{"id":"a4d10b03-3b4b-4ac6-a38c-55a67b1db7b1","arxiv_id":"1908.07266","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For certain parameter ranges, normalized confluent hypergeometric, Lommel, and generalized Struve functions satisfy zf'/f ≺ e^z or 1+zf''/f' ≺ e^z in the unit disk.","lead":"New parameter conditions put confluent hypergeometric, Lommel, and Struve functions into the exponential starlike and convex families on the unit disk. The paper applies the same differential subordination method previously used for Bessel functions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2.3 as printed is ill-posed: it states (Phi-1)^{c/a} in K_e, but for non-integer c/a this is not analytic at 0; the proof and Example 2.5 actually establish (c/a)(Phi-1) in K_e.","rationale":"The reader's note already identified that Theorem 2.3 as printed states a power function while the proof and Example use the constant multiple; I agree with that observation and elevate it to the load-bearing concern. The paper's main contribution for the confluent hypergeometric case is the exponential-convexity theorem, and the printed statement is false, not merely awkward. My independent check of the surrounding estimates found no comparable internal inconsistency: the admissibility lower bounds in Sections 2-4 are arithmetically consistent once the intended normalized functions are used, and the stated conditions do appear to make the displayed constants nonnegative. The remaining risk is the one the reader emphasized: the proofs of Theorems 3.1, 3.2 and 4.3 depend on external lower bounds from [26] and [18] that are not proved in this paper. If those cited lemmas are correct and their hypotheses are met, the intended corrected theorems should hold. Because the verdict CONDITIONAL already correctly asks for the theorem statement to be fixed and for the external citations to be checked, my stress-test does not move the verdict.","tokens_in":18914,"tokens_out":34460,"duration_ms":335064,"concrete_test":"Verify the stated failure by taking a=0.9, c=1.9. Expand Phi(0.9;1.9;z)-1 = (0.9/1.9)z + O(z^2). Then (Phi-1)^{19/9} behaves like const * z^{19/9}(1+O(z)), which is not analytic in any neighborhood of 0 and has no finite derivative at 0; hence it violates the normalization f in A required by K_e. Separately, replace the power by (c/a)(Phi-1), re-run the proof of Theorem 2.3, and check that Lambda'(0)=1 and that the differential equation (2.5) is exactly the equation satisfied by p=1+zLambda''/Lambda'.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The statement of Theorem 2.3 is internally inconsistent with its proof and with Example 2.5. The theorem asserts that (Phi(a;c;z)-1)^{c/a} belongs to K_e, while the proof defines p from the function (c/a)(Phi-1) and Example 2.5 uses 2(Phi-1). The difference is not cosmetic. Since Phi-1 has a simple zero at z=0, the power (Phi-1)^{c/a} is not a single-valued analytic function near 0 unless c/a is a positive integer, and it is certainly not normalized with f'(0)=1. A concrete admissible parameter pair is a=0.9, c=1.9: it satisfies a>-1, c>=a, and (e-1)|c-2|+|a| = (e-1)(0.1)+0.9 approx 1.0718, which is below (e-1)^2(e+1)/e^2 approx 1.4857. Here c/a = 19/9 is not an integer, so (Phi-1)^{19/9} has a branch point at 0 and cannot belong to K_e. Thus the central confluent-hypergeometric claim, as written, is false; the correct claim is the constant-multiple version (c/a)(Phi-1) in K_e. The proof's differential equation (2.5) is derived for p=1+zPhi''/Phi', which is exactly the exponential-convexity expression of (c/a)(Phi-1), so the intended theorem is very likely salvageable, but the printed theorem needs correction before the result can stand.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript derives sufficient conditions on the parameters of the confluent hypergeometric function Phi(a;c;z), the Lommel function of the first kind h_{mu,nu}, and the generalized Struve function u_{nu} for these functions to lie in the exponential convex class K_e or the exponential starlike class S_e^*. The method is the differential-subordination admissibility lemma of Naz et al. [16]. Section 2 treats the confluent hypergeometric function, Section 3 the Lommel function, and Section 4 the generalized Struve function, with several examples and graphical illustrations.","tokens_in":19247,"tokens_out":34268,"duration_ms":291574,"significance":"If the results are correct, they provide new, explicit parameter ranges for exponential convexity and starlikeness of these special functions, and the derivations are analytic and parameter-free. The proofs in Sections 3 and 4 are internally consistent, and the constants in those admissibility estimates match the stated conditions. The paper does not rely on fitted parameters or on ad-hoc axioms tailored to these functions, since the key lemma is a published general criterion [16]. However, the central confluent-hypergeometric theorem is not in a usable form as printed: its statement is false as written, and its proof contains an algebraic inconsistency in the key estimate. Because this is the first and most prominent result, the manuscript requires substantial revision before it can be considered further.","major_comments":[{"comment":"The theorem states that (Phi(a;c;z)-1)^{c/a} belongs to K_e, but the proof and Example 2.5 concern the function (c/a)(Phi-1). These are not the same: since Phi-1 has a simple zero at z=0, the power (Phi-1)^{c/a} is not single-valued and analytic near 0 when c/a is not an integer, and it is not normalized with f'(0)=1 when c/a is an integer different from 1. For instance, with a=1,c=2 the stated function is (Phi-1)^2, whose derivative at 0 is 0, so it cannot belong to K_e. The proof's p=1+zPhi''/Phi' is exactly the exponential-convexity expression for (c/a)(Phi-1), so the intended statement is evidently the constant-multiple version. The theorem must be restated accordingly.","section":"Section 2, Theorem 2.3"},{"comment":"The proof concludes |Psi| > 1/e - 1/e^2 + 1 - (e-1)|c-2| - |a| - e >= 0 whenever (2.3) holds. But 1/e - 1/e^2 + 1 - e = -(e - 1 - 1/e + 1/e^2) = -(e-1)^2(e+1)/e^2, so the right-hand side is always negative; subtracting the additional nonnegative terms makes it strictly negative. Thus the stated lower bound cannot be nonnegative under condition (2.3), and the claimed inequality has the wrong sign. Consequently the admissibility condition Psi notin Omega is not established. The assertion that the auxiliary function g attains its global minimum at theta=pi is also not justified by the second derivative test alone, since g depends on m>=1. The proof of Theorem 2.3 is therefore invalid as written.","section":"Section 2, proof of Theorem 2.3 after (2.6)"},{"comment":"The paragraph defines g_delta(z):=Phi(1;1+delta;z) and h_delta(z):=zPhi(1;1+delta;z) and claims membership in K_e and S_e^*, respectively. But Phi(1;1+delta;0)=1, so Phi is not in the normalized class A, and zPhi is not in A either. What follows from the (corrected) Theorem 2.3 is, for a=1,c=1+delta, the membership of (1+delta)(Phi-1) in K_e, not of Phi itself. This passage should be rewritten to define the normalized functions and to state the correct conclusion.","section":"Section 2, integral-representation paragraph"}],"minor_comments":[{"comment":"The phrase 'Milen conjecture' should be 'Milin conjecture'.","section":"Introduction"},{"comment":"The proof uses 'using (3.2)' to assert that mu > 2, but this implication is not immediate from the displayed inequality and should be justified or stated as a separate consequence.","section":"Section 3, Theorem 3.1"},{"comment":"The symbol g(theta) is defined with different meanings in Theorem 2.3 and Theorem 3.1; this is confusing and the second function should be given a different name.","section":"Sections 2 and 3"},{"comment":"Example 2.5 uses Lambda(1;2;z)=2(Phi(1;2;z)-1), which is inconsistent with the statement of Theorem 2.3 as printed but consistent with the intended corrected version; the text should be aligned after the theorem is fixed.","section":"Example 2.5"},{"comment":"The figures are described only by placeholder text; ensure the final version contains the actual graphics and that they are legible.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The Section 2 results as printed are not reliable: the theorem statement and proof both need correction, and the integral-representation consequences are misstated. The Lommel and Struve sections appear internally sound, so the paper has salvageable content. The authors should be asked to correct Theorem 2.3, provide a valid estimate for the admissibility condition, and re-derive the integral-representation claims before resubmission."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main takeaway: the paper does what it says—sufficient conditions for three special function families to be exponential convex/starlike—and the intended mathematics is largely correct, but Theorem 2.3 as printed is false. The statement should be (c/a)(Φ−1) ∈ K_e, not (Φ−1)^{c/a} ∈ K_e; the power is not analytic at 0 for noninteger c/a and is not normalized. The proof and Example 2.5 actually use the constant-multiple version, so the fix is straightforward but mandatory.\n\nWhat is genuinely new: parameter ranges for the confluent hypergeometric, Lommel, and generalized Struve functions in the classes K_e and S_e^*, extending the authors' earlier Bessel-function paper. The method is the same differential subordination lemma from their own [16], but the admissibility estimates are new for these functions and are checked with care. The lower-bound lemmas imported from Yagmur and Orhan are external; I didn't verify them, but they are published results and look like the right tools. No fitted parameters, no circularity.\n\nSoft spots, in order: (1) The misstated Theorem 2.3. It matters, because the printed claim is not a harmless typo—for a=0.9, c=1.9 the hypotheses hold but the printed function is multivalued. It's exactly the kind of error a referee should catch. (2) The integral-representation paragraph calls Φ(1;1+δ;z) ∈ K_e, but that function isn't normalized; the result applies to the constant multiple. This is a smaller mislabeling in an example. (3) Display (3.2) has a broken parenthesis; the intended inequality is legible but should be cleaned up. (4) Example 2.5 uses a=1,c=2, where c/a is an integer, so it doesn't exercise the issue; an example with noninteger ratio would have exposed the bug.\n\nBottom line: this is a solid specialist paper once the theorem statement is corrected. It doesn't open a new direction, and the significance is catalogue-level, but that's a legitimate contribution in this area. I'd send it to a serious referee with a clear request to verify the corrected statement. The referee will likely spend most time on the external lemmas, not on the authors' own estimates.","headline":"The paper is a competent extension of the authors' own exponential-starlike/convex program to three more special function families, but Theorem 2.3 as printed states the wrong function and must be corrected.","tokens_in":19727,"tokens_out":3752,"would_cite":false,"duration_ms":34880,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["30C10","30C45","30C80"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper establishes explicit parameter inequalities under which normalized confluent hypergeometric, Lommel, and generalized Struve functions belong to the exponential convex class $K_e$ or the exponential starlike class $S_e^*$, by…","keywords":["differential subordination","exponential function","exponential convex","exponential starlike","confluent hypergeometric function","Lommel function","Struve function","unit disk"],"falsifier":"Take a parameter pair that satisfies the hypotheses of Theorem 2.3 with equality in (2.3), such as $a=-0.5$ and $c=2-\\frac{(e-1)^2(e+1)/e^2-0.5}{e-1}$, and compute $\\sup_{|z|<1}|\\log p(z)|$ for $p=1+z\\Phi''(a;c;z)/\\Phi'(a;c;z)$ on a fine grid in the unit disk; a single point with value above $1$ would disprove the membership claim for that parameter range. Alternatively, for the Lommel theorems, evaluate the cited lower bound $|h'_{\\mu,\\nu}(z)|\\ge 1-4M/(N(2M-3))$ at a parameter pair meeting the first two hypotheses of Theorem 3.1 but lying near the boundary; finding a zero of $h'_{\\mu,\\nu}$ would break the analyticity step before subordination is applied.","tokens_in":18725,"feed_emoji":"📐","tokens_out":13920,"duration_ms":130463,"temperature":0.7,"pith_summary":"This paper supplies explicit parameter conditions under which three classical special functions—the normalized confluent hypergeometric function, the Lommel function of the first kind, and the generalized Struve function of the first kind—are exponential convex or exponential starlike in the unit disk. A function is exponential convex when $1+zf''(z)/f'(z)$ lies inside the image of the unit disk under $e^z$, and exponential starlike when $zf'(z)/f(z)$ does; equivalently, the logarithm of the corresponding quotient has modulus less than $1$. The conditions are simple inequalities in the parameters, for instance $(e-1)|c-2|+|a|\\le (e-1)^2(e+1)/e^2$ for a normalized multiple of the confluent hypergeometric function, and each is proved by reducing a second-order differential equation to a subordination to $e^z$. These are sufficient conditions, so they enlarge the known explicit parameter ranges for geometric properties of these special-function families.","feed_headline":"New bounds put Kummer, Lommel, Struve in exp-convex class","feed_subtitle":"Explicit inequalities on parameters decide when these special functions map inside the disk of e^z.","key_machinery":"The load-bearing mechanism is the admissibility criterion for subordination to the exponential function. A normalized analytic $f$ is exponential convex exactly when $p=1+zf''(z)/f'(z)$ is subordinate to $e^z$, and exponential starlike exactly when $p=zf'(z)/f(z)$ is; the paper writes both conditions as $p\\in P_e$, where $P_e=\\{p:p(0)=1,\\ p(z)\\prec e^z\\}$. Lemma 1.1 states that if a differential expression $\\Psi(r,s,t;z)$ avoids zero for every boundary candidate $r=e^{e^{i\\theta}}$, $s=me^{i\\theta}r$ with $m\\ge 1$ and $\\operatorname{Re}(1+t/s)\\ge m(1+\\cos\\theta)$, then the analytic solution $p$ of $\\Psi(p(z),zp'(z),z^2p''(z);z)=0$ must lie in $P_e$. The proofs convert the special function's differential equation into exactly such a $\\Psi$ and then use elementary estimates—$|e^{e^{i\\theta}}-1|\\le e-1$, lower bounds on $\\operatorname{Re}((e^{e^{i\\theta}}-e^{-e^{i\\theta}})e^{-i\\theta})$ whose minimum is $2\\sin 1$, and similar trigonometric extrema—to show the parameter inequalities force $|\\Psi|>0$ on the boundary. Thus the inequalities are precisely the conditions under which the boundary of the exponential subordination class repels the differential expression.","core_discovery":"The central claim is that, for each of the three families, a moderate order/size condition on the parameters together with one explicit inequality places the normalized function in the exponential classes. For the confluent hypergeometric function $\\Phi(a;c;z)$, under $a>-1$, $c\\ge a$ (or $a\\le -1$, $c\\ge \\sqrt{1+(1+a)^2}$) and $(e-1)|c-2|+|a|\\le (e-1)^2(e+1)/e^2$, the function $\\frac{c}{a}(\\Phi(a;c;z)-1)$ belongs to $K_e$; shifting the parameters by one turns this into $z\\Phi(a;c;z)\\in S_e^*$. For the normalized Lommel function $h_{\\mu,\\nu}$, condition (3.2) gives $h_{\\mu,\\nu}\\in K_e$, condition (3.4) gives $f_{\\mu,\\nu}(z)=\\int_0^z h_{\\mu,\\nu}(t)/t\\,dt\\in K_e$ and hence $h_{\\mu,\\nu}\\in S_e^*$, and condition (3.7) gives $h_{\\mu,\\nu}(z)/z\\in P_e$. For the normalized generalized Struve function $u_\\nu$, condition (4.10) gives $u_\\nu\\in P_e$, while condition (4.13) gives $6\\kappa(1-u_\\nu)/c\\in K_e$; the paper's final theorem shows this last membership is preserved under convolution with arbitrary convex functions, hence under the two standard integral transforms it applies. In each theorem the auxiliary function $p=1+zf''/f'$ or $p=zf'/f$ is shown to satisfy a second-order differential equation, and the parameter inequality is exactly what prevents the associated differential expression from vanishing on the boundary candidates of $P_e$.","pith_inferences":["The same boundary-repulsion scheme should transfer to other normalized hypergeometric solutions of second-order differential equations; each new family only requires a second-order equation whose coefficients are polynomial in $z$ and a fresh set of trigonometric extrema.","Because the paper's inequalities are sufficient and not necessary, numerical exploration of the parameter space could identify larger regions; any widening would improve corollaries such as the explicit lower bound on $\\nu$ in Corollary 4.4.","The convolution closure used in Theorem 4.6 suggests a cheap way to generate many exponential convex functions: convolve the special functions with arbitrary convex functions, so the parameter conditions imply whole families of examples rather than single functions.","Replacing the target function $e^z$ by another standard simply connected image domain would produce analogous classes; the estimates would change only through the boundary shape of the target domain."],"forward_implications":["If Theorem 2.3 is correct, every parameter pair satisfying (2.3) yields an explicit exponential convex function, and the shifted inequality in Corollary 2.4 yields an exponential starlike function, both with no further numerical work.","If Theorem 3.1 and Theorem 3.2 are correct, the normalized Lommel functions and their integral transforms supply two linked families in $K_e$ and $S_e^*$ for the parameter ranges in (3.2) and (3.4).","If Theorem 4.3 and Theorem 4.6 are correct, the generalized Struve functions give exponential convex functions that remain exponential convex after convolution with any convex function, including the Alexander and Libera integral operators.","Corollaries 4.4 and 4.5 translate the general Struve condition into a single explicit lower bound on the order $\\nu$, so the normalized Struve and modified Struve functions are exponential convex and their derivatives exponential starlike above that bound.","Theorem 3.3 adds the derivative quotient $h_{\\mu,\\nu}(z)/z$ itself to the class $P_e$ under condition (3.7), which is a direct starlikeness-type statement for the Lommel family."],"supporting_citations":[{"why":"Provides Lemma 1.1, the admissibility criterion used in every theorem to convert non-vanishing of $\\Psi$ on boundary candidates into $p\\prec e^z$.","marker":"[16]"},{"why":"Supplies the lower-bound theorem ($\\operatorname{Re}((c/a)\\Phi')>0$) used in Theorem 2.3 to guarantee $\\Phi'\\neq 0$ and hence analyticity of $p$.","marker":"[12]"},{"why":"Supplies the estimates $|h'_{\\mu,\\nu}|>0$ and $\\operatorname{Re}(h_{\\mu,\\nu}/z)>0$ that make the auxiliary function $p$ analytic in Theorems 3.1 and 3.2.","marker":"[26]"},{"why":"Supplies the nonvanishing estimate $|u'_\\nu|>0$ used in Theorem 4.3 and the recursive identity used in Corollary 4.2.","marker":"[18]"},{"why":"Establishes that the classes $K_e$ and $S_e^*$ are closed under convolution with convex functions, used in Theorem 4.6.","marker":"[11]"}],"fun_headline_variants":["Parameter bounds nail exp-convexity for Kummer, Lommel, Struve","Explicit conditions place special functions in exp-convex class","When do Kummer, Lommel, Struve become exp-convex?","Three special functions made exp-convex by clean inequalities","New inequalities guarantee exp-convexity for special functions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument depends on cited lower-bound estimates from the literature that guarantee the needed derivatives do not vanish and certain real parts stay positive under the stated parameter restrictions, for example estimates of the forms $|h'_{\\mu,\\nu}(z)|>0$, $\\operatorname{Re}(h_{\\mu,\\nu}(z)/z)>0$, and $|u'_\\nu(z)|>0$; if any of these estimates is invalid or its hypotheses are misapplied, the auxiliary function $p$ need not be analytic and the subordination argument cannot begin.","fun_headline_variants_meta":{"raw":{"variants":["Parameter bounds nail exp-convexity for Kummer, Lommel, Struve","Explicit conditions place special functions in exp-convex class","When do Kummer, Lommel, Struve become exp-convex?","Three special functions made exp-convex by clean inequalities","New inequalities guarantee exp-convexity for special functions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000373,"raw_usage":{"total_tokens":2021,"prompt_tokens":1003,"completion_tokens":1018,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":619,"completion_tokens_details":{"reasoning_tokens":930}},"tokens_in":619,"tokens_out":1018,"duration_ms":8961,"temperature":1.0,"reasoning_tokens":930,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:22:44.089917+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a parameter pair that satisfies the hypotheses of Theorem 2.3 with equality in (2.3), such as $a=-0.5$ and $c=2-\\frac{(e-1)^2(e+1)/e^2-0.5}{e-1}$, and compute $\\sup_{|z|<1}|\\log p(z)|$ for $p=1+z\\Phi''(a;c;z)/\\Phi'(a;c;z)$ on a fine grid in the unit disk; a single point with value above $1$ would disprove the membership claim for that parameter range. Alternatively, for the Lommel theorems, evaluate the cited lower bound $|h'_{\\mu,\\nu}(z)|\\ge 1-4M/(N(2M-3))$ at a parameter pair meeting the first two hypotheses of Theorem 3.1 but lying near the boundary; finding a zero of $h'_{\\mu,\\nu}$ would break the analyticity step before subordination is applied.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides Lemma 1.1, the admissibility criterion used in every theorem to convert non-vanishing of $\\Psi$ on boundary candidates into $p\\prec e^z$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the lower-bound theorem ($\\operatorname{Re}((c/a)\\Phi')>0$) used in Theorem 2.3 to guarantee $\\Phi'\\neq 0$ and hence analyticity of $p$."},{"cited_title":"Ya˘ gmur, Hardy space of Lommel functions, Bull","cited_arxiv_id":null,"evidence_quote":"Supplies the estimates $|h'_{\\mu,\\nu}|>0$ and $\\operatorname{Re}(h_{\\mu,\\nu}/z)>0$ that make the auxiliary function $p$ analytic in Theorems 3.1 and 3.2."},{"cited_title":"Orhan and N","cited_arxiv_id":null,"evidence_quote":"Supplies the nonvanishing estimate $|u'_\\nu|>0$ used in Theorem 4.3 and the recursive identity used in Corollary 4.2."},{"cited_title":"Mendiratta, S","cited_arxiv_id":null,"evidence_quote":"Establishes that the classes $K_e$ and $S_e^*$ are closed under convolution with convex functions, used in Theorem 4.6."}],"review_version":1}