Pith. sign in

REVIEW 3 major objections 5 minor 34 references

Hunt's Hypothesis (H) for Markov Processes: Survey and Beyond

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read For degenerate multidimensional Lévy processes with finite total jump mass outside the Gaussian range, Hunt's hypothesis (H) holds exactly when the projection onto the residual subspace satisfies (H).

desk verdict Useful survey plus a plausible but not fully self-contained new projection criterion; Theorem 3.6 depends on an ambiguous 'drift coefficient' from an unpublished preprint. read the letter →

arxiv 1908.06825 v1 pith:H6646Z4E submitted 2019-08-19 math.PR

Ze-Chun Hu , Wei Sun This is my paper · ORCID
classification math.PR MSC 60G5160J4560J2531C15
keywords Hunt'shypothesis(H)LévyprocessessemipolarsetspolarprojectionsenergyGetoor'sconjecturepotentialtheory
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper surveys what is known about Hunt's hypothesis (H) for Markov processes — the assertion that every semipolar set is polar, i.e. that countable unions of sets essentially never hit at time zero are essentially never hit at all — and adds new criteria for multidimensional Lévy processes. Its central new theorem is a projection criterion: when the Gaussian covariance matrix $Q$ is degenerate and the Lévy measure has finite total mass outside the Gaussian range $\sqrt{Q}\,\mathbb{R}^n$, the process satisfies (H) if and only if its projection onto the orthogonal complement of $\sqrt{Q}\,\mathbb{R}^n$ satisfies (H), which is also equivalent to every one-dimensional projection satisfying (H). The proof uses the Lévy–Itô decomposition to split off a compound Poisson residual and shows that (H) for the residual projection forces the adjusted drift into the Gaussian range, where an earlier equivalence applies. A second set of results shows that finite energy for a product of independent Lévy processes passes to its marginals, and that a singular marginal forces the other marginal to avoid semipolar sets. The survey also records that Getoor's conjecture — that essentially all Lévy processes except the extremely nonsymmetric uniform-motion cases satisfy (H) — remains open, with these results as partial steps.

What carries the argument

The central machinery is the Lévy–Itô decomposition combined with orthogonal projection. Hunt's hypothesis (H) is the statement that every semipolar set (a countable union of sets that are essentially never hit at time zero) is polar (essentially never hit at any finite time). For the main theorem, the process is written $X_t = X^{(1)}_t + X^{(2)}_t$, where $X^{(1)}$ carries the Gaussian part and the compensated jumps inside $\sqrt{Q}\,\mathbb{R}^n$, and $X^{(2)}$ is a compound Poisson process built from jumps outside that subspace; the finite-mass condition makes $X^{(2)}$ compound Poisson. Projecting onto $(\sqrt{Q}\,\mathbb{R}^n)^\perp$ removes the Gaussian part, leaving a drift plus compound Poisson process, and validity of (H) for that projection forces the adjusted drift $b'$ into $\sqrt{Q}\,\mathbb{R}^n$. The energy results use the $\lambda$-energy integrand $\operatorname{Re}(1/(\lambda+\psi(z)))$ and Kanda's comparison estimates to transfer finiteness and vanishing limits between a product process and its marginals; the named Kanda–Forst condition, the inequality $|\operatorname{Im}\psi(z)| \le M(1+\operatorname{Re}\psi(z))$, controls the skew part of the Lévy exponent and is used as one factor's hypothesis in the energy theorem.

What would settle it

Exhibit a degenerate-$Q$ Lévy process with $\mu(\mathbb{R}^n \setminus \sqrt{Q}\,\mathbb{R}^n)<\infty$ such that its projection onto $(\sqrt{Q}\,\mathbb{R}^n)^\perp$ satisfies (H) while the adjusted drift $b'$ does not lie in $\sqrt{Q}\,\mathbb{R}^n$; the theorem predicts that no such process exists, so producing one, or directly showing that $X$ fails (H) while the residual projection satisfies it, would refute the criterion.

Watch

Extended reading notes

Core claim

The paper's core new result is Theorem 3.6. Let $X$ be an $\mathbb{R}^n$-valued Lévy process with degenerate Gaussian covariance $Q$, and suppose the Lévy measure $\mu$ satisfies $\mu(\mathbb{R}^n \setminus \sqrt{Q}\,\mathbb{R}^n)<\infty$, so that jumps outside the Gaussian range form a compound Poisson process. Then three statements are equivalent: (i) $X$ satisfies Hunt's hypothesis (H); (ii) for every one-dimensional subspace $A$, the projection of $X$ onto $A$ satisfies (H); (iii) the projection of $X$ onto $(\sqrt{Q}\,\mathbb{R}^n)^\perp$ satisfies (H). The direction (i)$\Rightarrow$(ii),(iii) is the earlier projection lemma; the new content is that (ii) or even (iii) alone forces (H) for the full process. The proof decomposes $X_t=X^{(1)}_t+X^{(2)}_t$, shows that (H) for the residual projection forces the adjusted drift $b'$ to lie in $\sqrt{Q}\,\mathbb{R}^n$, and invokes the earlier equivalence (Theorem 2.6) between (H) and that drift condition. Theorem 3.7 gives a companion statement with any subspace $S$ satisfying $\sqrt{Q}\,\mathbb{R}^n\subsetneq S\subsetneq \mathbb{R}^n$ and $\mu(\mathbb{R}^n\setminus S)<\infty$. The energy part adds that for a product $Z_t=(X_t,Y_t)$ of independent Lévy processes with resolvent densities, finite $\lambda$-energy of a compactly supported measure passes to its marginals; that a finite-energy product measure with a singular factor has the other factor charging no semipolar set; and that if $X$ satisfies (H) and $Y$ satisfies the Kanda–Forst condition, the product measure's $\lambda$-energy tends to zero as $\lambda\to\infty$.

Load-bearing premise

The load-bearing premise is the finite-mass condition $\mu(\mathbb{R}^n \setminus \sqrt{Q}\,\mathbb{R}^n)<\infty$, which ensures that the residual process $X^{(2)}$ is a compound Poisson process; without that condition the projection equivalence of Theorem 3.6 is not established.

Editorial extensions

If this is right

  • Under the finite-mass condition, (H) for a degenerate multidimensional Lévy process is equivalent to (H) for its projection onto $(\sqrt{Q}\,\mathbb{R}^n)^\perp$.
  • In the same setting, (H) for every one-dimensional projection is equivalent to (H) for the single residual projection, so a single projection suffices.
  • The coordinate-axis version fails: there is a two-dimensional process whose projections on the two coordinate axes both satisfy (H) while the process itself does not, because the projection on the diagonal $y=-x$ is a uniform motion.
  • For products of independent Lévy processes, finite $\lambda$-energy of a product measure forces finite $\lambda$-energy on each marginal, and a singular marginal forces the other marginal to avoid semipolar sets.
  • If one factor of a product satisfies (H) and the other satisfies the Kanda–Forst skew-control condition, then the $\lambda$-energy of any finite-energy product measure vanishes as $\lambda\to\infty$.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Extension: for concrete models, the projection criterion reduces (H) to a linear-algebra check — whether $b'$ lies in $\sqrt{Q}\,\mathbb{R}^n$ — plus a one-dimensional check on the residual projection, whenever the finite-mass condition holds.
  • Extension: the coordinate-axis counterexample implies that any automated or experimental test of (H) by projections must scan all directions, or target the orthogonal complement, and cannot rely on coordinate axes alone.
  • Extension: the energy-transference propositions suggest that verifying Getoor's conjecture could be approached by decomposing a Lévy process into independent summands with complementary energy behavior, if such decompositions can be constructed.
  • Extension: the boundary of Theorem 3.6 can be probed by building degenerate-$Q$ examples with $\mu(\mathbb{R}^n \setminus \sqrt{Q}\,\mathbb{R}^n)=\infty$; the paper leaves the projection equivalence open in that case.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper surveys the state of the art on Hunt's hypothesis (H) for Markov processes and Getoor's conjecture for Lévy processes, and contributes new results on multidimensional Lévy processes from the viewpoints of projections and energy. The main new claims are Theorem 3.6, an equivalence between (H) for a degenerate-Gaussian Lévy process with finitely many jumps outside the range of the Gaussian part and (H) for all of its one-dimensional projections; Theorem 3.7, a related criterion using a subspace strictly between the Gaussian range and the whole space; and several energy results in Section 4, including Propositions 4.5-4.8 on product processes.

Significance. If Theorem 3.6 is correct, it gives a genuinely useful reduction: for a wide class of multidimensional Lévy processes, Hunt's hypothesis can be checked on one-dimensional projections, complementing the classical Kanda-Forst and Rao criteria. The survey portion is well organized and collects many results from the authors' earlier work. The proof of Lemma 3.4 is detailed and correct, and Example 3.5 correctly shows that checking coordinate-axis projections alone is insufficient. However, the central new theorem depends on an unproved and unpublished proposition, and the proof of Proposition 4.8 contains a nontrivial analytical gap; these issues prevent the paper from being accepted in its present form.

major comments (3)
  1. [§3.2.2, Theorem 3.6] Both converses in Theorem 3.6 rely on Proposition 3.2, which is stated in §3.1 without proof and cited to the authors' unpublished preprint [23]. The statement 'its drift coefficient equals zero' is ambiguous: for a process of the form d t plus a compound Poisson process, the coefficient a in the Lévy-Khintchine representation (a,0,µ) includes the small-jump compensation term and is generally nonzero even when the canonical trajectory drift d is zero. The proof of Theorem 3.6 silently uses the canonical-drift reading to conclude b'_i=0 and P2 b'=0. Please include a precise definition of 'drift coefficient' in Proposition 3.2 and either prove the proposition in this paper or cite a published version; as it stands, both (ii)⇒(i) and (iii)⇒(i) of Theorem 3.6 rest on an unverified convention.
  2. [§4.1, Proposition 4.8] The proof after equation (4.12) uses a reverse Fatou step to move limsup inside the y-integral. The integrand involves E^{λ+ReΨ(y)}_X(η), but ReΨ(y) ≤ 0 and can be unbounded below, so λ+ReΨ(y) is not positive for all y; the energy E^λ_X is defined only for positive parameter, and the monotonicity of λ ↦ E^λ_X(η) cannot be applied pointwise on the set where λ+ReΨ(y) ≤ 0. A uniform domination argument or a truncation splitting the y-integral is needed. As written, the conclusion lim_{λ→∞} E^λ_Z(µ)=0 is not fully justified.
  3. [§3.2.3, Theorem 3.7] The final step 'Following the proof of [20, Theorem 1.2, (ii) ⇒ (i)], we can show that F is a polar set of X' is a deferred key argument rather than a proof. In fact, once (X^(1)) is known to satisfy (H) and X^(2) is compound Poisson, the desired conclusion follows directly from Theorem 2.14 (or Proposition 3.9) together with independence of X^(1) and X^(2). Please replace the deferred reference with an explicit citation of Theorem 2.14 and spell out the independence argument.
minor comments (5)
  1. [§3.1, Proposition 3.3] The phrase 'drift coefficient of Y equals zero' inherits the same ambiguity as Proposition 3.2; please use the notation b' from §2.1.2 so that the canonical drift is unambiguous.
  2. [§2.1.2, Theorem 2.6] The definitions of b and b' appear just before Theorem 2.6 and are used later without being recalled; a short reminder in §3 would improve readability.
  3. [References] Reference [23] is listed as an arXiv preprint from 2019; please provide its publication status or a more recent citation if it has appeared.
  4. [§5, Question 6 and references] There are several typographical/OCR issues, e.g., 'Dose any' in Question 6, 'Bwownian' in reference [8], and broken ligatures such as 'exis ting' in the abstract; these should be corrected in the final version.
  5. [§4.1, Proposition 4.6] The sentence 'If ξ charges a semipolar set, then it charges a compact set K ... such that K ⊂ {y ∈ R^m : E^y[exp(-λT_K)] < δ}' would benefit from a brief justification using the definition of semipolar sets as countable unions of thin sets and the regularity of capacity.

Circularity Check

1 steps flagged · score 4.0 of 10

Theorem 3.6's converses lean on Proposition 3.2 from the authors' unpublished preprint [23] without proof; the rest of the derivation is independent.

  1. self citation load bearing [Section 3.2.2, proof of Theorem 3.6, after decomposition (3.4); Proposition 3.2 statement]
    "Proposition 3.2 ([23, Proposition 5.3]) Let X be a Lévy process on Rn (n ≥ 1) with Lévy-Khintchine exponent (a, 0, µ) satisfying ∫ Rn(|x| ∧ 1)µ(dx)< ∞. If X satisfies (H), then its drift coefficient equals zero. ... By (ii), we find that the projection process (PiXt)t≥0 satisfies (H) and hence b′ i = 0 for any i = k +1,...,n."

    The inference 'projection process satisfies (H) and hence b'_i = 0' is exactly Proposition 3.2 applied to the finite-variation projection PiX_t = (P_i b')t + PiX^(2)_t, where PiX^(2)_t is compound Poisson. The paper does not prove Proposition 3.2 here; it cites [23, Proposition 5.3], an unpublished preprint by the same authors. Both converse directions of the paper's main projection theorem, (ii) implies (i) and (iii) implies (i), therefore rest on a load-bearing self-citation whose proof is not included in the present paper. This is not a full reduction of the theorem to its own conclusion, because Proposition 3.2 does not assume Theorem 3.6, but the central new implication is not self-contained and depends on an unverified-in-this-paper result from the authors' own prior work.

full rationale

The paper is a survey plus new projection and energy results. The central new theorem, Theorem 3.6, is derived through a Lévy-Itô decomposition into a Gaussian part on sqrt(Q)R^n and a compound-Poisson residual, then by showing the residual drift vanishes and invoking the published Theorem 2.6 from [20]. This chain is not circular: Theorem 2.6 states an independent characterization of (H) under the same finiteness condition, and its assumptions do not include the projection criterion being proved. Lemma 3.1 from [22], used for the forward direction, is likewise a published result with its own proof. The counterexample in Example 3.5 is self-contained. The energy section applies Theorem 4.2 from [24] as an input rather than assuming its conclusion. No fitted parameter is renamed a prediction, and no displayed equation reduces to its own input. The one load-bearing step that raises the score is the use of Proposition 3.2 from the authors' unpublished preprint [23] to conclude that a one-dimensional projection satisfying (H) must have zero drift; without that proposition, the proof of Theorem 3.6's converses is incomplete. The unpublished status and the ambiguity of 'drift coefficient' flagged in Proposition 3.2 are correctness risks rather than circularity, but they make the self-citation load-bearing. Hence a score of 4 is appropriate rather than 0.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

The paper introduces no free parameters or invented entities. It relies on a substantial body of prior results, many from the same authors, which are treated as black boxes. The most critical external inputs are Theorem 2.6 from [20], Proposition 3.2 from the preprint [23], and Kanda's energy lemmas.

assumptions (7)
  • standard math Standard potential theory of Markov processes (Blumenthal-Getoor framework, resolvents, capacities)
    Used throughout Section 2 and in the definitions of thin, semipolar, polar sets.
  • domain assumption Kanda-Forst theorem (Theorem 2.1)
    Used as a sufficient condition for (H) in many of the paper's proofs; stated without proof.
  • domain assumption Rao's extension of the Kanda-Forst condition (Theorem 2.3)
    Used for stable processes and in the authors' earlier work.
  • domain assumption The authors' prior Theorem 2.6 (from [20])
    Establishes equivalence between (H) and condition (S) under μ(R^n \ √Q R^n)<∞; used in proofs of Theorems 3.6 and 3.7.
  • domain assumption Proposition 3.2 (from the preprint [23])
    States that for processes of the form drift plus compound Poisson with finite variation, (H) forces zero drift; used in Theorems 3.6 and 3.7. [23] is an arXiv preprint.
  • domain assumption Kanda's lemmas in [28]
    Used in Section 4 for energy and capacity estimates (Lemmas 2.1-2.4 of Kanda 1991).
  • domain assumption Assumption that processes have resolvent densities
    Made in Section 4 and in several surveyed theorems; needed for the equivalence between (H) and energy limits.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Hunt's Hypothesis (H) for Markov Processes: Survey and Beyond." pith.science (2026). https://pith.science/paper/H6646Z4E

@misc{pith2026190806825,
  author       = {Pith},
  title        = {Pith review of: Hunt's Hypothesis (H) for Markov Processes: Survey and Beyond},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H6646Z4E}},
  note         = {Machine review of arXiv:1908.06825}
}
read the original abstract

The goal of this paper is threefold. First, we survey the existing results on Hunt's hypothesis (H) for Markov processes and Getoor's conjecture for L\'{e}vy processes. Second, we investigate (H) for multidimensional L\'{e}vy processes from the viewpoints of projections and energy, respectively. Third, we present a few open questions for further study.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 34 canonical work pages

  1. [23]

    Two Theorems on Hunt's Hypothesis (H) for Markov Processes

    Hu Z.-C., Sun W., Wang L.-F.: Two theorems on Hunt’s hypothesis (H ) for Markov processes. arXiv:1903.00050 (2019)

    work page Pith review arXiv 2019

  2. [20]

    Hu Z.-C., Sun W.: Hunt’s hypothesis (H) and Getoor’s conjecture for L´ evy processes. Stoch. Proc. Appl. 122, 2319-2328 (2012)

    work page 2012

  3. [1]

    Cambridge University Press, Camb ridge (1996)

    Bertoin J.: L´ evy Processes. Cambridge University Press, Camb ridge (1996)

    work page 1996

  4. [2]

    Academic Press, New York and London (1968)

    Blumenthal R.M., Getoor R.K.: Markov Processes and Potential Th eory. Academic Press, New York and London (1968). 25

    work page 1968

  5. [3]

    Blumenthal R.M., Getoor R.K.: Dual processes and potential theo ry. Proc. 12th Biennial Seminar of the Canadian Math. Congress, 137-156 (1970)

    work page 1970

  6. [4]

    Bochner S.: Harmonic Analysis and the Theory of Probability. Univ. California Press, Berkeley and Los Angeles (1955)

    work page 1955

  7. [5]

    S´ eminare de Probabilit´ es V, Lect

    Bretagnolle J.: R´ esults de Kesten sur les processus ` a accroiss ements ind´ ependants. S´ eminare de Probabilit´ es V, Lect. Notes in Math., Vol. 191, Springer-Verlag, Berlin, 21-36 (1971)

    work page 1971

  8. [6]

    Fitzsimmons P.J.: On the equivalence of three potential principles f or right Markov pro- cesses. Probab. Th. Rel. Fields 84, 251-265 (1990)

    work page 1990

Show all 34 references
  1. [7]

    P otential Anal

    Fitzsimmons P.J.: On the quasi-regularity of semi-Dirichlet forms. P otential Anal. 15, 151-185 (2001)

    2001

  2. [8]

    Fitzsimmons P.J.: Gross’s Bwownian motion fails to satisfy the polarit y principle. Rev. Roumaine Math. Pures Appl. 59, 87-91 (2014)

    2014

  3. [9]

    Fitzsimmons P.J., Kanda M.: On Choquet’s dichotomy of capacity for Markov processes. Ann. Probab. 20, 342-349 (1992)

    1992

  4. [10]

    Forst G.: The definition of energy in non-symmetric translation in variant Dirichlet spaces. Math. Ann. 216, 165-172 (1975)

    1975

  5. [11]

    De Gruyter (2011)

    Fukushima M., Oshima Y., Takeda M.: Dirichlet Forms and Symmetric M arkov Processes (Second revised and extended edition). De Gruyter (2011)

    2011

  6. [12]

    Proba- bility Theory and Its Applications, XXVI, 4, 757-768 (1981)

    Glover J.: Energy and the maximum principle for nonsymmetric Hun t processes. Proba- bility Theory and Its Applications, XXVI, 4, 757-768 (1981)

    1981

  7. [13]

    Seminar on Sto chastic Processes, 1982, Birkh¨ auser, 195-202 (1983)

    Glover J.: Topics in energy and potential theory. Seminar on Sto chastic Processes, 1982, Birkh¨ auser, 195-202 (1983)

    1983

  8. [14]

    Glover J., Rao M.: Hunt’s hypothesis (H) and Getoor’s conjectur e. Ann. Probab. 14, 1085-1087 (1986)

    1986

  9. [15]

    Glover J., Rao M.: Nonsymmetric Markov processes and hypothe sis (H). J. Theor. Probab. l, 371-380 (1988)

    1988

  10. [16]

    Acta Math

    Han X.-F., Ma Z.-M., Sun W.: hˆh-transforms of positivity preserving semigroups and associated Makov processes. Acta Math. Sinica, English Series 27, 369-376 (2011)

    2011

  11. [17]

    Hansen W., Netuka I.: Hunt’s hypothesis (H) and triangle proper ty of the Green function. Expo. Math. 34, 95-100 (2016)

    2016

  12. [18]

    Hartman P., Wintner A.: On the infinitesimal generators of integr al convolutions. Amer. J. Math. 64, 273-298 (1942)

    1942

  13. [19]

    Hawkes J.: Potential theory of L´ evy processes. Proc. Lond on Math. Soc. 3, 335-352 (1979). 26

    1979

  14. [21]

    Hu Z.-C., Sun W.: Further study on Hunt’s hypothesis (H) for L´ e vy processes. Sci. China Math. 59, 2205-2226 (2016)

    2016

  15. [22]

    Hu Z.-C., Sun W.: Hunt’s Hypothesis (H) for the sum of two indepen dent L´ evy processes. Commun. Math. Stat. 6, 227-247 (2018)

    2018

  16. [24]

    Potential Anal

    Hu Z.-C., Sun W., Zhang J.: New results on Hunt’s hypothesis (H) fo r L´ evy processes. Potential Anal. 42, 585-605 (2015)

    2015

  17. [25]

    Kanda M.: Two theorems on capacity for Markov processes with stationary independent increments. Z. Wahrsch. verw. Gebiete 35, 159-165 (1976)

    1976

  18. [26]

    Kanda M.: Characterization of semipolar sets for processes wit h stationary independent increments. Z. Wahrsch. verw. Gebiete 42, 141-154 (1978)

    1978

  19. [27]

    Surikaiseki Kenkyusho Kokyuroku 502, 161-170 (198 3)

    Kanda M.: Characterization of semipolar sets for L´ evy proces ses by Fourier transform of measures. Surikaiseki Kenkyusho Kokyuroku 502, 161-170 (198 3)

  20. [28]

    Nagoya Math

    Kanda M.: Notes on energy for space-time processes over L´ e vy processes. Nagoya Math. J. 122, 63-74 (1991)

    1991

  21. [29]

    Memoirs of the American Mathematical Society, No

    Kesten H.: Hitting probabilities of single points for processes with stationary independent increments. Memoirs of the American Mathematical Society, No. 93 , American Mathemat- ical Society, Providence, R.I. (1969)

    1969

  22. [30]

    Port S.C., Stone C.J.: The asymmetric Cauchy process on the line. Ann. Math. Statist. 40, 137-143 (1969)

    1969

  23. [31]

    Rao M.: On a result of M. Kanda. Z. Wahrsch. verw. Gebiete 41, 3 5-37 (1977)

    1977

  24. [32]

    London Math

    Rao M.: On polar sets for L´ evy processes, J. London Math. So c. 35, 569-576 (1987)

    1987

  25. [33]

    Rao M.: Hunt’s hypothesis for L´ evy processes. Proc. Amer. Math. Soc. 104, 621-624 (1988)

    1988

  26. [34]

    Silverstein M.L.: The sector condition implies that semipolar sets ar e quasi-polar. Z. Wahrsch. verw. Gebiete 41, 13-33 (1977). 27

    1977

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.