REVIEW 2 major objections 4 minor 9 references
On definition of quantum tomography via the Sobolev embedding theorem
T0 review · 2 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read For kernels in a carefully chosen Sobolev space, optical tomograms and Wigner functions are well-defined functions, and the two standard tomogram formulas coincide.
desk verdict A useful sufficient condition for tomograms beyond Schwartz class, with one genuinely load-bearing proof gap in Lemma 2 that a serious revision should fix. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central object is the space $V(\mathbb{R}^{2n}) = W_2^{n+1}(\mathbb{R}^{2n}) \cap \mathcal{F}[W_2^{n+1}(\mathbb{R}^{2n})]$, consisting of functions that have $n+1$ Sobolev derivatives and whose Fourier transforms also have $n+1$ Sobolev derivatives. This symmetry makes $V$ invariant under Fourier and partial Fourier transforms, so the characteristic function and the Wigner function inherit membership in $V$ from the kernel. The Sobolev embedding theorem then supplies the crucial trace operation: restricting a function in $V$ to an $n$-dimensional hyperplane yields a continuous, integrable function, and that trace is exactly the optical tomogram.
What would settle it
Check the step in Lemma 2 by testing whether every compactly supported smooth function $f$ with $|x|^{n+1}\mathcal{F}[f]\in L^2(\mathbb{R}^{2n})$ satisfies $\int_{\mathbb{R}^n} |F_\alpha(t)|^2 \prod_{j=1}^n (t_j^2+1)\,dt<\infty$ for each hyperplane $\alpha$; a single smooth function in $V$ whose hyperplane trace is not in $L^1$ would make $\omega_\rho$ non-integrable and disprove Theorem 1.
Extended reading notes
Core claim
The paper's central claim is Theorem 1: if the kernel $\rho(q,q')$ of an integral operator belongs to $V(\mathbb{R}^{2n})$, then both the characteristic function $f_\rho$ and the Wigner function $W_\rho$ belong to $V(\mathbb{R}^{2n})$, and the optical quantum tomogram defined by the partial Fourier transform formula (13) and by the Radon-transform formula (14) is correctly defined, with $\omega_\rho(\cdot,\alpha)\in C(\mathbb{R}^n)\cap L^1(\mathbb{R}^n)$. The two formulas agree by the Fourier slice theorem. The proof rests on the Sobolev embedding theorem, which supplies a trace of a function in $V(\mathbb{R}^{2n})$ to an $n$-dimensional hyperplane, and on the Fourier invariance of $V$, which keeps every intermediate object in the same regularity class. This extends tomography from Schwartz-class kernels to a broader function class while keeping the inversion formulas intact.
Load-bearing premise
The proof assumes that restricting a function in $V(\mathbb{R}^{2n})$ to a hyperplane preserves the weighted $L^2$ growth that makes the restriction integrable, and the cited Sobolev trace theorem does not by itself justify weighted restrictions.
Editorial extensions
If this is right
- Every kernel in $V(\mathbb{R}^{2n})$ has an optical tomogram that is both continuous and integrable, so the tomogram can be treated as an ordinary probability density without extra regularity assumptions on the state.
- The characteristic function, Wigner function, and tomogram form a closed cycle of invertible transforms: formula (15) recovers the characteristic function from the tomogram, and formula (12) recovers the kernel from the characteristic function.
- Transition probabilities between states with kernels in $V$ can be computed directly from their tomograms by the absolutely convergent integral (17).
- For pure states whose wavefunction lies in $V(\mathbb{R}^n)$, the tomogram equals the squared modulus of the fractional Fourier transform of the wavefunction, extending this relation beyond Schwartz functions.
- The construction does not use positivity of the kernel, so the tomogram is well defined for a class of non-positive trace-class-type operators as well as for quantum states.
Reading between the lines
- A natural extension the paper does not explore is whether the Sobolev index $n+1$ can be lowered; testing kernels in $W_2^s$ for $s<n+1$ would show whether the condition is close to optimal.
- The weighted trace step in Lemma 2 could be replaced by a direct weighted trace inequality for the product weight $\prod_{j=1}^n (t_j^2+1)$; proving or disproving such an inequality would clarify whether the integrability conclusion is a genuine consequence of the Sobolev embedding or an artifact of the proof.
- The same $V$-space method can likely be adapted to other tomographic schemes, such as symplectic tomography with arbitrary linear canonical transformations, where hyperplanes are replaced by more general Lagrangian planes.
- A numerical experiment with a kernel on the boundary of $V$, such as one whose trace decays exactly like $|t|^{-(n+1)}$, could test in low dimensions how sharp the $L^1$ integrability conclusion actually is.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proposes a Sobolev-type space V(R^{2n}) = W^{n+1}_2(R^{2n}) ∩ F[W^{n+1}_2(R^{2n})] and claims that for density-matrix kernels in this space, the Wigner function and the optical quantum tomogram are well-defined and related by the Radon-transform formula. The main result (Theorem 1) states that for ρ ∈ V(R^{2n}), the tomogram defined by (13) and (14) exists in C(R^n) ∩ L1(R^n). The paper also states a transition-probability formula (Theorem 2) and discusses pure states and fractional Fourier transforms (Theorem 3). The central technical tool is Lemma 2, which asserts that the trace of a function in V(R^{2n}) to a hyperplane belongs to C(R^n) ∩ L1(R^n).
Significance. If the central claim held as stated, V(R^{2n}) would be a useful and fairly broad class of kernels (properly containing the Schwartz space) under which tomographic and Wigner-function descriptions are rigorously justified. The approach via Sobolev embedding is natural, and the paper is explicit and self-contained in its definitions. However, the proof of the key trace lemma contains an unsupported step, and a secondary theorem rests on an incorrect equivalence. The main contribution is therefore not yet established; the paper requires substantial revision before its claims can be accepted.
major comments (2)
- [Lemma 2, proof] The step 'Since f ∈ V(R^{2n}) we get F[f] ∈ W^{n+1}_2(R^{2n}). It gives rise |x|^{n+1}f(x) lies in L2(R^{2n}). Hence, |t|^{n+1}Fα(t) belongs to L2(R^n)' is not justified by the cited Sobolev embedding theorem. The trace theorem quoted in Section 2 provides a trace for functions in W^{n+1}_2(R^{2n}); it does not apply to the weighted function |x|^{n+1}f, which is only known to lie in L2(R^{2n}). Restrictions of general L2 functions to lower-dimensional hyperplanes are not defined, and while the continuity of f makes the pointwise restriction of |x|^{n+1}f well-defined, its L2 norm on the hyperplane is not controlled by the global L2 norm. This unsupported 'hence' is the load-bearing step for the Cauchy-Schwarz estimate (9) and for the conclusions Fα ∈ C(R^n) ∩ L1(R^n) and, consequently, ωρ ∈ C(R^n) ∩ L1(R^n) in Theorem 1. The authors should either supply a proof of this weighted trace statement using the full V(R^{2n}) structure, or modify the definition of V so that the trace of the weighted function is provably L2.
- [Theorem 3, proof] The assertion 'The condition ψ ∈ V(R^n) is equivalent to ρ ∈ V(R^{2n})' for ρ(x,y)=ψ(x)ψ*(y) is not correct as stated. For n=1, V(R) requires only second-order Sobolev regularity, while V(R^2) requires third-order derivatives in the two variables. A term such as ∂_x^3ρ = ψ'''ψ* need not be square-integrable for ψ ∈ W^2_2(R), so the product kernel may fail to belong to W^3_2(R^2). Since the proof of Theorem 3 relies on this equivalence to apply Theorem 1, the theorem is not established under the stated hypothesis. The authors should either prove a suitable product property for V or prove Theorem 3 by a direct argument that does not require ρ ∈ V(R^{2n}).
minor comments (4)
- [Theorem 2, proof] The justification that the λ-integral in (17) converges is incomplete. The statement that fρ1(λ cosα,λ sinα) and fρ2(λ cosα,λ sinα) belong to L1(R^n) in λ does not by itself control ∫ |λ| |f1| |f2| dλ. The authors should invoke interpolation to obtain |λ|^{1/2}f ∈ L2 from |λ|^{n+1}f ∈ L2 and f ∈ L2, which would make the iterated integral well-defined.
- [Lemma 2, statement] The statement contains a typo: 'y1,...y2' should read 'y1,...,yn'.
- [Introduction] The phrase 'for all ρ > 0 with Tr{ρ}=1 and the kernels ρ(q,q′) ∈ L2(R2n)' is slightly ambiguous; it would be clearer to write 'for all positive trace-class operators ρ with unit trace whose kernels lie in L2(R2n)'.
- [Section 2, definition of W^ν_2] The definition of W^ν_2 via |x|^ν F[ψ] ∈ L2, together with ψ ∈ L2, is equivalent to the standard Bessel-potential space for integer ν, but the paper does not state this. A brief remark would help avoid confusion, especially because the subsequent use of the Sobolev embedding theorem assumes the standard Sobolev regularity.
Circularity Check
No significant circularity: the paper's derivation is self-contained and all load-bearing steps cite external Sobolev embedding theorems, not the paper's own conclusions.
full rationale
The paper claims sufficient conditions on density-matrix kernels ρ ∈ V(R^{2n}) under which Wigner functions, optical quantum tomograms, and the linking formulas are correctly defined. The central argument is: (i) define V(R^{2n}) = W^{n+1}_2(R^{2n}) ∩ F[W^{n+1}_2(R^{2n})]; (ii) use Lemma 1 to show V is invariant under Fourier and partial Fourier transforms; (iii) use the classical Sobolev embedding theorem, explicitly attributed to Sobolev [5] and Nikol'skii [6], to obtain trace and continuity properties; and (iv) from these conclude F_α ∈ C(R^n) ∩ L^1(R^n), giving the tomogram ω_ρ ∈ C(R^n) ∩ L^1(R^n). The tomogram is not defined in terms of the desired conclusion; it is defined by the Fourier transform formula (13) and the Radon-transform formula (14), whose coincidence follows from the external Fourier slice theorem [8]. No parameters are fitted, no prediction is made from a fitted subset, and no load-bearing step is justified by a citation to the present authors' prior work. The only potentially vulnerable point is Lemma 2's assertion that |t|^{n+1}F_α(t) ∈ L^2(R^n) follows from |x|^{n+1}f(x) ∈ L^2(R^{2n}) and the cited Sobolev trace theorem; that is a mathematical proof gap or correctness concern about the applicability of the trace theorem to weighted spaces, not a circularity. The derivation does not presume the tomogram regularity it purports to establish, and the supporting results are independently stated external theorems. Therefore the circularity score is 0.
Assumptions & free parameters
assumptions (3)
- standard math Sobolev embedding theorem and the trace theorem as stated in Section 2, citing Sobolev 1938 and Nikol'skii 1961.
- domain assumption For a state with L2 kernel, the characteristic function lies in L2 ∩ C and formula (11) extends to Hilbert-Schmidt operators, as borrowed from Holevo [3].
- domain assumption The density operators under study have kernels in V(R^{2n}); positivity is explicitly waived.
Cite this review
Pith. "Pith review of On definition of quantum tomography via the Sobolev embedding theorem." pith.science (2026). https://pith.science/paper/JISGPGJL
@misc{pith2026190806793,
author = {Pith},
title = {Pith review of: On definition of quantum tomography via the Sobolev embedding theorem},
year = {2026},
howpublished = {\url{https://pith.science/paper/JISGPGJL}},
note = {Machine review of arXiv:1908.06793}
}
read the original abstract
We obtain sufficient conditions on kernels of quantum states under which Wigner functions, optical quantum tomograms and linking their formulas are correctly defined. Our approach is based upon the Sobolev embedding theorem. The transition probability formula and the fractional Fourier transform are discussed in this framework.
Reference graph
Works this paper leans on
-
[1]
Symplectic tomography as classical approach to quantum systems
S. Mancini, V. I. Manko, and P. Tombesi, “Symplectic tomography as classical approach to quantum systems”, Phys. Lett. A, 213 1 (1996)
work page 1996
-
[2]
An introduction to the tomographic picture of quantum mechanics
A. Ibort, V. I. Man’ko, G. Marmo, A. Simoni, and F. Ventriglia, “An introduction to the tomographic picture of quantum mechanics ”, Phys. Scr., 79, 065013 (2009)
work page 2009
-
[3]
A. S. Holevo, Probabilistic and statistical aspects of quantum theory (Edi- zioni della Normale, 2011)
work page 2011
-
[4]
M. Reed, B. Simon, Methods of Modern Mathematical Physics II. Fourier analysis, Self-Adjointness (Academic Press, 1975)
work page 1975
-
[5]
Sur un theoreme d’analyse fonctionnelle
S. Soboleff, “Sur un theoreme d’analyse fonctionnelle”, Rec. Mat h. [Mat. Sbornik] N.S., 4(46):3 ( 471–497 (1938)
work page 1938
-
[6]
S. M. Nikol’skii, “On imbedding, continuation and approximation the- orems for differentiable functions of several variables”, Russian M ath. Surveys, 16:5, 55–104 (1961)
work page 1961
-
[7]
Uber die Bestimmung von Funktionen durch ihre Integ ral- werte langs gewisser Mannigfaltigkeiten
J. Radon, “Uber die Bestimmung von Funktionen durch ihre Integ ral- werte langs gewisser Mannigfaltigkeiten”, Ber. Verh. Sachs. Akad ., 69, 262 (1917)
work page 1917
-
[8]
Helgason, The Radon transform (Birkhauser, Boston, Basel, Stuttgart, 1980)
S. Helgason, The Radon transform (Birkhauser, Boston, Basel, Stuttgart, 1980)
work page 1980
Show all 9 references
-
[9]
The fractional order Fourier transform and its app lication to quantum mechanics
V. Namias, “The fractional order Fourier transform and its app lication to quantum mechanics”, J. Inst. Math. Appl., 25, 241–265 (1980). 10
1980
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.