{"id":"2cd463a5-652d-4f80-9b9a-213fef7160dd","arxiv_id":"2412.17447","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A ZTF-based search finds 23 ellipsoidal ELM WD binaries (17 new), and radio non-detections of 11 systems put the fraction with millisecond pulsar companions below about 15%.","lead":"This paper searched about 12,000 candidate extremely low-mass white dwarfs for the brightness changes caused by a heavy orbiting companion, found 23 such systems, and checked six of them for radio pulses from a millisecond pulsar. No pulses were found, and the authors conclude that fewer than about 15 percent of these systems harbor a millisecond pulsar companion.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The <15% fraction bound is model-loaded: per-source completeness PL,j in Eq. (5) is taken from an assumed MSP luminosity function, so an unmeasured shift in that function or in the adopted distances changes the headline upper limit materially.","rationale":"The reader's conditional verdict already identifies the same load-bearing dependency: the binomial bound in Eq. (5) rests on an assumed log-normal MSP luminosity function, Pbeam=0.7, Peff=0.8, and adopted spectrophotometric distances. My stress test confirms that this is the correct soft spot. The concern is real and materially affects how the headline should be read, but it is explicitly acknowledged in Section 5 and the central claim is an upper limit, so it does not invalidate the paper. The strongest defense is that the bound is presented as an estimate and the authors list the conditions under which it would change. The paper would be materially strengthened by propagating luminosity-function and distance uncertainties into the reported 15% number and by quoting the result under a few alternative luminosity functions. The secondary issues noted by the reader, such as sdB/sdA contamination and the J0756+6704 modeling tension, are worth addressing but are not as close to the quantitative headline as the luminosity-function dependence of Eq. (5). Because the reader's CONDITIONAL verdict already captures exactly this caveat, no change in verdict is needed.","tokens_in":24369,"tokens_out":7322,"duration_ms":73551,"concrete_test":"Recompute Eq. (5) using (a) the empirical 1400 MHz luminosity function for binary MSPs from the ATNF catalogue (or the Huang & Wang 2020 distribution) instead of FG06, and (b) Gaia-parallax distances for J1048-0000, J1401-0817, and J0745+1949 in place of Table 2's spectrophotometric distances, keeping Pbeam=0.7 and Peff=0.8 fixed. Report the new fMSP,ELL and its 1sigma range. If the upper limit stays below 15%, the concern is resolved; if it moves to 20% or above, the headline should be restated as conditional on the luminosity function.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 5's headline bound, fMSP,ELL < 15^{+6}_{-3}%, is not a direct binomial count. Equation (5) converts zero detections into a limit through the product Pbeam x PL,j x Peff x f, where PL,j is the probability that an MSP's 1400 MHz luminosity exceeds L1400,j under a log-normal distribution with mu=-1.1, sigma=0.9 (Faucher-Giguere & Kaspi 2006). This luminosity function is not measured for the MSP/ELM-WD population; the MSPs in question are recycled systems whose luminosities could be systematically dimmer than field pulsars. The result is also sensitive to distance: L1400,j = Smin d^2, and for J1048-0000, J1401-0817, and J0745+1949 the spectrophotometric distances used in Table 2 differ from Gaia-parallax distances by factors of roughly 2-4 (Fig. 4), with the paper itself acknowledging (Section 5) that underestimated distances overestimate PL,j and therefore underestimate fMSP,ELL. Recomputing Eq. (5) with a luminosity function shifted 0.5-1 dex dimmer raises the 50% upper limit from ~15% toward 20-30% or higher; with alternative distance estimates the shift is at least a few percent. Because the abstract and conclusion present the 15% number without this conditioning, the central claim is weaker than it appears unless these population parameters are independently verified.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper searches for ellipsoidal variability among ~12,000 ELM white dwarfs and candidates using ZTF DR15 light curves, identifies 23 such systems (17 newly discovered), and selects nine high-priority targets for which it estimates companion masses from ellipsoidal amplitudes and radial-velocity mass functions. Targeted FAST observations of six of these systems yield no radio pulsations; combining these with three GBT non-detections and two previously known GBT-observed ellipsoidal systems, the paper derives an upper limit on the fraction of ellipsoidal ELM WDs orbited by MSPs of f_MSP,ELL < 15^{+6}_{-3}%.","tokens_in":24635,"tokens_out":5501,"duration_ms":50314,"significance":"If the headline bound holds, the result would strengthen the view that short-period ellipsoidal ELM WDs are predominantly double-degenerate systems rather than MSP/ELM binaries, and it would provide a useful photometric-selection-based complement to previous spectroscopic searches. The paper's strengths include a clearly described search and fitting procedure, a Monte Carlo treatment of the mass-inclination degeneracy, use of public survey data, and an explicitly conditional upper limit. The central claim is, however, sensitive to external population assumptions and to the sample's contamination by non-ELM objects, so the 15% number should be read as conditionally derived rather than as a model-independent measurement.","major_comments":[{"comment":"The headline bound f_MSP,ELL < 15^{+6}_{-3}% is computed through Eq. (5) using the per-source completeness PL,j derived from a log-normal MSP luminosity function with mu=-1.1 and sigma=0.9 (Faucher-Giguere & Kaspi 2006), together with Pbeam=0.7 and Peff=0.8. The paper acknowledges in Section 5 that a dimmer MSP luminosity function or underestimated spectrophotometric distances would raise the bound, but it does not quantify the sensitivity. For J1048-0000, J1401-0817, and J0745+1949, the spectrophotometric distances in Table 2 differ from Gaia-parallax distances by factors of 2-4 (Fig. 4), and the luminosity function of recycled MSPs in ELM WD binaries is not measured independently. Plausible shifts of the luminosity function by 0.5-1 dex would move the 50% upper limit from ~15% toward 20-30%. Because the abstract and conclusion report the 15% value without this conditioning, the central claim is weaker than presented; the authors should either add a sensitivity analysis over luminosity function parameters, Pbeam, Peff, and alternative distances, or explicitly present the bound as conditional on those assumptions.","section":"Section 5, Eq. (5), Table 3, Fig. 4"},{"comment":"The 23-source sample is assembled from 'ELM WDs or their candidates' and contains several objects that the paper itself identifies as likely non-ELM contaminants: J0238+4123 (Teff ~80,000 K, likely sdB), J2029+0701 (likely sdB), J1257+4220 (hot DA), and J1048-0000/J1401-0817 (possibly sdA-type). The fraction f_MSP,ELL is then estimated for 11 systems that include J0745+1949, which the paper retains only as a proto-WD candidate not on the cooling track. If the denominator includes non-ELM objects, the derived upper limit on the fraction of genuine ELM WDs around MSPs is diluted, and the direct comparison with previous fNS estimates becomes ambiguous. The authors should either restrict the fraction to spectroscopically confirmed ELM WDs, or estimate and propagate the contamination rate into the bound.","section":"Section 4.2, Table 1"},{"comment":"The quoted uncertainty on the upper limit, 15^{+6}_{-3}%, is not derived in the text. Equation (5) contains Pbeam, Peff, and PL,j, each of which carries uncertainty, but the paper does not state how the asymmetric error bar is computed, nor whether it includes the systematic uncertainties in the luminosity function and beaming fraction. Without this information, the precision implied by the quoted error bar is unjustified. Please specify the error propagation procedure, or quote the bound without asymmetric uncertainties and list the dominant systematic rather than a statistical error.","section":"Section 5, Eq. (5)"}],"minor_comments":[{"comment":"There is a typo: 'Chandard CSC 2.1' should be 'Chandra CSC 2.1'.","section":"Section 4.5"},{"comment":"The telescope name is typeset with an internal space ('F AST') in multiple places; please fix the formatting to 'FAST'.","section":"Sections 4.5 and 5"},{"comment":"The sentence beginning 'Its Teff and log g1 differ from those derived by Brown et al. (2020), whose estimated parallax (~1.443)' has an unclear antecedent; 'whose' appears to refer to Brown et al. rather than to the star, so please rephrase.","section":"Section 4.2"},{"comment":"The text states that the 23 sources include '7 sources from sample S1, 17 sources from sample S2, and 3 sources from sample S3', which sums to 27 before overlap removal; please clarify that these are the counts before de-duplicating overlapping sources.","section":"Section 3.2, Table 1"},{"comment":"The statement that four targets have companion masses exceeding 1 Msun is based on posterior medians, but the 1-sigma uncertainties in Table 3 are very wide (e.g., J1401-0817 has M2 = 1.167^{+0.902}_{-0.297}); the wording should say 'median companion masses above 1 Msun' to avoid overstating the confidence in this result.","section":"Abstract and Section 4.4, Table 3"},{"comment":"The two additional GBT-observed sources J0056-0611 and J0112+1835 are included in the fraction estimate even though, as footnote 12 notes, their ellipsoidal variability was not found in the ZTF data; please clarify how these sources satisfy the same ellipsoidal selection criteria as the other nine systems.","section":"Section 5"}],"recommendation":"major_revision","confidential_remarks":"The manuscript addresses a timely question and the search work is competently presented. The main concern is that the headline fraction is presented without fully acknowledging the model dependence of the luminosity completeness correction; a sensitivity analysis would make the bound much more robust. The sample contamination issue is also substantive but fixable. I see no citation or novelty problems; the paper is within the scope of ApJ and should be reconsidered after revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The useful part of this paper is the 17 newly identified ellipsoidal variables among ELM WD candidates and the radio non-detections for six of them. The selection is straightforward: ZTF period search, visual lightcurve classification, five-parameter ellipsoidal fitting, and PRESTO on FAST/GBT data. The methods are standard and described well enough to follow, and the non-detections look credible. If you work on ELM WD binaries or MSP recycling, the candidate list and luminosity upper limits are worth having.\n\nThe headline claim, fMSP,ELL < 15^{+6}_{-3}%, is more fragile than the abstract suggests. Equation (5) converts zero detections into a limit using an assumed MSP luminosity function (log-normal with mu=-1.1, sigma=0.9), a beaming fraction, and a search efficiency, none of which are measured for this specific population. For J1048-0000, J1401-0817, and J0745+1949 the spectrophotometric distances differ from Gaia parallaxes by factors of 2-4, and the authors concede in Section 5 that underestimating distances would push the limit upward. The stress-test note is right about this, but it is not a hidden defect: the conditioning is stated in Section 5. The problem is that the abstract and conclusion present the 15% without that caveat, which overstates the strength of the result for casual readers.\n\nOther soft spots are minor. The 23-source sample is assembled by visual classification that can admit sdA/sdB contaminants; the authors acknowledge this themselves. The companion-mass posteriors are wide, so \"four targets exceed 1 M⊙\" really means \"four have medians above 1 M⊙,\" which is worth softening. The J0756+6704 fit only survives 0.45% of Monte Carlo trials, so that target's mass estimate should carry a warning label.\n\nAll told, this is a solid, honest observational paper. The search is reproducible, the non-detections are real, and the catalog contribution is genuinely new. The fraction bound is conditional, but the paper says so internally. I would send this to a referee. The referee should ask for a machine-readable candidate table and a version of the abstract that carries the conditioning, but neither is a blocker.","headline":"Competent null-result search with a model-loaded upper limit; the 17 new systems and radio upper limits are the durable part, not the headline fraction.","tokens_in":25244,"tokens_out":1983,"would_cite":false,"duration_ms":19888,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper estimates that fewer than 15% of short-period ellipsoidal extremely low-mass white dwarfs are orbited by millisecond pulsars, implying that most are double white dwarfs.","keywords":["extremely low-mass white dwarfs","ellipsoidal variables","millisecond pulsars","radio pulsar search","white dwarf binaries","close binary stars","orbital period search"],"falsifier":"A single convincing radio-pulse detection from any of the 11 ellipsoidal ELM WDs used in the fraction estimate, for example a longer FAST or GBT integration reaching a 1.4-GHz flux limit near 1 $\\mu$Jy on one of the six newly observed targets, would remove the non-detection basis for the <15% bound. A measurement of the actual MSP luminosity function from a complete sample, or accurate parallaxes for J1048-0000, J1401-0817, and J0745+1949, would provide a direct check on the completeness values that set the bound.","tokens_in":24079,"feed_emoji":"🔭","tokens_out":11892,"duration_ms":101235,"temperature":0.7,"pith_summary":"The paper asks how many extremely low-mass white dwarfs (ELM WDs, helium-core white dwarfs with masses below 0.3 solar masses) that show ellipsoidal brightness variations on sub-day orbits are actually orbited by recycled millisecond pulsars rather than by other white dwarfs. Screening roughly 12,000 ELM WDs or candidates in public time-domain data, the authors identify 23 systems with ellipsoidal-like variability and orbital periods under one day, 17 of them new, and estimate companion masses for nine high-priority candidates, four of which exceed one solar mass. Targeted radio observations of six of these systems with the Five-hundred-meter Aperture Spherical radio Telescope found no pulsed signals, with flux upper limits near 8 microjanskys at 1.4 GHz. Combining these non-detections with prior Green Bank Telescope non-detections of similar systems, the paper estimates that the fraction of ellipsoidal ELM WDs that host millisecond pulsars is below $15^{+6}_{-3}\\%$. If the bound stands, the short-period ellipsoidal ELM WD population is mostly double white dwarfs, which matters for binary evolution models and for what future gravitational-wave observatories will see.","feed_headline":"Most ellipsoidal ELM white dwarfs are not pulsar companions","feed_subtitle":"A radio search of 11 such systems pushes the millisecond-pulsar fraction below 15%, pointing to double white dwarfs.","key_machinery":"The argument rests on two mechanical steps. First, the ellipsoidal-variability amplitude $A_{\\rm EV} = 3\\pi^2(15+\\mu_1)(1+\\tau_1)M_2 R_1^3 \\sin^2 i / [5 P_{\\rm orb}^2 (3-\\mu_1) G M_1 (M_1+M_2)]$, with limb-darkening $\\mu_1$ and gravity-darkening $\\tau_1$, is combined with the mass function $f_1(M_2)=P_{\\rm orb}K_1^3/(2\\pi G)=M_2^3\\sin^3 i/(M_1+M_2)^2$ in Monte Carlo draws to map each target's companion mass and inclination. Second, the headline fraction comes from the binomial non-detection formula $\\prod_{j=1}^{N}(1-P_{\\rm beam}P_{L,j}P_{\\rm eff}f_{\\rm MSP,ELL})>1/2$, where $P_{\\rm beam}=0.7$ is the pulsar beaming fraction, $P_{\\rm eff}=0.8$ the search-success rate, and $P_{L,j}$ is the luminosity completeness of each source computed from an assumed log-normal millisecond-pulsar luminosity distribution; requiring this product to exceed one-half turns 11 radio non-detections into the upper limit on $f_{\\rm MSP,ELL}$. The candidate selection itself uses Lomb-Scargle periodograms and a five-parameter harmonic fit to the phase-folded ZTF light curves to separate the $\\cos 2\\phi$ ellipsoidal term from Doppler beaming, reflection, and the first orbital harmonic.","core_discovery":"The central claim is a demographic bound: among ELM WDs selected to show ellipsoidal variability with orbital periods shorter than one day, the fraction with millisecond-pulsar companions is below $15^{+6}_{-3}\\%$, estimated from the absence of radio pulsations in 11 such systems. The paper builds this from a sample of roughly 12,000 ELM WDs or candidates, finds 23 ellipsoidal variables (17 newly discovered), and uses the ellipsoidal amplitude--orbital period relation, solved together with each system's mass function, to estimate unseen companion masses for nine high-priority targets; four have companion masses above 1 $M_\\odot$. A FAST radio search of six of these targets produced no convincing pulsed signals, no X-ray counterparts appear in archival catalogues, and, combined with earlier GBT non-detections, the 11-system binomial calculation puts the MSP fraction below the quoted bound. The paper's conclusion is that most of these short-period ellipsoidal systems are double white dwarfs rather than MSP/ELM binaries.","pith_inferences":["If the bound holds under deeper searches, the ellipsoidal-selection method mostly isolates double-degenerate systems, implying that future radio follow-up of such targets should expect a low yield and that the systems are better exploited as WD+WD gravitational-wave sources.","Applying the same binomial non-detection method to a larger sample, such as the southern-sky ELM candidates once new time-domain data are available, would push the constraint toward the roughly 10% level previously estimated for low-mass WD binaries generally.","Radial-velocity monitoring of the four candidates with companion masses above 1 $M_\\odot$ could separate a massive CO-core white dwarf from a neutron star companion without needing a radio detection, testing the companion-mass estimates from ellipsoidal amplitudes."],"forward_implications":["If the <15% bound holds, most short-period ellipsoidal ELM WDs are double white dwarfs; the recycled-pulsar channel is a minority outcome among these photometrically selected systems.","The four high-priority candidates with companion masses above 1 $M_\\odot$ are the most promising targets for deeper radio and X-ray follow-up to look for MSP companions.","The 17 newly discovered ellipsoidal variables enlarge the census of ultracompact WD binaries with sub-day periods, several of which are candidate low-frequency gravitational-wave sources for future space-based detectors.","Extending this search to the roughly 4,300 ELM candidates south of the current survey footprint, with southern time-domain surveys, should tighten or revise the fraction estimate."],"supporting_citations":[{"why":"It supplies the binomial non-detection method, expressed in Eq. (5), that turns the absence of radio pulsations into an upper limit on the MSP fraction.","marker":"van Leeuwen et al. 2007"},{"why":"It applies the same binomial method to low-mass WDs and provides the framework for interpreting radio non-detections that this paper extends to ellipsoidal ELM WDs.","marker":"Agüeros et al. 2009"},{"why":"It provides the log-normal millisecond-pulsar luminosity distribution ($\\mu=-1.1$, $\\sigma=0.9$) used to compute each source's luminosity completeness $P_{L,j}$.","marker":"Faucher-Giguère & Kaspi 2006"},{"why":"It gives the radio beaming fraction $P_{\\rm beam}=0.7\\pm0.2$ assumed for millisecond pulsars in the fraction calculation.","marker":"Kramer et al. 1998"},{"why":"It supplies three GBT radio non-detections among the 11 systems and a prior estimate below 10% for the neutron-star companion fraction in low-mass WD binaries.","marker":"Athanasiadis et al. 2021"},{"why":"It provides the two additional ellipsoidal ELM WDs with GBT non-detections that bring the total to 11 systems in the fraction estimate.","marker":"Bell et al. 2018"},{"why":"It supplies the ellipsoidal-variability amplitude formula (Eq. 1) and the five-parameter light-curve model used to estimate companion masses.","marker":"Hermes et al. 2014"},{"why":"It provides the sample S1 ELM WDs and atmospheric parameters, and it includes GBT non-detections for some of the targets used in the fraction estimate.","marker":"Brown et al. 2020"}],"fun_headline_variants":["Most ellipsoidal ELM WDs aren't pulsar companions","Pulsar fraction below 15% in ellipsoidal ELM white dwarfs","No radio pulsations found in 11 ellipsoidal ELM WD binaries","FAST and GBT silence pushes MSP fraction below 15% for ELM WDs","Ellipsoidal ELM WDs: double white dwarfs outnumber pulsar pairs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The limiting assumption is that the millisecond-pulsar population in these binaries has the same luminosity distribution, beaming fraction, and search success as the standard values adopted in the calculation (a log-normal luminosity distribution with mean $-1.1$ and width $0.9$, beaming 0.7, efficiency 0.8); if the true pulsars are dimmer, beamed away from Earth more often, or the distances to the targets are overestimated, the same 11 non-detections would permit a true MSP fraction well above 15%.","fun_headline_variants_meta":{"raw":{"variants":["Most ellipsoidal ELM WDs aren't pulsar companions","Pulsar fraction below 15% in ellipsoidal ELM white dwarfs","No radio pulsations found in 11 ellipsoidal ELM WD binaries","FAST and GBT silence pushes MSP fraction below 15% for ELM WDs","Ellipsoidal ELM WDs: double white dwarfs outnumber pulsar pairs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000966,"raw_usage":{"total_tokens":4164,"prompt_tokens":1054,"completion_tokens":3110,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":670,"completion_tokens_details":{"reasoning_tokens":3004}},"tokens_in":670,"tokens_out":3110,"duration_ms":21233,"temperature":1.0,"reasoning_tokens":3004,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T05:27:27.292616+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A single convincing radio-pulse detection from any of the 11 ellipsoidal ELM WDs used in the fraction estimate, for example a longer FAST or GBT integration reaching a 1.4-GHz flux limit near 1 $\\mu$Jy on one of the six newly observed targets, would remove the non-detection basis for the <15% bound. A measurement of the actual MSP luminosity function from a complete sample, or accurate parallaxes for J1048-0000, J1401-0817, and J0745+1949, would provide a direct check on the completeness values that set the bound.","supporting_citations":[{"cited_title":"M., Berezina , M., Antoniadis , J., et al","cited_arxiv_id":null,"evidence_quote":"It supplies three GBT radio non-detections among the 11 systems and a prior estimate below 10% for the neutron-star companion fraction in low-mass WD binaries."}],"review_version":1}