real axis

    Case 5 Now we choose xn =nL0 with n ∈ N. This case nat- urally describes a quantum system moving on the semi- infinite real axis. Our fifth position operator is then X5 = ∑ n∈ N xn|xn⟩ ⟨xn|. (19) Just as in Case 4 we start in physicists’ notation with |p⟩ = ∑ n∈ N exp(ipxn) |xn⟩. (20) While we do have ∫ Ωp |p⟩ ⟨p|dp = ∑ n∈ N |xn⟩ ⟨xn|= 11H5, (21) the operat...

    work page

  6. [6]

    momentum

    Case 6: finite-dimensional Hilbert space The finite-dimensional case is straightforward since the coordinate and momentum operators will both be bounded. The coordinate operator may be used to de- scribe different sorts of discrete physical coordinates. We could use a discretized angle φ as generalized coordinate and hence angular momentum as our canonically...

    work page

  7. [7]

    That is, whenever xn < xk < xn′ we have n < k < n′

    Cases 4 and 5 revisited We assumed in the preceding subsections that the la- beling preserves the spatial order of the coordinates xn. That is, whenever xn < xk < xn′ we have n < k < n′. Although this is very natural it is not necessary, and there is one reason for not following this prescription in Case 5 wheren ∈ N. Namely, since we needed a Neumark ext...

    work page

  8. [8]

    finest” possi- ble measurements. That is, we do not include POVMs that are “coarser

    Discrete position on a finite interval revisited In Case 6 we defined a finite number of discrete co- ordinates in a finite interval. We may, however, define infinitely many of them. In that case we need to let |xn+1 − xn| go to zero for |n| → ∞ . This case is not really physically different from the Cases 4 or 5, how- ever. Namely, we just interpret x as merely...

    work page

  9. [9]

    Thus, the POVM {Πα} de- scribes a physically acceptable simultaneous measure- ment of X3 and ˜P3

    we can use the projector 113 := ∑ m |pm⟩ ⟨pm| to convert the POVM { 1 π |α ⟩ ⟨α |} on L2(Ω1, dx) to a POVM on L2(Ω3, dx) by defining Πα := 113 1 π |α ⟩ ⟨α |113, (42) which satisfies ∫ C Πα d2α = 11311H1 113 = 113, (43) simply because Ω 3 ⊂ Ω1. Thus, the POVM {Πα} de- scribes a physically acceptable simultaneous measure- ment of X3 and ˜P3. Since Case 4 is i...

    work page

  10. [10]

    These four operators thus form a POVM on H and thus describe a generalized measurement on the quantum system described by H

    and the fact that H ⊂ H ′. These four operators thus form a POVM on H and thus describe a generalized measurement on the quantum system described by H. The Neumark extension theorem states that one can always run this procedure in reverse, starting with the four positive operators Π n on H that satisfy (

    work page

  11. [11]

    parti- cle in a box

    and ending up with four orthogonal projectors |n⟩ ⟨n|ontoH ′. The relation between H = C2 and H ′ = C4 can be understood in two physically distinct ways. If we view H as a subspace of H ′ (which is what we have assumed so far) then we can view our physical system as being one system described by a 4-dimensional Hilbert space, 9 but where we restrict atten...

    work page

  12. [12]

    paradox” of the type discussed in [ 2], in the sense that a mathematical property that at first sight might seem unimportant and “merely a technical detail

    and ( 10). We find ψ (x) = i ∑ m exp(iθ0) − 1 (θ0 + 2πm ) exp(ikmx)√ L , (52) where the right-hand side now is in the Hilbert space by construction. (The convergence is not point-wise, but in the L2 norm.) The expectation value ⟨ ˜P n 3 ⟩ is formally calculated by using the right-hand side expansion, which yields ⣨ ˜P n 3 ⟩ = 4 sin2(θ0/ 2) L2 ∑ m (θ0 + 2πm...

    work page 1991

  13. [13]

    Since there are well-documented dangers in Dirac’s formalism [ 2], in the following we will, when appropriate, indicate such dangers

    All this uses physicists’ notation, namely, Dirac’s bra - ket notation. Since there are well-documented dangers in Dirac’s formalism [ 2], in the following we will, when appropriate, indicate such dangers. See Ref. [ 6] and Ref. [8] for excellent pedagogical expositions of the rigor- ous mathematics behind quantum mechanics. Both books also provide a guid...

    work page

  14. [14]

    Mathematical surprises and Dirac's formalism in quantum mechanics

    F. Gieres, Mathematical surprises and Dirac’s formalism in quantum mechanics, Reports on Progress in Physics 63, 1893 (2000) , arXiv:quant-ph/9907069

    work page internal anchor Pith review Pith/arXiv arXiv 2000

  15. [15]

    Self-adjoint extensions of operators and the teaching of quantum mechanics

    G. Bonneau, J. Faraut, and G. Valent, Self-adjoint extensions of operators and the teaching of quantum mechanics, American Journal of Physics 69, 322 (2001) , arXiv:quant-ph/0103153

    work page internal anchor Pith review Pith/arXiv arXiv 2001

  16. [16]

    Canonical Quantization and Impenetrable Barriers

    P. Garbaczewski and W. Karwowski, Im- penetrable barriers and canonical quantiza- tion, American Journal of Physics 72, 924 (2004) , math-ph/0310023

    work page internal anchor Pith review Pith/arXiv arXiv 2004

  17. [17]

    H. E. Brandt, Positive operator valued mea- sure in quantum information processing, American Journal of Physics 67, 434 (1999)

    work page 1999

  18. [18]

    B. C. Hall, Quantum Theory for Mathematicians , Vol. 267 (Springer Science & Business Media, 2013)

    work page 2013

  19. [19]

    Becnel and A

    J. Becnel and A. Sengupta, The Schwartz space: Tools for quantum mechanics and infinite dimensional analysis, Mathematics 3, 527 (2015)

    work page 2015

  20. [20]

    Talagrand, What Is a Quantum Field Theory? (Cam- bridge University Press, 2022)

    M. Talagrand, What Is a Quantum Field Theory? (Cam- bridge University Press, 2022)

    work page 2022

  21. [21]

    S. J. Summers, On the Stone—von Neumann uniqueness theorem and its ramifications, in John von Neumann and the Foundations of Quantum Physics (Springer, 2001) pp. 135–152

    work page 2001

  22. [22]

    C. W. Helstrom, Quantum detection and estimation the- ory, Journal of Statistical Physics 1, 231 (1969)

    work page 1969

  23. [23]

    X. Wang, X. Zhan, Y. Li, L. Xiao, G. Zhu, D. Qu, Q. Lin, Y. Yu, and P. Xue, Generalized quantum mea- surements on a higher-dimensional system via quantum walks, Physical Review Letters 131, 150803 (2023)

    work page 2023

  24. [24]

    M. H. Al-Hashimi and U.-J. Wiese, Canonical quantiza- tion on the half-line and in an interval based upon an alternative concept for the momentum in a space with boundaries, Physical Review Research 3, 033079 (2021) , arXiv:2103.01715

    work page arXiv 2021

  25. [25]

    M. H. Al-Hashimi and U.-J. Wiese, Alternative mo- mentum concept for a quantum mechanical particle in a box, Physical Review Research 3, L042008 (2021) , arXiv:2012.09596

    work page arXiv 2021

  26. [26]

    Judge, On the uncertainty relation for lz and φ, Physics Letters 5, 189 (1963)

    D. Judge, On the uncertainty relation for lz and φ, Physics Letters 5, 189 (1963) . 14

    work page 1963

  27. [27]

    Judge, On the uncertainty relation for angle variabl es, Il Nuovo Cimento 31, 332 (1964)

    D. Judge, On the uncertainty relation for angle variabl es, Il Nuovo Cimento 31, 332 (1964)

    work page 1964

  28. [28]

    Bouten, N

    M. Bouten, N. Maene, and P. V. Leuven, On an uncertainty relation for angular variables, Il Nuovo Cimento 37, 1119 (1965)

    work page 1965

  29. [29]

    A. A. Evett and H. M. Mahmoud, Uncertainty relation with angle variables, Il Nuovo Cimento 38, 295 (1965)

    work page 1965

  30. [30]

    Vourdas, Quantum systems with finite Hilbert space, Reports on Progress in Physics 67, 267 (2004)

    A. Vourdas, Quantum systems with finite Hilbert space, Reports on Progress in Physics 67, 267 (2004)

    work page 2004

  31. [31]

    In some cases the construction of a physical separable Hilbert space as an appropriate subspace of an initial non-separable Hilbert space for infinitely many degrees of freedom, alluded to in the Appendix on QFT, depends explicitly on the dynamics, namely, when the choice be- tween unitarily inequivalent representations is informed by the dynamics

    work page

  32. [32]

    She and H

    C. She and H. Heffner, Simultaneous measurement of noncommuting observables, Physical Review 152, 1103 (1966)

    work page 1966

  33. [33]

    Kraus, A

    K. Kraus, A. B¨ ohm, J. D. Dollard, and W. Wootters, States, Effects, and Operations Fundamental Notions of Quantum Theory: Lectures in Mathematical Physics at the University of Texas at Austin (Springer, 1983)

    work page 1983

  34. [34]

    S. J. van Enk, Photodetector figures of merit in terms of POVMs, Journal of Physics Communications 1, 045001 (2017)

    work page 2017

  35. [35]

    Daubechies, A

    I. Daubechies, A. Grossmann, and Y. Meyer, Pain- less nonorthogonal expansions, Journal of Mathematical Physics 27, 1271 (1986)

    work page 1986

  36. [36]

    We use abbreviated notation here: ⟨Q†φ|ψ⟩ stands for the inner product ( Q∗(φ), ψ) in math notation

    work page

  37. [37]

    Linear stochastic wave-equations for continuously measured quantum systems

    P. Goetsch and R. Graham, Linear stochastic wave equations for continuously measured quan- tum systems, Physical Review A 50, 5242 (1994) , arXiv:gr-qc/9404054

    work page internal anchor Pith review Pith/arXiv arXiv 1994

  38. [38]

    H. M. Wiseman, Quantum trajectories and quantum measurement theory, Quantum and Semiclassical Optics: Journal of the European Optical Society Part B 8, 205 (1996)

    work page 1996

  39. [39]

    C. S. Jackson and C. M. Caves, Simultaneous measurements of noncommuting observables: Pos- itive transformations and instrumental Lie groups, Entropy 25, 1254 (2023) , arXiv:2306.06167

    work page arXiv 2023

  40. [40]

    C. S. Jackson and C. M. Caves, How to perform the coherent measurement of a curved phase space by con- tinuous isotropic measurement. i. Spin and the Kraus- operator geometry of SL(2, C), Quantum 7, 1085 (2023) , arXiv:2107.12396

    work page arXiv 2023

  41. [41]

    C. S. Jackson and C. M. Caves, Simultaneous mo- mentum and position measurement and the instrumen- tal Weyl-Heisenberg group, Entropy 25, 1221 (2023) , arXiv:2306.01045

    work page arXiv 2023

  42. [42]

    M. J. W. Hall, The quantum description of optical phase, Quantum Optics: Journal of the European Optical Soci- ety Part B 3, 7 (1991)

    work page 1991

  43. [43]

    K. Blow, R. Loudon, S. J. Phoenix, and T. Shep- herd, Continuum fields in quantum optics, Physical Review A 42, 4102 (1990)

    work page 1990

  44. [44]

    Separable Hilbert space in Loop Quantum Gravity

    W. Fairbairn and C. Rovelli, Separable Hilbert space in loop quantum gravity, Journal of Mathematical Physics 45, 2802 (2004) , arXiv:gr-qc/0403047

    work page internal anchor Pith review Pith/arXiv arXiv 2004

  45. [45]

    E. Witten, Why does quantum field theory in curved spacetime make sense? and what happens to the algebra of observables in the thermodynamic limit?, in Dialogues Between Physics and Mathematics: C. N. Yang at 100 (Springer Nature, 2022) p. 241, arXiv:2112.11614

    work page arXiv 2022

  46. [46]

    Haag, On quantum field theories, Matematisk-fysiske Meddelelser 29, no

    R. Haag, On quantum field theories, Matematisk-fysiske Meddelelser 29, no. 12, 1 (1955)

    work page 1955

  47. [47]

    Roman, Introduction to Quantum Field Theory (Wi- ley, 1969)

    P. Roman, Introduction to Quantum Field Theory (Wi- ley, 1969)

    work page 1969

  48. [48]

    R. F. Streater and A. S. Wightman, PCT, Spin and Statistics, and All That , Vol. 30 (Princeton University Press, 2000)

    work page 2000

  49. [49]

    Even though [the Hamiltonians for the two field theories] both make perfect sense, it is the (implicit) assumption that they operate on the same state space that does not

    As Talagrand [ 8] says on page 356: “Even though [the Hamiltonians for the two field theories] both make perfect sense, it is the (implicit) assumption that they operate on the same state space that does not.”

    work page

  50. [50]

    Simulating quantum field theory with a quantum computer

    J. Preskill, Simulating quantum field theory with a quan - tum computer, in The 36th Annual International Sym- posium on Lattice Field Theory - LATTICE2018 (2019) p. 024, arXiv:1811.10085

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  51. [51]

    C. W. Bauer, Z. Davoudi, A. B. Balantekin, T. Bhattacharya, M. Carena, W. A. De Jong, P. Draper, A. El-Khadra, N. Gemelke, M. Hanada, et al. , Quantum simulation for high-energy physics, PRX Quantum 4, 027001 (2023) , arXiv:2204.03381

    work page arXiv 2023