How many hidden pulsars can archival Parkes data reveal in the Magellanic Clouds?

Reprocessing Parkes Magellanic Cloud archives with deep harmonic folding uncovered six pulsars, five in the LMC, one a Milky Way foreground source.

Direct answer

Reprocessing the original Parkes Multibeam Magellanic Cloud surveys with candidate folding pushed near the noise floor and high harmonic sums yielded six new pulsars. Five belong to the Large Magellanic Cloud, raising its known radio pulsar count to 49, and one lies toward the Small Magellanic Cloud but is a Milky Way foreground object.

1sources cited

This article was generated with WisPaper-powered search and paper analysis.

Six new pulsars from data already in hand

The paper asks whether the original Parkes Multibeam surveys of the Magellanic Clouds still contain detectable pulsars that earlier processing missed. The authors reprocessed both the PM survey and a high-resolution survey of the Clouds and report six new pulsars. Five are associated with the Large Magellanic Cloud (LMC), and the sixth, PSR J0051−75, sits toward the Small Magellanic Cloud (SMC) but is a Galactic foreground pulsar rather than an SMC member.

The LMC count matters because the paper states that these discoveries bring the total number of radio pulsars known in the LMC to 49. The new objects are faint: the five LMC pulsars have 1.4-GHz flux densities of roughly 26–60 µJy, and the paper describes these as lower-limit estimates derived from the radiometer equation because no timing positions or interferometric localizations were yet available.

One of the five, PSR J0520−67, has a spin period of 3.8 s, which the authors describe as the slowest pulsar known in the Magellanic Clouds. That single object already shows why re-searching old data can change a population census rather than merely add marginal entries.

Folding faint candidates at high harmonic sums

The method is the reason these pulsars were recoverable. Pulsar searches look for a periodic train of pulses in noisy radio data; a standard approach transforms the time series into the frequency domain and sums harmonics of a trial spin frequency to accumulate signal. The authors folded candidates near the noise floor and summed harmonics up to high orders, a strategy they describe as previously effective at finding many pulsars in archival data.

PSR J0443−68 shows how thin the margin was. It was found very near the adopted FFT noise floor with a spectral signal-to-noise ratio of only 5.8, and only when harmonic summing reached the 32nd harmonic. The paper notes it might not have been detectable above the noise at all with traditional summing limited to 16 harmonics. Historically, most pipelines did not fold candidates of such low significance, with typical thresholds of 5–8.

The authors also ran a time-domain search with a Fast Folding Algorithm, which folds the full time series at many trial periods and is less affected by red noise and better suited to narrow, non-sinusoidal pulses. All six pulsars were detected through the FFT-based searches, and all six were also recovered in the FFA results, though the FFA found no additional compelling candidates. No acceleration trials were applied in the time-domain search, so that stage was sensitive primarily to isolated pulsars.

Distance excess separates Cloud pulsars from foreground

Deciding whether a detection belongs to the LMC, the SMC, or the Milky Way depends on the dispersion measure (DM), the frequency-dependent delay that free electrons impose on radio pulses. A pulsar behind more ionized gas shows a larger DM, so the authors compare each DM with the maximum Galactic line-of-sight DM predicted by three electron-density models, NE2001, YMW16, and NE2025.

For the five LMC pulsars, the ratio of measured DM to that maximum foreground value ranges from 1.1 to 2.7. The case is strongest for PSR J0520−67 and PSR J0541−68, whose DMs are nearly a factor of three above the maximum foreground prediction. The other three lie closer to the threshold but remain systematically above the maximum predicted by all three models. For PSR J0051−75, the ratio is about 0.5, below the foreground ceiling, which is why the paper assigns it to the Milky Way rather than the SMC.

Follow-up observations support the reality of the detections, with caveats. PSR J0443−68 was not redetected in an initial 1.5 hr Parkes observation but was confirmed in a later 3 hr observation with the Ultra-wideband receiver at a signal-to-noise ratio of 70. PSR J0543−67 was not detected in two 2.5 hr follow-up observations separated by one month; the authors argue it is likely genuine because its emission properties resemble PSR J0443−68 and a known pulsar in the same dataset, PSR J0451−67, shows similar band-limited behavior. None of the six shows evidence of acceleration, suggesting they are likely isolated systems.

Scintillation and nulling explain some non-detections

Four of the new pulsars show scintillation, the frequency-dependent flickering caused by scattering in intervening plasma. PSR J0443−68 switches on and off across the 1150–1700 MHz band, and its flux density rose from about 26 µJy in the discovery observation to about 75 µJy in confirmation, indicating a highly variable source. During confirmation it entered a null state lasting about 36 min, and its absence from the entire initial 1.5 hr follow-up suggests the null state can persist for at least that long.

The paper is careful about what it can measure. A Gaussian fit to the frequency autocorrelation gives a scintillation bandwidth of about 1.6 MHz, within the range previously reported for Magellanic Cloud pulsars, but the time-domain fit yields a scintillation timescale shorter than the subintegration duration, so that timescale cannot be reliably measured. After removing the zero-lag component, the remaining autocorrelation structure is weak, so the authors say the bandwidth should also be treated with caution.

For PSR J0543−67, the disappearance of signal after about 1.5 hr persists in all three sub-bands tested, which is suggestive of broadband nulling. With the available signal-to-noise ratio, however, the authors cannot conclusively distinguish intrinsic nulling from scintillation. This is a case where a non-detection in follow-up is consistent with the source being real but intermittent, not evidence that the discovery was spurious.

Magellanic pulsars resemble Galactic canonical pulsars

The paper compares the 1.4-GHz luminosity distribution of all current Magellanic Cloud pulsars with that of Galactic non-recycled pulsars. Using an updated LMC sample that includes the new discoveries and recent MeerKAT/TRAPUM detections, the fitted power-law slopes agree at about the 15% level. The authors attribute the apparently steeper slope reported in earlier work to small-number statistics and heterogeneous survey depths rather than to a genuinely different population.

A duty-cycle comparison points the same way. Magellanic Cloud pulsars appear on average slightly narrower at a given period, but the paper argues this is plausibly explained by selection effects, since distance-limited surveys preferentially detect the brightest and narrowest sources, rather than by a fundamentally distinct emission geometry.

The flux-density trend across surveys supports the idea that deeper processing keeps finding fainter objects. LMC pulsars from the original Parkes survey have a mean 1.4-GHz flux density of about 149 µJy; four added in a later reprocessing average about 82 µJy; the five reported here average only about 38 µJy; and the known MeerKAT LMC sample averages about 11 µJy. The authors read this progression as evidence that their reanalysis was effective at uncovering pulsars previously too weak to detect in the same data.

Deeper surveys could add hundreds of normal pulsars

Population synthesis simulations address what is still missing. Under a Galactic log-normal luminosity prescription, a MeerKAT survey with TRAPUM-like sensitivity is predicted to detect 201±13 normal pulsars in the LMC and 35±6 in the SMC. These are model expectations for adopted survey regions and sensitivity, not objects already found.

Millisecond pulsar predictions depend heavily on the assumed luminosity function. If Magellanic Cloud millisecond pulsars follow the same log-normal distribution as normal Galactic pulsars, the simulations yield 5±2 in the LMC and 3±2 in the SMC. Using a Galactic millisecond-pulsar power-law luminosity function instead reduces the expected yield to essentially zero in the SMC and about 0.5 in the LMC. The paper explicitly labels the millisecond-pulsar yields as upper limits.

The simulations also imply that the PM survey has likely recovered most of the normal pulsars accessible to its sensitivity, while the underlying population is far from exhausted. The authors adopt a detection threshold of signal-to-noise ratio 10 for these forecasts, noting that no Magellanic Cloud pulsar has ever been discovered below a folded signal-to-noise ratio of 10 and that several candidates near 9 could not be confirmed.

The archival reanalysis shifts the priority to sensitivity

The paper's most consequential comparison is not the count of six but the gap between what the PM survey could reach and what remains. The five new LMC pulsars are fainter than those found in earlier passes over the same data, and one of them, PSR J0443−68, sat at a spectral signal-to-noise ratio of 5.8 that would have been discarded under conventional thresholds. That makes the choice of folding depth and harmonic order a direct determinant of the census, not a technical detail.

At the same time, the paper's own limits define how far that conclusion travels. The flux densities are lower-limit estimates without timing positions, one new source lacks follow-up confirmation, and the millisecond-pulsar forecasts swing from a handful to nearly zero depending on the adopted luminosity function. The simulations are conditioned on assumed survey coverage, with 75% of the adopted regions treated as covered by coherent-beam tiling.

The practical priority that follows is specific: the PM archive has been squeezed close to its sensitivity limit for normal pulsars, so further gains in the Clouds depend on deeper surveys rather than more reprocessing of the same data. The predicted LMC yield of 201±13 normal pulsars, against 49 known radio pulsars there, is the number that should frame that decision, with millisecond-pulsar expectations treated as an upper bound rather than a target.

Sources used in this answer

1

Discovery of Hidden Extragalactic Pulsars in the Parkes Archival Data

Reprocessing the original Parkes Multibeam Magellanic Cloud surveys with candidates folded near the noise floor and high harmonic sums uncovered six pulsars, five in the LMC (bringing its known radio pulsar total to 49) and one Galactic foreground source toward the SMC.