Draft version July 29, 2025

Typeset using LATEX twocolumn style in AASTeX631

The Timing Evolution of the Magnetar Swift J1818.0–1607 During a Period of Reduced Activity

Moaz Abdelmaguid ,1, 2, 3 Paulo C. C. Freire ,4 Joseph D. Gelfand ,2, 3, 5 Yogesh Maan ,6 Samayra Straal ,2 and
J. A. J. Alford 2, 3
1 Department of Physics, New York University, 726 Broadway, New York, NY 10003, USA
2 New York University Abu Dhabi, PO Box 129188, Abu Dhabi, UAE
3 Center for Astrophysics & Space Science (CASS), NYU Abu Dhabi, PO Box 129188, Abu Dhabi, UAE
arXiv:2507.19698v1 [astro-ph.HE] 25 Jul 2025

4 Max-Planck-Institut für Radioastronomie, Auf dem Hügel 69, 53121 Bonn, Germany
5 Center for Cosmology and Particle Physics (CCPP, Affiliate), New York University, 726 Broadway, New York, NY 10003, USA
6 National Centre for Radio Astrophysics, Tata Institute of Fundamental Research, Pune 411007, Maharashtra, India

(Accepted July 25, 2025)

Submitted to ApJ

ABSTRACT
We report results from an observational campaign of the radio-loud magnetar Swift J1818.0–1607 using
the Green Bank Telescope (GBT) at 2.0 GHz which began in 2021 November, during a period of reduced
activity approximately 20 months after its 2020 March outburst. Over the ∼ 60 day duration reported here, the
integrated pulse profile remained consistently stable, exhibiting a single, narrow peak with a small precursor
component and no evidence of a postcursor one. This pulse profile is in sharp contrast to the double-peaked
morphology observed during an observing campaign ∼ 120 days preceding ours. Along with this change in
the integrated pulse profile shape, we also measure a slower spin-down rate (ν̇) compared to the end of that
preceding campaign. Together, these differences suggest that a mode-switching event likely occurred between
the end of that campaign and the start of ours. Finally, we derived a phase-connected timing solution from
our data, from which we inferred a characteristic age of τc ∼ 2500 years—about 2.5 times older than the most
recent published estimate— and a surface dipole magnetic field strength of B f ield ∼ 1 × 1014 G, nearly three
times weaker.These updated estimates reflect the short-term variations in the magnetar’s spin-down rate, from
which both its age and magnetic field strength are inferred, rather than intrinsic changes in the magnetar itself.

Keywords: Magnetars (992); Radio Pulsars (1353); Pulsars (1306); Neutron stars (1108); Non–thermal radiation
sources (1119)

1. INTRODUCTION peaters (Kouveliotou et al. 1987) and anomalous X-ray pul-

Magnetars are neutron stars believed to have extremely sars, whose X-ray luminosities (LX ) were found to far ex-
strong surface magnetic fields, ranging from 1013 to 1015 G, ceed their spin-down power (Ė), implying an alternative en-
and typically have long spin periods (1 to 12 seconds) and ergy source beyond rotational losses (Duncan & Thompson
small characteristic ages (e.g, Kaspi & Beloborodov 2017, 1992).
and the references therein). Unlike rotation-powered pulsars Magnetars exhibit a wide range of observational transient
(RPPs), whose emission is driven by the loss of rotational phenomena, including short (0.1 s), bright X-ray bursts, and
energy, magnetars are believed to be powered by the decay giant flares that last for seconds to minutes in duration longer.
of their ultra-strong magnetic fields. This idea was first pro- These outbursts are usually accompanied by a change in
posed in the 1990s to explain the mysterious soft gamma re- their spectral and timing behavior (Kaspi & Beloborodov
2017). They also show significant variability in their spin-
down rates, along with discrete glitching events and changes
Corresponding author: Moaz Abdelmaguid in pulse profile shapes that point to a highly dynamic and
[email protected] evolving magnetosphere (Lower et al. 2021; Rajwade et al.
2022).
2 Abdelmaguid et al.

Table 1. Summary of past & present monitoring campaigns of Swift J1818.0–1607 including this work

MJD Telescope Frequency (GHz) Reference

58922 – 59015 ∗ 1.37 – 6 (Champion et al. 2020b)

58936 – 59092 TMRT 2.24 – 8.69 (Huang et al. 2021)
58977 – 59128 Parkes 2.4 (Lower et al. 2020)
59053 – 59105 Lovell & MK II 1.53 (Rajwade et al. 2022)
59092 - 59190 GBT 0.8 – 35 (Lewis et al. 2025)
59104 – 59365 Lovell & MK II 1.53 (Fisher et al. 2024)
59117 – 59400 Effelsberg 6.0 (Rajwade et al. 2022)
59520 – 59578 GBT 2.0 This Work

Note— TMRT: Shanghai Tian Ma Radio Telescope, GBT: Green Bank Telescope
∗ Observations reported by Champion et al. (2020b) were made using many telescopes; Efflsberg at 1.37, 2.55, 4.85 & 6 GHz, Lovell at 1.53 GHz and Nançay
Radio Telescope (NRT) at 1.48 GHz. MK II is the Mark II radio telescope located at Jodrell Bank Observatory.

Of the ∼30 currently identified magnetars, pulsed radio at 1.4 GHz by Rajwade et al. (2022), along with simultane-
emission has only been detected in only six of them (Olausen ous observations at 2.2 and 8.5 GHz by Huang et al. (2021)
& Kaspi 2014). The appearance of this radio emission ap- and Bansal et al. (2023), revealed additional mode-switching
pears to be correlated with outbursts observed at higher en- episodes with pulse profiles that differed markedly from the
ergies (Camilo et al. 2006, 2007a) and typically features sig- initial two. These observations further emphasize the magne-
nificant flux and spectral variations (Lower et al. 2020). Ad- tar’s complex emission behavior across multiple frequencies.
ditionally, the pulsed radio emission of magnetars display in- Table 1 summarizes previous radio monitoring campaigns of
tegrated profiles that vary within hours to days, along with a this source, including the observations presented in this work.
relatively flat radio spectra, characterized by spectral indices Given the magnetar’s dynamic nature, continued long–
α > − 0.8 (Torne et al. 2015, 2017), where the flux density term monitoring of Swift J1818.0–1607 is essential for a
S ν ∝ να . This is in contrast to RPPs, which usually have sta- comprehensive understanding of the later stages of magne-
ble integrated profiles and steeper radio spectra, with typical tar evolution. In particular, observations during periods of
values of α ∼ − 1.6 (Jankowski et al. 2018). reduced activity provide an opportunity to explore whether
Swift J1818.0−1607 was discovered in 2020, following magnetar magnetospheres settle into a stable configuration
its bright X-ray outburst detected by the Swift–Burst Alert post-outburst, or continue to evolve over longer timescales.
Telescope (BAT) (Gehrels et al. 2004). Subsequent ob- Here, we present radio observations of Swift J1818.0–1607
servations with the Neutron Star Interior Composition Ex- with the 100-m Green Bank Telescope (GBT), during a
plorer (NICER) revealed X-ray pulsations with a period of period of reduced activity. We concentrate on a subset of
1.36 s (Enoto et al. 2020; Esposito et al. 2020), establish- our broader monitoring campaign, conducted between 2021
ing it as the magnetar with the shortest known period to November and 2021 December, during which we could de-
date. Shortly afterwards, radio pulsations were detected by rive a phase-connected timing solution. In §2, we present the
the Efflesberg telescope (Karuppusamy et al. 2020), mak- timeline of our observations and describe the data reduction
ing it the fifth known “radio-loud” magnetar. The initial ob- procedures. The results are presented in §3, including the
served spin-down rate suggests a characteristic age of τc ∼ timing solution (§3.1), pulse profile evolution (§3.2), flux
250 – 1000 years (Champion et al. 2020b), implying that density and spectral indices measurements (§3.3). In §4, we
Swift J1818.0−1607 is one of the youngest known magne- compare these findings with earlier studies on this source and
tars. other magnetars, and discuss their implications. We finally
Similar to other radio-loud magnetars, Swift J1818.0–1607 conclude with a summary and future outlook in §5.
shows significant variability in its pulse profile morphology,
spin-down rate, and flux across a range of timescales (Lewis 2. OBSERVATIONS AND DATA REDUCTION
et al. 2025; Fisher et al. 2024; Rajwade et al. 2022; Huang We observed Swift J1818.0–1607 between 2021 March 18
et al. 2021; Lower et al. 2020; Champion et al. 2020b; Hu (MJD 59291) and 2023 November 12 (MJD 60260) with the
et al. 2020). Furthermore, it has displayed numerous distinct GBT in three frequency ranges; L band (1.1 − 1.9 GHz), S
emission modes, reflecting a highly dynamic and evolving band (1.6 − 2.4 GHz) & C band (4.65 − 6.15 GHz). Figure
magnetosphere. During its 2020 outburst, (Lower et al. 2021) 1 shows a timeline of our entire multi-frequency monitoring
identified two such emission modes below 4 GHz that al- campaign. The measurements from the full set of observa-
ternated on minute-long timescales. Subsequent monitoring tions will be presented in future work; this paper focuses
 3

Table 2. Log of the S band dense-set observations highlighted in

red in Figure 1

MJD Date Duration Fc

- (YYYY MM DD) (Minutes) (GHz)

59520.76 2021 Nov 02 15 2.0

59536.71 2021 Nov 18 15 2.0
59550.88 2021 Dec 02 15 2.0
59559.84 2021 Dec 11 15 2.0
59564.80 2021 Dec 16 15 2.0
59568.72 2021 Dec 20 20 2.0
59572.63 2021 Dec 24 15 2.0
59576.59 2021 Dec 28 11.4 2.0
59577.76 2021 Dec 29 24.92 2.0
Figure 1. A summary of our entire monitoring campaign highlight- 59578.80 2021 Dec 30 15 2.0
ing the dense–set presented in this paper.
Note— Fc denotes the central observing frequency. All
observations were conducted with a bandwidth of 800 MHz.
on the closely spaced S band-only observations conducted
between 2021 Nov 2 (MJD 59520) & 2021 Dec 30 (MJD
1. Using the DSPSR software library (van Straten & Bailes
59578), henceforth referred to as the “dense set” (2021B
2011), we fold the data using the phase-connected tim-
semester; Project code GBT21B-354, PI: Samayra Straal).
ing solution presented in Table 3. This results in an
These observations were carried out in SEARCH mode,
”archive” file.
and the data were recorded using the VErsatile GBT As-
tronomical Spectrometer (VEGAS) backend (Prestage et al.
2015) with a bandwidth of 800 MHz at a central frequency 2. Using the pazi tool in the PSRCHIVE software pack-
of ν = 2 GHz and a 40.96 µs time resolution. The 800 MHz age 1 (Hotan et al. 2004), we manually removed all RFI
bandwidth was divided into 2048 frequency channels for bet- to minimize baseline variations through the following
ter radio frequency interference (RFI) mitigation and more steps:
accurate dispersive delay correction. Each epoch included a Scrunch the updated archive” file in frequency to re-
65 s obsevations of 3C286 used for absolute flux and polar- duce the number of channels, then use the pazi rou-
ization calibration. A log of the dense set observations show- tine to identify and flag RFI-affected sub-integrations.
ing relevant details is shown in Table 2. Use the resulting psrsh command to apply these flags
We analyzed the raw data from each epoch using the pulsar back to the original, full-resolution archive file. This
analysis software PRESTO (Ransom et al. 2002). Below, we produces an archive” file cleaned of RFI in both fre-
outline the detailed steps used to extract the times of arrival quency and time domains.
(ToAs) at each epoch, which are then used to construct the
phase-connected timing solution described in §3.1:
• Scrunch the data in time and use the pazi routine
1. Automatically removing RFI using the rfifind rou- to remove the frequency channels contaminated
tine in PRESTO, by RFI and then generate a psrsh command to
2. Folding the data from each epoch with the prepfold print-out a script to reproduce this interactive ses-
routine, using the timing solution derived by Rajwade sion results and finally apply this to the original
et al. (2022) as the initial ephemeris. ”archive” file from 1.
3. Use the pygaussfit.py routine to generate a tem- • Scrunch the updated “archive” file from the pre-
plate pulse profile by fitting a Gaussian to the highest vious step in frequency and use the pazi routine
S/N observation, which will then be correlated with the to remove the RFI-affected sub-integrations. Use
folded profiles from each epoch. the resulting psrsh command to the original .ar
4. Use the get toas.py routine to extract at least 3 ToAs file. This produces an “archive” file cleaned of
per epoch. RFI in both frequency and time domains.

We used on-source and off-source scans of 3C286 to flux

1 https://psrchive.sourceforge.net/
calibrate the data, following the procedure outlined below:
4 Abdelmaguid et al.

Table 3. Timing parameters of Swift J1818.0−1607 derived from

the observations listed in Table 2

Parameter Value

Right Ascension (J2000) (hh:mm:ss) 18:18:00.23

Declination (J2000) (dd:mm:ss) –16:07:53.00
DM (pc cm−3 ) 710 ± 1
Date Range (MJD) 59536.713 – 59578.804
Epoch of Frequency 59564.805772
F0 (Hz) 0.7326046915(5)
F1 (Hz s−1 ) − 4.4855(11) × 10−12
F2 (Hz s−2 ) 1.2286(85) × 10−19
EPHEM DE200

Note— Both the DM and the the coordinates of the magnetar were
held fixed.
Figure 2. Timing residuals for Swift J1818.0−1607 from the GBT
• Remove RFI from the ON & OFF calibrator observing campaign at 2.0 GHz. The reference MJD is 59564.8.
scans as well as the calibrator scan of the noise
diode using the pazi routine. (2020), as was done by Rajwade et al. (2022)2 . Since the
magnetar was observed at a single frequency band during the
dense set, constraining the DM proved difficult. Each obser-
3. Using the PSRCHIVE software package (Hotan et al.
vation was initially folded using PRESTO, allowing for a DM
2004), we followed the calibration procedure outlined
search around the value reported in the most recent timing
by van Straten et al. (2012) to estimate the system
solution by Rajwade et al. (2022). The resulting folds con-
equivalent flux density (SEFD). Here is a brief sum-
sistently yielded a DM of 710 ± 1 pc cm−3 , in agreement with
mary of the steps:
the value reported by Rajwade et al. (2022). We therefore
adopted this value and fixed the DM in subsequent steps. The
• Create a calibrator database that has the ON and derived phase coherent ephemeris is listed in Table 3 and the
OFF scans of the calibrator source (3C286) using timing residuals are shown in Figure 2. We were unable to
the pac routine. extend the timing solution to later or earlier epochs due to
• Make the “.fluxcal” file and update the cre- significant timing noise, which made maintaining a phase-
ated database by adding this file to it using the connected solution increasingly difficult.
fluxcal routine.
3.2. Pulse Profile Evolution
4. Check that the derived calibrator solutions “.fluxcal” Using the updated ephemeris derived in §3.1 and shown
file is free from RFI. If not, use pazi to remove con- in Table 3, we refolded the data at each epoch to obtain the
taminated channels. corresponding integrated pulse profile. Left panel of Figure
3 shows the temporal evolution of the pulse profile for the
dense−set presented in this paper (∼ 60 days). To make com-
5. Calibrate the data using the derived calibrator solution
parison easier, the profiles have been normalized. The pro-
–“.fluxcal” file– from the calibrator database.
files are characterized by a dominant single-peak structure
with an excess on the leading edge, indicative of a precursor
3. ANALYSIS AND RESULTS component.
3.1. Timing Solution To better characterize the asymmetric morphology, and
With the extracted ToAs, obtained using the procedure accurately measure the full-width half-maximum (FWHM
outlined in §2, we used the TEMPO2 pulsar timing software or W50 ), we fit each profile using a two-component Gaus-
(Hobbs et al. 2006) to derive a phase-connected timing so-
lution following the technique described by Freire & Ridolfi 2 This position was adopted because our timing analysis was conducted
(2018). We kept the RA and DEC values fixed at the best- in 2022, prior to the publication of the updated VLBI position from
known X-ray coordinates reported by Blumer & Safi-Harb Ding et al. (2024).
 5

Figure 3. Pulse Profile Evolution at 2.0 GHz over our monitoring period of ∼ 60 days. All profiles have been normalized for easier comparison.
Right Panel: An example of a double gaussian fit to the pulse profile at MJD 59536, showing the contributions from a narrow main component
(blue dotted curve) and a broader pre-cursor component (green dotted curve). The red dotted curve represents the sum of these individual
components.

Figure 4. Period averaged flux densities as a function of MJD. The y-axis is shown on a logarithmic scale. Our 2.0 GHz measurements are
plotted alongside earlier observations at 2.25 & 8.6 GHz by Huang et al. (2021) and at 1.4 GHz by Rajwade et al. (2022) and future epochs
at 8.6 GHz by Ding et al. (2024). The segments labeled Mode 3 and Mode 4 correspond to distinct mode-switching events, as identified by
Rajwade et al. (2022).

sian model. The primary narrow Gaussian (blue dot- 3.3. Flux Density and Spectral Index
ted curve in Figure 3) accounts for the main pulse Following the calibration procedure outlined in §2, we
component, while the broader, low-amplitude secondary estimated the period-averaged flux density by dividing the
Gaussian (green dotted curve in Figure 3) captures data from each epoch into four sub-bands and integrating the
the pre-cursor component. The model was fitted using area under the flux-calibrated profiles of each sub-band fol-
scipy.optimize.curve fit, which uses the non-linear lowed by dividing by the number of bins. The overall period-
least-squares minimization method. As shown in Figure 3, averaged flux density across the entire frequency band was
this model captures the profile shape more effectively than a then determined by averaging the flux density values from
single Gaussian. The resulting pulse widths, listed in Table 4, the four sub-bands. The random errors on our flux density es-
range from W50 ∼ 18 – 23 ms and exhibit minimal variation timates are tiny, however, the flux density of our primary cal-
across epochs, indicating a stable pulse morphology over the ibrator 3C286 itself could have uncertainties of the order of
monitoring period. The fitted parameters for each Gaussian 5% (Perley & Butler 2017). Moreover, the presence of some
component are summarized in Table A.2. faint residual RFI could also affect the measurements. So,
6 Abdelmaguid et al.

Figure 5. Power–law spectral indices as a function of MJD estimated using the period averaged flux densities along with older data reported
by Huang et al. (2021)
Table 4. Pulse width, flux density, and spectral index measure- dex at each epoch by fitting a power−law model to the mea-
ments for Swift J1818.0−1607 during our monitoring campaign at sured flux densities across the 4 sub-bands above. We per-
2.0 GHz. formed this fit using the non-linear least squares optimization
MJD Pulse Width (FWHM) Sν α method available through scipy.optimize.curve fit.
- (ms) (mJy) - We assumed that, at a given frequency ν, the flux den-
sity, S ν follows the relationship S ν ∝ να . The correspond-
59520.76 18.6(8) 0.25(3) −0.37(9) ing spectral index values (α) at each epoch are provided in
59536.71 18.9(2) 0.21(2) −0.15(16) Table 4, and its evolution over time is illustrated in Fig-
59550.88 18.6(8) 0.27(3) −0.17(31) ure 5 along with α measurements at earlier epochs (Huang
59559.84 20.0(2) 0.27(3) −0.20(20) et al. 2021). Our spectral index measurements indicate that
59564.80 22.6(8) 0.23(2) −0.30(17) Swift J1818.0−1607 continues to exhibit a flat radio spec-
59568.72 21.3(5) 0.27(3) 0.24(24) trum, with α ≳ − 1, consistent with previous results reported
59572.63 18.6(8) 0.21(2) −0.77(8) by Huang et al. (2021), particularly between MJD 59015 and
59576.59 18.6(8) 0.33(3) 0.33(23) 59090, as shown in Figure 5.
59577.76 22.6(8) 0.26(3) 0.72(30)
59578.80 20.0(2) 0.31(3) −0.10(23) 4. DISCUSSION
Note— The table lists the fitted full-width at half-maximum 4.1. Spin-down Behaviour & Mode Switching
(FWHM) pulse widths, period-averaged flux densities (S ν ) at 2.0
Mode switching refers to the abrupt transition between two
GHz, and spectral indices (α) at each observing epoch. Values are
quoted with 1 − σ uncertainties in parentheses, representing the
or more distinct emission states, thought to arise as a re-
uncertainty in the last digit(s). sult of changes in the neutron star’s magnetosphere (Timo-
khin 2010; Wang et al. 2007). Swift J1818.0−1607 has ex-
hibited multiple distinct emission modes, reflected in both
conservatively, we assume the corresponding flux density un- short- and long-term variations in its pulse profile. Between
certainites to be 10%. 2020 June and 2021 August, mode switching was observed at
The period-averaged flux density values each epoch are 1.4 GHz (Rajwade et al. 2022) and simultaneously at 2.2 and
listed in Table 4. Our measurements indicate a relatively low 8.5 GHz (Huang et al. 2021; Bansal et al. 2023). On minute
flux density for the source, ranging between S ν ∼ 0.2 − 0.3 timescales, Lower et al. (2020) reported two mode-switching
mJy. Figure 4 shows a plot of the evolution of our flux den- behaviors below 4 GHz, characterized by morphologies dis-
sity measurements with time along with measurements from tinct from those seen at higher frequencies. Since the closest
earlier epochs by Huang et al. (2021) & Rajwade et al. (2022) published observations to our dataset are those of Rajwade
and some future epochs by Ding et al. (2024). This will be et al. (2022), we compare our results to this work. In their
discussed later in §4.1. study, they identified distinct emission states, including the
Finally, given the large bandwidth of 800 MHz in the segments labeled Mode 3 and Mode 4 (Figures 4 & 6). These
GBT’s S-band receiver, we estimated the in-band spectral in- modes were characterized by differences in pulse profile mor-
 7

Figure 6. Spin frequency evolution of Swift J1818.0−1607 over time, showing measurements from different studies (Champion et al. 2020b;
Huang et al. 2021; Rajwade et al. 2022), and this work. The segments labeled Mode 3 and Mode 4 correspond to distinct spin-down states
associated with mode-switching events as identified by Rajwade et al. (2022), each with a characteristic frequency derivative (ν̇) indicated by
the fitted slopes. Data from Huang et al. (2021) was provided by Zhi-Peng Huang, who also shared the dataset from Champion et al. (2020b),
which had been extracted using the engauge digitizer app. Data from Rajwade et al. (2022) was provided by Kaustubh Rajwade.

phology, spin-down behaviour and flux density. Crucially, As mentioned earlier and shown in Figure 3, the integrated
Rajwade et al. (2022) noted that these transitions between pulse profiles throughout our ∼ 60-day observing campaign
different modes or ”mode switches” were correlated with a consistently feature a single, narrow peak with little varia-
change in ν̇ (see Figure 2 in Rajwade et al. 2022), similar tion in shape, that closely resembles the pulse profile seen in
to what is observed for other mode-changing pulsars (Lyne emission mode 3 (Rajwade et al. 2022; Figure 3c). In con-
et al. 2010). trast, the closest published pulse profile to our data (near
While different from the mode-switching seen in Swift MJD 59400, Rajwade et al. 2022, Figure 3d) displayed a
J1818.0−1607, the radio-loud magnetar PSR J1622−4950 double-peaked structure with a precursor component. This
showed another form of coupled spin and profile evolution, indicates that the magnetar has since transitioned to a single-
marked by a steady decline in its spin-down rate from 2011 peak profile and remained in that state throughout our moni-
to 2014, accompanied by a gradual narrowing of its pulse toring period.
profile (Scholz et al. 2017). However, other radio magnetars To compare the spin–down behavior of Swift J1818.0−1607
do not show clear evidence of a similar secular decrease in across different epochs, we first measured the spin frequency
profile width. Additionally, the magnetar’s spin-down rate at each individual epoch in our dataset. We then fitted linear
reached a minimum just before the source became radio- slopes to the spin frequency data for Mode 3 (MJD 59170 –
silent, rising again when pulsations reactivated (Scholz et al. 59300) and Mode 4 (MJD 59300 – 59426), as well as for our
2017; Camilo et al. 2018). In contrast, XTE J1810−197 dis- own data spanning MJD 59520 – 59578. The comparison
played significant pulse profile variability throughout its ac- shown in Figure 6 shows that the spin-down rate observed in
tive phase but no evidence of a systematic or gradual nar- our dataset (ν̇ = − 4.6 × 10−12 Hz s−1 ) is notably similar to
rowing (Camilo et al. 2016; Levin et al. 2019). Similar to that of Mode 3 (ν̇ = − 3.1 × 10−12 Hz s−1 ), suggesting a likely
PSR J1622−4950, its spin-down rate also reached minimum return to a similar emission state following the more rapid
values just before radio silence and increased again upon re- spin-down episode seen in Mode 4 (ν̇ = − 2.2×10−11 Hz s−1 ).
activation (Camilo et al. 2016; Levin et al. 2019). Concurrently, our flux density measurements show that
A related behavior has been observed in PSR J1119−6127, Swift J1818.0−1607 remained in a persistently low-flux state
a high magnetic field pulsar with magnetar-like characteris- with minimal variability throughout our observations (Fig-
tics. Following its outbursts in 2016, an increase in the spin- ure 4). This overall behaviour closely resembles the rela-
down rate was observed to be correlated with an increase in tively steady trend seen in Mode 3, albeit at systematically
its radio flux density (Dai et al. 2018). Altogether, the corre- lower flux levels, and contrasts with the declining flux char-
lated changes in spin-down rate and radio emission observed acteristic of Mode 4 observed immediately prior to our cam-
in Swift J1818.0−1607 are not anomalous, but rather align paign. It is important to note that the flux density measure-
with trends seen in other magnetars and high B-field pulsars. ments shown in Figure 4 span a range of observing frequen-
cies (1.4 − 8.6 GHz). Since magnetar emission can exhibit
8 Abdelmaguid et al.

frequency-dependent behavior, these spectral variations may a mostly consistent and comparatively narrower pulse profile
introduce biases when comparing flux measurements taken widths (W50 ∼ 16 − 18 ms) at higher frequencies between 6
at different epochs. This pattern of highly variable spin-down and 22 GHz (MJD 59092 − 59190). This narrowing of pulse
rates is not exclusively limited to Swift J1818.0−1607. For width at higher frequencies is consistent with what has been
example, 1E 1048.1−5937 exhibited repeated increases and observed in many RPPs (Pilia et al. 2016) and was also ob-
decreases in its spin-down rate—by up to an order of magni- served in XTE J1810−197 (Eie et al. 2021). In canonical pul-
tude—within ∼ 100 – 600 days following its multiple X-ray sars, the evolution of pulse width with frequency is explained
outbursts in 2002, 2007, 2012 (Archibald et al. 2015). Fol- by radius-frequency mapping, which predicts that the pulse
lowing its 2016 outburst, 1E 1048.1−5937 showed a monotic profile becomes narrower at higher radio frequencies (Cordes
decline in the magnitude of its spin-down variations; how- 1978).
ever, its long-term spin-down evolution remains dominated
by recurrent fluctuations (Archibald et al. 2020).
1E 1547.0−5408, another magnetar, underwent a rapid 4.3. Flux Density & Spectral Indices
spin-down rate increase after its 2008, 2009 and 2022 out-
Our flux density measurements for Swift J1818.0−1607 in-
bursts, with significant variability persisting thereafter (Dib
dicate a continuation of the declining trend observed around
et al. 2012; Kuiper et al. 2012; Lower et al. 2023). Sim-
MJD 59300 (Figure 4), with the source exhibiting a relatively
ilar variability in the spin-down behaviour has also been
stable, low-level radio emission state throughout our observ-
observed in PSR J1622−4950 (Scholz et al. 2017; Camilo
ing campaign. Placing Swift J1818.0−1607 in the context of
et al. 2018), and in XTE J1810−197 (Camilo et al. 2016;
other radio magnetars can help clarify its potential evolution-
Levin et al. 2019; Huang et al. 2023). Additionally, the high-
ary trajectory.
magnetic-field pulsar PSR J1846−0258 showed erratic spin
PSR J1622−4950, the third discovered radio-loud magne-
evolution during its 2020 outburst, including a spin-up fol-
tar, displayed a downward trend approximately two years af-
lowed by a steady decline in spin-down rate (Sathyaprakash
ter its discovery in 2009 (Levin et al. 2010), before enter-
et al. 2024). These examples demonstrate that spin-down
ing a low-flux regime by 2013, and stopped emitting any
variability is a common feature among magnetars in the
detectable radio emission entirely around five years post-
aftermath of outbursts, especially when they are observed
discovery (Scholz et al. 2017), before it was revived again
over long timescales. Thus, the variability observed in
in 2017 (Camilo et al. 2018). A similar pattern was seen
Swift J1818.0−1607 appears consistent with the broader be-
in XTE J1810−197, which exhibited bright and highly vari-
havior of magnetar population. However, given that only
able radio emission (Camilo et al. 2007b,c) after its 2003
a comparatively short time has passed since its outburst in
outburst (Ibrahim et al. 2004), followed by a steady de-
2020, continued long-term monitoring will be essential to de-
cline and eventual disappearance in late 2008 (Camilo et al.
termine whether it follows similar evolutionary trends.
2016), before it got reactivated again in 2018 (Gotthelf et al.
Therefore, taken together; the change in pulse profile,
2019). In contrast, 1E 1547.0−5408 was initially a bright and
spin−down rate (ν̇) and flux density between Mode 4 and
persistently detectable radio source after its 2007 discovery
the onset of our observations suggests that another mode
(Gelfand & Gaensler 2007; Camilo et al. 2007a). Follow-
switching episode likely occurred during the ∼100-day gap
ing two X-ray outbursts in 2008 and 2009 (Dib et al. 2012;
(MJD 59426 to MJD 59520) between the last published tim-
Kuiper et al. 2012), its radio emission became markedly spo-
ing campaign (Rajwade et al. 2022) and the start of ours.
radic, with extended periods of non-detection lasting several
This underscores the importance of continued monitoring of
months (Camilo et al. 2009; Burgay et al. 2009). More re-
Swift J1818.0–1607 to capture potential future transitions or
cently, during another outburst in 2022, observations showed
changes in its activity, given the magnetar’s highly dynamic
that the persistent radio emission from 1E 1547.0−5408 dis-
and evolving behavior.
appeared at least 22 days before the initial Swift−BAT detec-
tion and was subsequently re-detected about two weeks later
4.2. Pulse Widths (Lower et al. 2023).
In our study of Swift J1818.0−1607, we observed pulse As of Febraury 2023, Swift J1818.0−1607 continues to
widths ranging from W50 ∼ 18−23 ms) at 2.0 GHz, implying be detected in radio (Ding et al. 2024), however continued
a stable integrated profile shape throughout the observation long-term monitoring will be crucial to determine whether
period. This is in contrast with the variability reported im- Swift J1818.0−1607 eventually follows the fading and reacti-
mediately following the magnetar’s outburst, where the pulse vation cycles seen in PSR J1622−4950 and XTE J1810−197,
profile transitioned from a broader (W50 ∼ 40−80 ms) to nar- or transitions into a more sporadic and intermittent emission
rower one (W50 ∼ 10 − 45 ms) at (1.37 − 2.55) GHz (Cham- state like 1E1547.0−5408 or follows a unique evolutionary
pion et al. 2020b). Additionally, Lewis et al. (2025) reports path. Our future work will shed light on the long-term evolu-
 9

Table 5. A comparison of the characteristic age (τc ), spin-down inferred magnetic field (B f ield ) and spin−down luminosity (Ė) estimated in this
paper with values reported in previous studies and observational campaigns.

Reference MJD τc Spin−down Inferred B f ield Ė

Champion et al. (2020a) 58922 ∼ 265 yr ∼ 3.4 × 1014 G ∼ 1.6 × 1036 erg s−1
Champion et al. (2020b) 58922 – 59015 ∼ 500 yr ∼ 2.5 × 1014 G ∼ 7 × 1035 erg s−1
Huang et al. (2021) 58936 – 59092 ∼ 500 yr ∼ 2.5 × 1014 G ∼ 7 × 1035 erg s−1
Rajwade et al. (2022) 59117 – 59400 ∼ 1000 yr ∼ 2 × 1014 G ∼ 4 × 1035 erg s−1
This Work 59520 – 59578 ∼ 2500 yr ∼ 1 × 1014 G ∼ 1 × 1035 erg s−1

Note—The characteristic age (τc ) reported by Champion et al. (2020a) represents the first estimate for Swift J1818.0−1607. Subsequent
values from Champion et al. (2020b) and Huang et al. (2021) were inferred by fitting a linear function to the spin frequency measured at each
epoch during their respective observing campaigns. Estimates from Rajwade et al. (2022) and this work are derived from timing solutions.

tion of Swift J1818.0−1607 and its place within the broader

population of radio magnetars.
Following its outburst, the pulsed radio spectrum of
Swift J1818.0−1607 was initially steep (α between − 3.6
and − 1.8) (Maan & van Leeuwen 2020; Lower et al. 2020),
but gradually flattened over time (Majid et al. 2020), with
a transition from a steep spectrum (α < − 1.48, MJD
58936−58944) to a flat one (α > − 0.63, MJD 59015–59092)
(Huang et al. 2021; Bansal et al. 2023). Lower et al. (2021)
notes that this observed spectral flattening was driven by
the emergence of a new profile component with an in-
verted spectrum, which persisted at high frequencies, while
the original steep−spectrum component gradually faded and
merged with it by MJD 59128. Our observations show that
Swift J1818.0−1607 continues to exhibit a flat spectrum
(α ≳ −1), in contrast to the steep spectrum seen shortly
after the outburst (Figure 5). A similar spectral evolution
Figure 7. P−Ṗ diagram showing the position of Swift
was observed in PSR J1119–1627, a RPP that has exhib- J1818.0−1607. The inset provides a zoomed−in view highlighting
ited magnetar-like outbursts (Archibald et al. 2016; Dai et al. the temporal evolution of its spin period and spin−down rate across
2018). These characteristics align more closely with RPPs different epochs, with four data points (from top to bottom) cor-
than with typical radio magnetars, leading to speculation that responding to measurements reported by Champion et al. (2020b),
Swift J1818.0−1607 may serve as an evolutionary link be- Huang et al. (2021), Rajwade et al. (2022), and this work.
tween the two populations (Hu et al. 2020).
Due to the limited bandwidth of these observations, we can 4.4. Inferred Properties of Swift J1818.0−1607 on the P−Ṗ
not investigate the possibility of a spectral turnover at higher Diagram
frequency, as seen by Huang et al. (2021), where in one of Using the derived timing solution (Table 3), we infer a sur-
their epochs (MJD 59076), they get an α = +0.5 below 32 face dipole magnetic field √ strength of B ∼ 1 × 1014 G, esti-
GHz, whereas Torne et al. (2020) reports α = − 1.4 between
mated using B = 3.2×10 19
PṖ G, where P is the spin period
86 and 154 GHz at the same epoch. Such turnover has also
and Ṗ is its time derivative. This value is consistent with the
been observed in another magnetar XTE J1810−197. How-
typical range observed in the magnetar population (Kaspi &
ever, we note that the spectral turnover in XTE J1810−197
Beloborodov 2017), but is ∼ 3 times lower than the value
occurs at substantially lower radio frequencies (MHz to a
of B ∼ 3.5 × 1014 G, originally reported after the outburst
few GHz) compared to the tens of GHz turnover reported for
(Champion et al. 2020a). Table 5 summarizes the evolution
Swift J1818.0−1607. As such, it remains unclear whether
of key spin-down parameters for Swift J1818.0−1607 across
these turnovers are directly comparable, as they may origi-
multiple observational campaigns (Champion et al. 2020a,b;
nate from distinct physical processes.
Huang et al. 2021; Rajwade et al. 2022), highlighting the de-
creasing trend in inferred magnetic field strength with time.
10 Abdelmaguid et al.

Figure 7 shows the position of Swift J1818.0−1607 in the 5. SUMMARY

P – Ṗ, highlighting its location within the broader pulsar and We carried out observations of the radio magnetar
magnetar population. The inset provides a zoomed-in view of Swift J1818.0–1607 with the GBT at a central frequency of
its temporal evolution, with four data points corresponding to 2.0 GHz in November 2021, approximately 20 months after
P – Ṗ measurements reported in previous studies (Champion its March 2020 outburst. From these observations, we derived
et al. 2020b; Huang et al. 2021; Rajwade et al. 2022), and a phase-connected timing solution spanning the duration of
this work. The downward trend in Ṗ over time reflects sig- our monitoring campaign.
nificant variability in the magnetar’s spin-down rate, possibly Using our phase-connected timing solution, we infer a
driven by evolving magnetospheric conditions. This apparent characteristic age of ∼ 2500 years for Swift J1818.0–1607,
movement in the P – Ṗ diagram underscores the importance significantly older than previous estimates (see Table 5). This
of continued long-term monitoring, to better understand the increase reflects the continued evolution of the magnetar’s
long-term spin-down properties of this magnetar. spin-down rate, and underscores the limitations of inferring
Using the same timing solution, we infer a characteris- its τc from P − Ṗ measurements, particularly for young, ac-
tic age of τc ∼ 2500 years for Swift J1818.0–1607, cal- tive magnetars with highly variable spin-down behavior.
culated under the assumption of a constant braking index Throghout our observing campaign, Swift J1818.0–1607
of n = 3, using the standard formula τc = P/2Ṗ. This exhibited a stable pulse profile, characterized by a single,
estimate is roughly an order of magnitude higher than the narrow peak (W50 ∼ 18 − 23 ms) at 2.0 GHz with a small pre-
first reported age of ∼ 265 years during the initial outburst cursor component and no evidence of a postcursor, and main-
phase (Champion et al. 2020a), and ∼ 2.5 times older than taned a magnetar-like flat radio spectrum. Our analysis sug-
the most recent estimate (τc ∼ 1000 years, Rajwade et al. gests that the magnetar likely underwent a mode−switching
2022). Table 5 summarizes the various age estimates for episode during the ∼100-day gap (MJD 59426 – 59520) pre-
Swift J1818.0–1607. Taken together, these results suggest ceding our observations, as indicated by changes in the pulse
that the magnetar has effectively aged by ∼ 2000 years over profile, spin−down rate, and flux density relative to the emis-
the past two years, reflecting a significant and ongoing evo- sion mode (Mode 4) reported in earlier epochs (Rajwade
lution in its spin−down behavior. et al. 2022). The observed variability in the spin-down rate of
It is important to note that the characteristic age may not Swift J1818.0–1607 is not surpising, as similar behavior has
accurately reflect the true age of the magnetar. This estimate been observed in other magnetars and high B-field pulsars
assumes a constant braking index of n = 3, which may not (Dib et al. 2012; Kuiper et al. 2012; Archibald et al. 2015,
hold in practice, as the braking index can differ from this 2016; Camilo et al. 2016; Scholz et al. 2017; Dai et al. 2019;
value or evolve over time (Johnston & Karastergiou 2017). Levin et al. 2019; Lower et al. 2023; Huang et al. 2023).
For instance, if the braking index is less than 3, as observed In future work, we plan on presenting a comprehensive
in a few magnetars (Gao et al. 2016), it would result in a analysis of multi-frequency data from the rest of our monitor-
change in the magnetar’s inferred age. Additionally, given ing campaign covering the period from MJD 59700 to MJD
that magnetars often exhibit irregular variations in their spin- 60300. Our primary focus will be on the long-term evolu-
down rate (ν̇), particularly during or shortly after outbursts, it tion of the pulsed radio emission from Swift J1818.0–1607
can lead to a wide range of age estimates, as demonstrated in in both frequency and time, by leveraging contemporaneous
this study. Therefore, obtaining independent constraints on multi-frequency observations in the L and C bands across
the age of Swift J1818.0–1607 is crucial. While no coinci- most epochs.
dent supernova remnant (SNR) has previously been identified
to directly constrain the magnetar’s age (Green 2025; Ferrand ORCID IDS
& Safi-Harb 2012), recent radio observations have revealed
Moaz Abdelmaguid: 0000-0002-4441-7081
diffuse radio emission around Swift J1818.0–1607 (Ibrahim
Paulo Freire: 0000-0003-1307-9435
et al. 2023), suggesting that it originates from the shell of the
Joseph Gelfand: 0000-0003-4679-1058
supernova remnant where the magnetar was formed. This hy-
Yogesh Maan: 0000-0002-0862-6062
pothesis is further corroborated by recent VLBI observations,
Samayra Straal: 0000-0003-4136-7848
which measured the magnetar’s 3−D velocity to be 190 km
J. A. J. Alford: 0000-0002-2312-8539
s−1 (Ding et al. 2024), indicating that the source has likely not
traveled far from where it was born. High-resolution follow-
up radio observations are essential to robustly characterize
the diffuse emission and determine whether its spectral and
morphological properties are consistent with an SNR shell.
 11

ACKNOWLEDGEMENTS jwade for sending their data which has been used in this pub-
The authors would like to thank the anonymous referee lication.
for the comments and suggestions that helped make this
FACILITIES & SOFTWARES
manuscript better.
M.A. is supported by a graduate research assistanship at Facilities: GBT
NYU Abu Dhabi, which is funded by the Executive Affairs Software: TEMPO2 (Hobbs et al. 2006),
Authority of the Emirate of Abu Dhabi, as adminstrated by PRESTO (Ransom et al. 2002), DSPSR (van Straten & Bailes
Tamkeen. The research of J.D.G is funded by NYUAD grant 2011), PSRCHIVE (Hotan et al. 2004) & Scipy (Virtanen
AD 022. Both M.A. and J.D.G. received support form the et al. 2020).
NYU Abu Dhabi Research Institute grant to the CASS. The
National Radio Astronomy Observatory and Green Bank Ob- DATA AVAILABILITY
servatory are facilities of the U.S. National Science Foun- The data underlying Figure 2 is listed in Table A.1 in the
dation operated under cooperative agreement by Associated Appendix. Those underlying Figures 4 and 5 are provided in
Universities. Table 4. Additionally, the data for Figure 6 is presented in Ta-
M.A. would like to thank Scott M. Ransom and Ryan S. ble A.3 in the Appendix. Finally, the data tracing the trajec-
Lynch for answering questions related to PRESTO. M.A. tory of Swift J1818.0−1607 on the P–Ṗ diagram in Figure 7
would also like to thank Zhi-Peng Huang and Kaustubh Ra- is listed in Table 5.

APPENDIX

In Table A.1, we provide the full set of extracted time-of-arrival (ToA) measurements used to construct the phase-connected
timing solution in Table 3. Their timing residuals are shown in Figure 2. Each row lists the observatory code, frequency, MJD
of the ToA, and the associated uncertainty in microseconds. Table A.3 lists the spin frequency measurements obtained from
our observing campaign, which were used in Figure 6. Each measurement corresponds to a single epoch, with σν denoting the
statistical uncertainty associated with each spin frequency measurement.

Table A.1. Time of arrival (ToA) measurements for Swift J1818.0−1607 taken
with the GBT. Each row lists the observatory code (GBT = 1), observing fre-
quency (MHz), MJD of the ToA, and the uncertainty in µs. These data were used
to construct the phase-connected timing solution listed in Table 3, and whose
residuals are shown in Figure 2.

Obs. Code Frequency (MHz) MJD Uncertainty (µs)

1 2000.195 59520.7650042251161 360.52

1 2000.195 59520.7660469857919 320.79
1 2000.195 59520.7670739406895 382.48
1 2000.195 59520.7681008908125 319.86
1 2000.195 59520.7711975653497 334.83
1 2000.195 59536.7135290693987 353.97
1 2000.195 59536.7152511941632 356.62
1 2000.195 59536.7169733116307 338.93
1 2000.195 59536.7204017445736 339.84
1 2000.195 59550.8868235680141 311.14
1 2000.195 59550.8885456768594 309.47
1 2000.195 59550.8868235630431 329.95
1 2000.195 59550.8888932390432 443.79
1 2000.195 59559.8345127531953 407.93
1 2000.195 59559.8365824132791 366.77
1 2000.195 59559.8376093264854 417.73
1 2000.195 59559.8386362598791 542.49
12 Abdelmaguid et al.

Obs. Code Frequency (MHz) MJD Uncertainty (µs)

1 2000.195 59559.8407059098786 321.52

1 2000.195 59559.8427755654863 410.58
1 2000.195 59559.8438024861103 425.01
1 2000.195 59564.8013005036540 401.33
1 2000.195 59564.8043970641237 402.85
1 2000.195 59564.8085205477038 370.73
1 2000.195 59564.8095474593230 380.80
1 2000.195 59568.7141911149129 420.68
1 2000.195 59568.7148862444787 440.61
1 2000.195 59568.7169558566322 408.80
1 2000.195 59568.7453303473052 343.72
1 2000.195 59572.6262417115981 768.09
1 2000.195 59572.6272686235020 465.28
1 2000.195 59572.6293382250276 370.71
1 2000.195 59572.6313920679945 683.49
1 2000.195 59576.5999669558361 453.98
1 2000.195 59576.6009938571775 363.49
1 2000.195 59576.6040903807802 337.17
1 2000.195 59577.7544138512075 341.59
1 2000.195 59577.7561358842822 317.18
1 2000.195 59577.7578579233058 437.16
1 2000.195 59577.7595799644018 396.98
1 2000.195 59577.7818085152489 537.15
1 2000.195 59577.7859477464216 325.39
1 2000.195 59577.7869746523945 396.94
1 2000.195 59577.7910980569357 444.58
1 2000.195 59578.8012713277590 303.80
1 2000.195 59578.8029933650074 358.54
1 2000.195 59578.8064374293458 376.16
1 2000.195 59578.8081436699751 569.32
 13

Table A.2. Pulse width, phase, and amplitude measurements for Swift J1818.0−1607 during our monitoring campaign at 2.0 GHz.

MJD Main Component Pre-cursor Component

- Width (ms) Phase Amplitude Width (ms) Phase Amplitude

59520.76 17.3(5) 0.375 0.81 48.0(3) 0.363 0.20

59536.71 18.6(8) 0.374 0.94 41.3(6) 0.359 0.12
59550.88 17.3(5) 0.374 0.84 41.3(6) 0.365 0.18
59559.84 17.3(5) 0.375 0.86 34.6(9) 0.366 0.21
59564.80 20.0(1) 0.375 0.90 36.0(3) 0.362 0.21
59568.72 20.0(3) 0.374 0.853 36.0(4) 0.364 0.17
59572.63 14.6(8) 0.376 0.72 37.3(6) 0.367 0.33
59576.59 17.3(5) 0.374 0.85 41.3(6) 0.364 0.18
59577.76 20.0(1) 0.374 0.83 41.3(6) 0.364 0.20
59578.80 18.6(8) 0.374 0.87 44.0(3) 0.366 0.16

Note—Gaussian fit parameters of the main and pre-cursor pulse components of Swift J1818.0−1607 observed at 2.0 GHz during our
monitoring campaign. For each epoch, we report the full-width at half-maximum (FWHM) in ms, the fitted pulse phase, and the normalized
amplitude of each component. Uncertainties are quoted at the 1σ level in parentheses and reflect the uncertainty in the last digit(s).

Table A.3. Spin frequency measurements of Swift J1818.0−1607 from our observing campaign. σν represents the statistical uncertainty
associated with each spin frequency measurement. These data were used to create Figure 6.

MJD ν (Hz) σν (Hz)

59520.76339 0.7326252354 6.07 × 10−7
59536.71396 0.7326148968 5.77 × 10−7
59550.88215 0.7326072100 3.94 × 10−7
59559.83677 0.7326073648 4.17 × 10−7
59564.79993 0.7326066743 4.96 × 10−7
59568.70900 0.7326099055 7.07 × 10−7
59568.73670 0.7325946103 1.16 × 10−6
59572.62728 0.7325980647 4.81 × 10−7
59576.59604 0.7326033346 8.00 × 10−7
59577.74870 0.7326008607 8.17 × 10−7
59577.77600 0.7325999324 6.20 × 10−7
59578.79600 0.7326012217 6.29 × 10−7

REFERENCES
Archibald, R. F., Kaspi, V. M., Ng, C. Y., et al. 2015, ApJ, 800, 33, Camilo, F., Halpern, J. P., & Ransom, S. M. 2009, The
doi: 10.1088/0004-637X/800/1/33 Astronomer’s Telegram, 1907, 1
Archibald, R. F., Kaspi, V. M., Tendulkar, S. P., & Scholz, P. 2016, Camilo, F., Ransom, S. M., Halpern, J. P., & Reynolds, J. 2007a,
ApJL, 829, L21, doi: 10.3847/2041-8205/829/1/L21 ApJL, 666, L93, doi: 10.1086/521826
Archibald, R. F., Scholz, P., Kaspi, V. M., Tendulkar, S. P., & Camilo, F., Ransom, S. M., Halpern, J. P., et al. 2006, Nature, 442,
Beardmore, A. P. 2020, ApJ, 889, 160,
892, doi: 10.1038/nature04986
doi: 10.3847/1538-4357/ab660c
Camilo, F., Reynolds, J., Johnston, S., et al. 2007b, ApJL, 659,
Bansal, K., Wharton, R. S., Pearlman, A. B., et al. 2023, MNRAS,
L37, doi: 10.1086/516630
523, 2401, doi: 10.1093/mnras/stad1520
Camilo, F., Ransom, S. M., Peñalver, J., et al. 2007c, ApJ, 669,
Blumer, H., & Safi-Harb, S. 2020, ApJL, 904, L19,
doi: 10.3847/2041-8213/abc6a2 561, doi: 10.1086/521548
Burgay, M., Israel, G. L., Possenti, A., et al. 2009, The Camilo, F., Ransom, S. M., Halpern, J. P., et al. 2016, ApJ, 820,
Astronomer’s Telegram, 1913, 1 110, doi: 10.3847/0004-637X/820/2/110
14 Abdelmaguid et al.

Camilo, F., Scholz, P., Serylak, M., et al. 2018, ApJ, 856, 180, Ibrahim, A. Y., Borghese, A., Rea, N., et al. 2023, ApJ, 943, 20,
doi: 10.3847/1538-4357/aab35a doi: 10.3847/1538-4357/aca528
Champion, D., Desvignes, G., Jankowski, F., et al. 2020a, The Jankowski, F., van Straten, W., Keane, E. F., et al. 2018, MNRAS,
Astronomer’s Telegram, 13559, 1 473, 4436, doi: 10.1093/mnras/stx2476
Champion, D., Cognard, I., Cruces, M., et al. 2020b, MNRAS, Johnston, S., & Karastergiou, A. 2017, MNRAS, 467, 3493,
498, 6044, doi: 10.1093/mnras/staa2764 doi: 10.1093/mnras/stx377
Cordes, J. M. 1978, ApJ, 222, 1006, doi: 10.1086/156218 Karuppusamy, R., Desvignes, G., Kramer, M., et al. 2020, The
Dai, S., Johnston, S., Weltevrede, P., et al. 2018, MNRAS, 480, Astronomer’s Telegram, 13553, 1
3584, doi: 10.1093/mnras/sty2063 Kaspi, V. M., & Beloborodov, A. M. 2017, ARA&A, 55, 261,
Dai, S., Lower, M. E., Bailes, M., et al. 2019, ApJL, 874, L14, doi: 10.1146/annurev-astro-081915-023329
doi: 10.3847/2041-8213/ab0e7a Kouveliotou, C., Norris, J. P., Cline, T. L., et al. 1987, ApJL, 322,
Dib, R., Kaspi, V. M., Scholz, P., & Gavriil, F. P. 2012, ApJ, 748, 3, L21, doi: 10.1086/185029
doi: 10.1088/0004-637X/748/1/3 Kuiper, L., Hermsen, W., den Hartog, P. R., & Urama, J. O. 2012,
Ding, H., Lower, M. E., Deller, A. T., et al. 2024, ApJL, 971, L13, ApJ, 748, 133, doi: 10.1088/0004-637X/748/2/133
doi: 10.3847/2041-8213/ad5550 Levin, L., Bailes, M., Bates, S., et al. 2010, ApJL, 721, L33,
Duncan, R. C., & Thompson, C. 1992, ApJL, 392, L9, doi: 10.1088/2041-8205/721/1/L33
doi: 10.1086/186413 Levin, L., Lyne, A. G., Desvignes, G., et al. 2019, MNRAS, 488,
Eie, S., Terasawa, T., Akahori, T., et al. 2021, PASJ, 73, 1563, 5251, doi: 10.1093/mnras/stz2074
doi: 10.1093/pasj/psab098 Lewis, E. F., Blumer, H., Lynch, R. S., & McLaughlin, M. A. 2025,
Enoto, T., Sakamoto, T., Younes, G., et al. 2020, The Astronomer’s arXiv e-prints, arXiv:2502.15200,
Telegram, 13551, 1 doi: 10.48550/arXiv.2502.15200
Esposito, P., Rea, N., Borghese, A., et al. 2020, ApJL, 896, L30, Lower, M. E., Johnston, S., Shannon, R. M., Bailes, M., & Camilo,
doi: 10.3847/2041-8213/ab9742 F. 2021, MNRAS, 502, 127, doi: 10.1093/mnras/staa3789
Ferrand, G., & Safi-Harb, S. 2012, Advances in Space Research, Lower, M. E., Shannon, R. M., Johnston, S., & Bailes, M. 2020,
49, 1313, doi: 10.1016/j.asr.2012.02.004 ApJL, 896, L37, doi: 10.3847/2041-8213/ab9898
Fisher, R., Butterworth, E. M., Rajwade, K. M., et al. 2024, Lower, M. E., Younes, G., Scholz, P., et al. 2023, ApJ, 945, 153,
MNRAS, 528, 3833, doi: 10.1093/mnras/stae271 doi: 10.3847/1538-4357/acbc7c
Freire, P. C. C., & Ridolfi, A. 2018, MNRAS, 476, 4794, Lyne, A., Hobbs, G., Kramer, M., Stairs, I., & Stappers, B. 2010,
doi: 10.1093/mnras/sty524 Science, 329, 408, doi: 10.1126/science.1186683
Gao, Z. F., Li, X. D., Wang, N., et al. 2016, MNRAS, 456, 55, Maan, Y., & van Leeuwen, J. 2020, The Astronomer’s Telegram,
doi: 10.1093/mnras/stv2465 13560, 1
Gehrels, N., Chincarini, G., Giommi, P., et al. 2004, ApJ, 611, Majid, W. A., Pearlman, A. B., Prince, T. A., et al. 2020, The
1005, doi: 10.1086/422091 Astronomer’s Telegram, 13649, 1
Gelfand, J. D., & Gaensler, B. M. 2007, ApJ, 667, 1111, Olausen, S. A., & Kaspi, V. M. 2014, ApJS, 212, 6,
doi: 10.1086/520526 doi: 10.1088/0067-0049/212/1/6
Gotthelf, E. V., Halpern, J. P., Alford, J. A. J., et al. 2019, ApJL, Perley, R. A., & Butler, B. J. 2017, ApJS, 230, 7,
874, L25, doi: 10.3847/2041-8213/ab101a doi: 10.3847/1538-4365/aa6df9
Green, D. A. 2025, Journal of Astrophysics and Astronomy, 46, 14, Pilia, M., Hessels, J. W. T., Stappers, B. W., et al. 2016, A&A, 586,
doi: 10.1007/s12036-024-10038-4 A92, doi: 10.1051/0004-6361/201425196
Hobbs, G., Edwards, R., & Manchester, R. 2006, Chinese Journal Prestage, R. M., Bloss, M., Brandt, J., et al. 2015, in 2015
of Astronomy and Astrophysics Supplement, 6, 189 URSI-USNC Radio Science Meeting, 4,
Hotan, A. W., van Straten, W., & Manchester, R. N. 2004, PASA, doi: 10.1109/USNC-URSI.2015.7303578
21, 302, doi: 10.1071/AS04022 Rajwade, K. M., Stappers, B. W., Lyne, A. G., et al. 2022,
Hu, C.-P., Begiçarslan, B., Güver, T., et al. 2020, ApJ, 902, 1, MNRAS, 512, 1687, doi: 10.1093/mnras/stac446
doi: 10.3847/1538-4357/abb3c9 Ransom, S. M., Eikenberry, S. S., & Middleditch, J. 2002, AJ, 124,
Huang, Z.-P., Yan, Z., Shen, Z.-Q., et al. 2021, MNRAS, 505, 1788, doi: 10.1086/342285
1311, doi: 10.1093/mnras/stab1362 Sathyaprakash, R., Rea, N., Coti Zelati, F., et al. 2024, ApJ, 976,
—. 2023, ApJ, 956, 93, doi: 10.3847/1538-4357/acf193 56, doi: 10.3847/1538-4357/ad8226
Ibrahim, A. I., Markwardt, C. B., Swank, J. H., et al. 2004, ApJL, Scholz, P., Camilo, F., Sarkissian, J., et al. 2017, ApJ, 841, 126,
609, L21, doi: 10.1086/422636 doi: 10.3847/1538-4357/aa73de
 15

Timokhin, A. N. 2010, MNRAS, 408, L41, van Straten, W., & Bailes, M. 2011, PASA, 28, 1,
doi: 10.1111/j.1745-3933.2010.00924.x doi: 10.1071/AS10021
van Straten, W., Demorest, P., & Oslowski, S. 2012, Astronomical
Torne, P., Eatough, R. P., Karuppusamy, R., et al. 2015, MNRAS,
Research and Technology, 9, 237,
451, L50, doi: 10.1093/mnrasl/slv063
doi: 10.48550/arXiv.1205.6276
Torne, P., Desvignes, G., Eatough, R. P., et al. 2017, MNRAS, 465, Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020, Nature
242, doi: 10.1093/mnras/stw2757 Methods, 17, 261, doi: 10.1038/s41592-019-0686-2
Torne, P., Liu, K., Cognard, I., et al. 2020, The Astronomer’s Wang, N., Manchester, R. N., & Johnston, S. 2007, MNRAS, 377,
Telegram, 14001, 1 1383, doi: 10.1111/j.1365-2966.2007.11703.x