WorksPhysics and Astronomy
Periodic Modulation of Linear Polarization Fraction in Repeating FRB 20240114A
Abstract
Long-term polarization changes in repeating fast radio bursts (FRBs) can probe their emission geometry and propagation environment. We searched for periodic variation in the degree of linear polarization (DOL) of FRB~20240114A using the polarization catalog from the Five-hundred-meter Aperture Spherical radio Telescope. The sample contains 6,099 bursts detected on 95 observing days over days. The daily mean DOL series was analyzed using phase distance correlation and Lomb–Scargle periodograms over trial periods of 10–240 days. The strongest peaks occur at and days, respectively. The quoted ranges correspond to the period intervals above the global thresholds. Global false-alarm probabilities were estimated from random permutations of the daily values among the fixed observing dates, with the full period search repeated for each permutation. The resulting probabilities are and , corresponding to equivalent Gaussian significances of and , respectively. As a supplementary check, the same analysis using daily median DOL yielded similar peaks near 120 days. A descriptive sinusoidal fit to all 95 daily mean DOL measurements at a fixed period of 119.91 days has a semi-amplitude of 3.23%. Possible explanations include precession-driven changes in viewing geometry and variations in propagation conditions within a binary system.
Introduction
Fast radio bursts (FRBs) are bright, millisecond-duration radio transients (Lorimer et al. [22]). Some sources generate repeated bursts (Spitler et al. [34]). The second CHIME/FRB catalog contains 3,641 sources, including 83 known repeaters (Chime/Frb Collaboration et al. [6]). Repeated observations of these sources allow changes in their emission and propagation environments to be investigated within a single object.
Polarization provides constraints on both the emission geometry and the intervening magnetized plasma. The diverse polarization position-angle swings observed in FRB 20180301A favor a magnetospheric origin for its radio emission (Luo et al. [23]). The Faraday rotation measure (RM) depends on the electron density and the magnetic-field component along the propagation path. In FRB 20121102A, an RM of order rad m and its temporal changes indicate an extreme, evolving magnetized
Corresponding author: Yongfeng Huang
[email protected] environment [27]. The degree of linear polarization (DOL) is the fraction of the observed burst intensity that is linearly polarized. [10] found decreasing linear polarization toward lower frequencies in several repeaters and modeled this behavior through RM scatter. In this picture, radiation traveling along different paths undergoes different Faraday rotations, reducing the net linear polarization when the signals are combined. Tracking DOL over time can therefore help test changes in the emission geometry or propagation conditions.
FRB 20240114A was discovered by CHIME/FRB on 2024 January 14 [32]. An observation with the Robert C. Byrd Green Bank Telescope (GBT) detected 365 bursts in 1.38 hours, demonstrating the high activity of this source. Among 301 bursts with measured RMs in the GBT sample, approximately 81% had DOL above 90% [44]. The Five-hundred-meter Aperture Spherical radio Telescope (FAST) polarization catalog of FRB 20240114A provides measurements for more than 6,000 bright bursts from observations spanning approximately 16 months. It also reports an RM decline of approximately 200 rad m over 200 days, demonstrating long-term evolution of the magnetized environment [41]. This extensive polarization sample provides a basis for investigating the long-term behavior of DOL.
Periodic activity has been detected in FRB 20180916B, with a cycle of 16.35 days [5]. For FRB 20240114A, Zhou et al. [50] reported a 143.40 7.19-day period in burst activity. Li et al. (2026) reported a 112.91-day modulation in burst central frequency and compared FAST polarization distributions between phase intervals defined by that period. These findings motivate a direct search for periodicity in DOL. Such a search allows the preferred modulation timescale and its significance to be evaluated from the polarization measurements themselves.
In this study, we analyze the long-term periodic modulation in the DOL of FRB 20240114A using the FAST polarization catalog. The daily mean DOL series is analyzed using phase distance correlation and Lomb–Scargle periodograms. Global significance is assessed by randomly reassigning the daily values among the fixed observing dates. Daily medians are further used as a supplementary check. The structure of our paper is organized as follows. § 2 describes the data, and § 3 presents the methods. The results are given in § 4, followed by a brief discussion on the physical interpretations in § 5. § 6 compares our results with previous activity and central-frequency periods, and § 7 summarizes the study.
Data
FAST Polarization Catalog
The data analyzed in this work were collected from the publicly released FAST polarization catalog of FRB 20240114A [41]. FAST monitored the source from 2024 January 28 to 2025 May 30 in 97 observing sessions, totaling 57.99 hr. The observations used the central beam of the FAST 19-beam receiver [28] in tracking mode and covered 1.0–1.5 GHz. The campaign detected 17,356 bursts. Polarization measurements were released for a subset of these detections. We used the file FAST_FRB20240114A_pol_catalog_v5.csv, publicly available from the Science Data Bank repository.1 It contains 6,134 burst entries. The burst-level DOL measurements tabulated by Wang et al. [41] were used directly in this analysis.
Sample Selection and Daily DOL Series
All 6,134 entries in this catalog had finite values of topocentric arrival time, , and DOL. Applying to the tabulated values excluded 35 bursts. Of these, 27 had and eight had . The final sample contains 6,099 bursts from 95 observing days. The selected bursts span MJD 60337.238–60824.896, giving a temporal baseline of 487.66 days.
The number of detected bursts varies substantially among observing days. A periodicity search based on all individual bursts would therefore give greater weight to days with more detections. The selected bursts were instead grouped according to the integer part of their topocentric MJD. For each observing day, the daily DOL was calculated as the unweighted arithmetic mean of the selected bursts’ DOL measurements. The representative time was defined as the median arrival time of those bursts. This yielded a daily-mean DOL series containing 95 measurements. Measurement noise can cause DOL estimates to exceed 100% [41]. Cataloged values above 100% were retained without clipping.
Methods
In this study, we use two methods to analyze periodicity in the daily-mean DOL series, the Phase distance correlation (PDC) method and the generalized Lomb–Scargle (GLS) method. Here we first describe the two methods one by one.
Two Methods of Period Searches
Phase distance correlation (PDC) tests for a general dependence between DOL and the phase defined by a trial period. The periodogram is based on distance correlation [36, 51]. For two daily measurements, and , where is the representative time of observing day , and is its daily-mean DOL, the distance between their DOL values is
For a trial period , the wrapped separation between their observing times is
Here, lies in and has units of days. The corresponding phase distance is
The DOL-distance and phase-distance matrices were then double-centered by subtracting their row and column means and adding their grand means. Let and denote the centered entries corresponding to and , respectively. The PDC statistic was calculated as their normalized inner product:
A larger PDC value indicates a stronger dependence between DOL and phase at the corresponding trial period. Because the statistic compares pairwise distance structures, it does not require a specified modulation profile.
The generalized Lomb–Scargle periodogram (GLS) searches for sinusoidal variation in unevenly sampled data [21, 31, 48, 39]. In this generalized formulation, the constant offset is fitted together with the sinusoidal terms. It is referred to as Lomb–Scargle below. At each trial frequency , we fitted the daily DOL values with a floating-mean sinusoidal model:
where , , and were determined by least squares. All observing days were assigned equal weight. We defined the Lomb–Scargle power as
where is the residual sum of squares from the best-fitting sinusoidal model at trial frequency , and is that from the constant model. A larger Lomb–Scargle power indicates a greater reduction in the residual sum of squares relative to the constant model.
The same frequency grid was used for both methods. The trial periods ranged from 10 to 240 days. The grid was uniform in frequency, with a spacing of
where is the time span of the daily series and is the frequency oversampling factor. This gives a frequency spacing of approximately and 9348 trial frequencies. The peak period was determined separately for each method.
Global Significance Assessment
We assessed the global significance of each peak using a permutation test. In each permutation, the daily DOL values were randomly reassigned among the fixed observing dates. The complete 10–240 day period range was searched again, and the largest PDC statistic or Lomb–Scargle power was retained. Repeating this procedure produced the null distribution of the maximum statistic over the full search range. The resulting significance therefore accounts for the search over all 9348 trial frequencies [11, 1]. This permutation test assumes that the daily DOL values are exchangeable among the observing dates under the null hypothesis. Null models with time-correlated variability were not tested.
For each method, random permutations were performed. The 90%, 95%, and 99% global thresholds were obtained from the corresponding quantiles of the permutation maxima. If permutation maxima were at least as large as the observed maximum, the global false-alarm probability (FAP) was calculated as
where . Adding one to both the numerator and denominator prevents the estimated FAP from being reported as exactly zero [30]. The global FAPs were converted to equivalent Gaussian significances using one-sided tail probabilities.
Results
Periodicity in the Daily Mean DOL Series
Figure 1 shows the periodograms of the daily mean DOL series. Both methods show their strongest peaks near 120 days. The phase distance correlation periodogram [51] reaches its maximum value of 0.1452 at 119.91 days. This value exceeds the 99% global threshold of 0.1262. The Lomb–Scargle periodogram [48] reaches its maximum power of 0.1275 at 122.31 days, exceeding its 99% global threshold of 0.1108. The period ranges above the global thresholds are and days for phase distance correlation and Lomb–Scargle, respectively.
Figure 1. Periodograms of the daily mean DOL series of FRB 20240114A. Panels (a) and (b) show the phase distance correlation statistic and Lomb–Scargle power, respectively, for trial periods of 10–240 days. Red dotted vertical lines and filled circles mark the maxima at and days, respectively. The quoted ranges correspond to the period intervals above the global thresholds. Dashed, dotted, and solid horizontal lines indicate the 90%, 95%, and 99% global thresholds, respectively. These thresholds were obtained from random permutations of the daily DOL values among the fixed observing dates. The maximum statistic over the full period range was retained for each permutation. Both observed maxima exceed their corresponding 99% global thresholds.
Figure 2 compares the observed maxima with their global null distributions. Using the estimator of Phipson & Smyth [30], the global FAPs are for phase distance correlation and for Lomb–Scargle. These correspond to equivalent Gaussian significances of and , respectively. Both periodograms therefore show a significant peak near 120 days in the daily mean DOL series.
Figure 2. Global null distributions of the periodogram maxima for the daily mean DOL series of FRB 20240114A. Panels (a) and (b) correspond to phase distance correlation and Lomb–Scargle, respectively. For each method, random permutations were performed by shuffling the daily DOL values among the fixed observing dates. The full 10–240-day period range was searched in each permutation, and the maximum statistic was retained. Gray histograms show the distributions of these maxima. Dashed, dash-dotted, and dotted vertical lines mark the 90%, 95%, and 99% global thresholds, respectively. Solid lines mark the maxima from the observed data. The corresponding global false-alarm probabilities are and for panels (a) and (b), respectively. Both observed maxima exceed their corresponding 99% global thresholds.
Zhou et al. [50] reported a period of days in the burst activity of FRB 20240114A. Li et al. (2026) reported a 112.91-day periodic modulation in its burst central frequency. These studies probe different observables from the DOL examined here. A detailed comparison between these reported periods and our approximately 120-day DOL modulation is presented in § 6.
Supplementary Check Using Daily Median DOL
We have also calculated the median of the selected burst DOL values on each observing day. Here the median means there are equal numbers of bursts with DOL larger than and less than that particular DOL value, which is also a characteristic DOL for the bursts on that day. As a supplementary check, the same analysis was repeated using the daily median of DOL values. The search range and permutation procedure are the same as for the daily mean periodograms. Figure 3 shows the resulting periodograms. The strongest peaks occur at 119.91 days for phase distance correlation and 123.55 days for Lomb–Scargle. These two periods are well consistent with the daily mean DOL results. For phase distance correlation, the period range above the global threshold is days.
Figure 3. Periodograms of the daily median DOL series of FRB 20240114A, used as a supplementary check of the daily mean results. Panels (a) and (b) show the phase distance correlation statistic and Lomb–Scargle power, respectively. The maxima at and 123.55 days both exceed their corresponding 99% global thresholds. The quoted range corresponds to the period interval above the global threshold for phase distance correlation. The plotting conventions are the same as for the daily mean periodograms in Figure 1.
Figure 4 compares the observed maxima with their global null distributions. Both maxima exceed their corresponding 99% global thresholds. The global FAPs are for phase distance correlation and for Lomb–Scargle. These correspond to equivalent Gaussian significances of and , respectively. In short, we found that the daily median analysis also yields peaks near 120 days, consistent with the daily mean results.
Figure 4. Global null distributions of the periodogram maxima for the daily median DOL series of FRB 20240114A, used as a supplementary check of the daily mean results. The panel arrangement, permutation procedure, and line conventions are the same as for the daily mean analysis in Figure 2. The global false-alarm probabilities are and for panels (a) and (b), respectively. Both observed maxima exceed their corresponding 99% global thresholds.
Phase-folded DOL and Evolution across Consecutive Cycles
We have folded the timing series constructed in § 2.2 by referring to the 119.91-day period identified through the phase distance correlation method in § 4.1. The zero phase was set to the representative time of the first observing day. The phase of each measurement was calculated as
where is the representative time of observing day , days, and is the reference time. Figure 5 shows the phase-folded daily mean DOL measurements.
Figure 5. Daily mean DOL measurements of FRB 20240114A folded at a fixed period of 119.91 days. Light-blue circles show the 95 daily mean measurements. Dark-blue points show the averages of these daily values in 12 equal-width phase bins. Error bars give the standard error of the mean in each bin. The red curve shows the fixed-period sinusoidal fit to the unbinned daily mean measurements. The measurements, binned profile, and fitted curve are repeated over a second phase cycle for visual continuity.
The unbinned daily mean measurements were then fitted with the sinusoidal model
The coefficients were determined by ordinary least squares. Each observing day was assigned equal weight, and the period was held fixed. As shown in Figure 5, the fit to all 95 daily mean measurements has a semi-amplitude of 3.23%. The fitted DOL is highest near the middle of the phase cycle and lowest near the phase boundary. For display, the daily measurements were averaged in 12 equal-width phase bins. The error bars give the standard error of the mean in each bin.
Figure 6 shows the individual-burst and daily mean DOL measurements over the full observing span of 487.66 days. It corresponds to approximately 4.07 cycles of the adopted period. The same fitted sinusoid is shown throughout the observing span, without refitting individual cycles. The measurements show substantial scatter around the fitted curve.
Figure 6. DOL evolution of FRB 20240114A over the full observing span of 487.66 days. This span covers approximately 4.07 cycles of the adopted 119.91-day period. Light-purple circles represent individual bursts with S/N > 20, and dark-purple diamonds represent the daily mean DOL values. The gray dashed curve is a single sinusoid fitted to all daily means, with the period fixed at the phase distance correlation peak. The same fit is shown throughout the observing span without refitting individual cycles. Vertical dotted lines mark the cycle boundaries.
Possible Physical Interpretations
The -day DOL modulation could be due to various mechanisms, including neutron-star free precession, binary-orbital propagation effects, and other long-term dynamical processes. Here we present some discussion on these potential explanations.
Neutron-star Free Precession
Free precession has previously been proposed as the clock underlying periodic activity windows in repeating FRBs [16, 47] (Li & Zanazzi 2021). It could also be one possible origin of the approximately 120-day DOL modulation in FRB 20240114A. A viable interpretation of the present result requires two links. First, the precession must satisfy the observed timescale. Second, the changing viewing geometry must produce a variation in DOL.
For an approximately biaxial precessing component, let , , and denote the principal moments of its effective inertia tensor, with . The effective ellipticity is
Here the inertia difference represents the stress-supported deformation fixed to the stellar body. The centrifugal bulge that follows the instantaneous spin axis is excluded from this deformation [14]. In this approximation, the free-precession period is
where is the spin period of the compact central engine and is the wobble angle between the symmetry axis and the angular-momentum vector [14, 47, 9]. If the 119.91-day phase distance correlation peak is identified with , then
For spin periods of 1–10 s and small wobble angles, , the required effective ellipticity is approximately –. As an observational comparison, the preferred free-precession model of XTE J1810−197 gives a late-time ellipticity of approximately [9]. This value falls within the required range. Comparison with whole-star deformation models assumes that most of the stellar moment of inertia participates in the precession [8].
For spin periods of 1–10 ms, the required is approximately –. At small wobble angles, these values are the required effective ellipticities. A rotating-star comparison follows from the relaxed-crust calculations of Cutler et al. [8]. Consider a star with a mass of and a radius of 12 km. Its crust is assumed to be relaxed at the current spin rate. Elastic stresses supply the body-fixed deformation, and nearly the whole stellar moment of inertia participates in precession. Combining their crustal rigidity and rotational-deformation estimates gives . Under these assumptions, this estimate meets the 119.91-day requirement at a spin period of approximately 8.5 ms, with .
Thus, millisecond rotation can accommodate the observed timescale in this specific crust and coupling configuration. Faster spins or a smaller participating moment of inertia shorten the precession period within this scaling.
During free precession, the angular-momentum direction remains approximately fixed, while the body-fixed symmetry and magnetic axes change their orientation relative to the observer. The line of sight can therefore cross different parts of the radio-emission beam ([14, 47] (Li & Zanazzi 2021)). DOL modulation can arise if the sampled regions have different polarization properties. This can occur if the DOL distribution of individual bursts depends on the sampled beam region. It can also occur if the geometry changes the relative detection rates of burst populations with different DOL distributions.
Observations of FRB 20240114A show that its polarization state is complex or time-dependent. The FAST catalog contains a broad, non-uniform distribution of intrinsic polarization position angle (PA) values ([41]). Independent observations also show rapid and stochastic PA variations between and within bursts ([25]).
A change in the relative intensities of polarization modes that are unresolved within a Stokes measurement is one possible response to the changing viewing geometry. If two such modes are fully linearly polarized, strictly orthogonal, and mutually
incoherent, their combined linear polarization fraction is
where and are the mode intensities ([26]. Here , , are the Stokes parameters of the combined signal, and . Thus is the DOL expressed as a percentage in this idealized model. Taking , the weaker mode contributes a fraction
of the total intensity. The descriptive sinusoid fitted to all 95 daily mean measurements spans approximately 92.25%–98.71% in DOL. As an amplitude illustration, representing these endpoints with the idealized model gives weaker-mode fractions of approximately 3.88% and 0.64%, respectively. A small change in mode balance could therefore produce a DOL variation comparable to the range represented by the descriptive fit.
Orthogonal PA jumps have been observed in FRB 20201124A ([29]. For FRB 20240114A, the PA-jump distribution is broad, and jumps near are not more common than jumps of other magnitudes ([25]. A precession interpretation based on the two-mode response requires mode weights that vary systematically with viewing phase.
The period relation above assumes an approximately biaxial effective precessing component. A real neutron star contains an elastic crust coupled to superfluid components in the inner crust and core. Crustal strain, internal coupling, and vortex pinning can modify the effective deformation and precession dynamics ([14, 8, 12]. In this simple framework, long-period solutions require weak effective pinning or efficient vortex unpinning. Hydrodynamic forces generated during precession provide a possible unpinning mechanism ([20].
The month-scale damping inferred for XTE J1810−197 highlights the possible evolution of precession after excitation ([9]. For FRB 20240114A, the 487.66-day observing span covers approximately 4.07 cycles. The expected evolution of the viewing geometry and DOL amplitude depends on the excitation and damping history.
Phase-dependent changes in the distributions of PA, circular polarization, and burst morphology would provide tests of the geometric interpretation. Changes in the relative occurrence rates of distinct burst populations would provide another test. Simultaneous wideband polarimetry would help separate geometric changes from frequency-dependent propagation effects. A spin-period measurement would constrain the relation between effective ellipticity and wobble angle through the period equation above.
Free precession therefore provides a possible connection between the approximately 120-day timescale and the DOL modulation when the changing line of sight samples emission regions or burst populations with different polarization properties.
Binary-orbital Propagation Effects
Orbital motion in a binary system could make the propagation conditions of FRB 20240114A vary on a long timescale. Binary models have been proposed for periodic activity in repeating FRBs ([13]). A companion wind or circumstellar disk can also change the RM, scattering, and polarization of radio bursts along the orbit ([40, 46, 49]). The FAST catalog shows that the RM of FRB 20240114A evolves with time ([41]). These RM variations indicate a changing magneto-ionic environment along the line of sight.
If the DOL period of 119.91 days is associated with an orbital period, the Kepler’s third law gives the semimajor axis of the relative orbit:
Here is the total mass of the binary. The normalization is adopted for illustration. DOL modulation requires the orbital motion to change either the DOL of individual bursts or the relative proportions of bursts with different DOL values in the detected sample.
One direct propagation effect is external Faraday dispersion. A uniform RM rotates the polarization position angle but does not reduce the linear polarization fraction when the rotation is adequately resolved and corrected. Depolarization instead occurs if unresolved propagation paths undergo different amounts of Faraday rotation. For a screen with a Gaussian RM distribution, the Burn relation is ([3, 10])
where is wavelength, is orbital phase, and is the unresolved RM dispersion. In this propagation-only model, the intrinsic fraction is taken to be independent of orbital phase. An orbit could then modulate DOL if it changes the RM dispersion along the line of sight.
Wang et al. [41] tested frequency-dependent depolarization in 18 bright, broadband bursts from FRB 20240114A. They found no systematic decrease in DOL toward lower frequencies and obtained a 95% upper limit of for those bursts. Substituting this limit into the Burn relation gives a maximum reduction of approximately 0.42% at 1.0 GHz and 0.083% at 1.5 GHz, assuming an intrinsic linear polarization fraction of 100%. If the upper limit applies throughout the long-term cycle, these values also bound the maximum phase-to-phase DOL variation produced by the screen. They are well below the 6.46% peak-to-peak variation of the descriptive sinusoid fitted to the daily mean DOL series. Under this assumption, a single external screen obeying this relation would be insufficient to reproduce the full fitted variation. This comparison is conditional because the published limit constrains the selected bursts, rather than every phase of the long-term cycle.
Other effects within a binary environment need not follow the Burn relation. If different polarization modes experience different optical depths near a cyclotron resonance, changes in their relative transmission could alter the observed DOL (Li et al. 2026). Alternatively, phase-dependent absorption, scattering, or plasma lensing could change which bursts enter the detected sample ([7, 24]). This would change the daily DOL statistics only if the affected burst populations have different DOL distributions.
Broadband polarization measurements covering the full long-term phase could test these alternatives. External Faraday dispersion predicts stronger low-frequency depolarization when DOL is lower. Sample selection instead requires changes in the properties and relative numbers of the detected bursts. An orbital explanation of the DOL modulation therefore requires propagation or sample-selection effects that recur with orbital phase.
Other Possible Origins of the Long-term Modulation
A torque acting on the neutron star could cause the direction of its angular momentum to precess. If is the angular momentum and is the torque, . A component of perpendicular to changes the direction of . Torques from a fallback disk and anomalous electromagnetic torques have both been proposed as drivers of forced precession in repeating FRBs ([37, 33]). If the emission has angularly dependent polarization, the changing viewing geometry could alter the relative contributions of polarized emission components and thereby modulate DOL. The precession period would depend on the strength and geometry of the torque.
Precession associated with a tilted disk provides another possible clock. In the weak-field limit, the local Lense–Thirring precession angular frequency at radius around a spinning compact object is
where is the magnitude of the compact object’s angular momentum. The period of an extended disk also depends on its radial structure. Chen [4] proposed a model in which a tilted debris disk drives precession of the neutron star and inner disk. Precession of the polar funnel in a super-Eddington accreting binary has also been proposed to yield periods from days to years [35]. To produce the observed DOL variation, such precession would have to change either the polarized emission visible to the observer or the propagation conditions along the line of sight.
Another possibility is that the approximately 120-day timescale is the neutron star’s rotation period itself. Particle winds, flare-induced angular-momentum loss, and fallback disks have been investigated as ways to produce ultra-long-period magnetars [2]. Disk-wind calculations also permit periods of tens to hundreds of days, but only under particularly strong angular-momentum extraction (Xu et al. 2021). In this interpretation, rotation could modulate DOL if different rotational phases expose emission with different polarization properties.
The energy budget places a further constraint on a rotation-powered version of this model. Taking a stellar moment of inertia and , the rotational energy is
This is far below the reported isotropic-equivalent energies of individual bursts from this source, approximately – [42]. Isotropic-equivalent energy is not the true emitted energy if the radiation is beamed. Nevertheless, the large difference disfavors the star’s current rotational energy as the main burst-energy reservoir unless the radio emission is extremely narrowly beamed and the conversion of rotational energy into radio emission is sufficiently efficient. Magnetic energy could power the bursts while rotation sets the phase of the DOL variation.
These mechanisms assign the approximately 120-day timescale to different physical clocks: forced precession of the neutron star, precession associated with a disk, or stellar rotation. Each also requires the observed polarization to change with phase in order to produce the DOL variation.
Discussion
Zhou et al. [50] reported a 143.40 7.19-day period in the burst activity of FRB 20240114A. Their analysis used 111 FAST observations spanning 612.33 days and incorporated weights proportional to the burst rate. Our DOL analysis spans 487.66 days and assigns equal weight to each observing day. Over this 487.66-day interval, the 119.91-day DOL model completes 4.07 cycles, whereas a 143.40-day model completes 3.40 cycles. The relative phase of these two fixed-period models would therefore drift by approximately 0.67 cycles. If the activity period represents an orbital period and the propagation conditions repeat each orbit, an associated DOL variation should also repeat with orbital phase. Activity and DOL maxima could occur at different phases, but a constant phase offset would not explain the difference between the fitted periods. A shared orbital interpretation would therefore need to reproduce both datasets with a common period, while allowing their modulation profiles and phases to differ.
An activity cycle can reflect changes in burst production, visibility, or both. If only the burst rate varies while the polarization distribution remains unchanged, the expected daily mean DOL also remains unchanged. A geometric connection is possible if changing viewing directions both alter detectability and sample emission regions with different polarization fractions (Li & Zanazzi 2021). In a binary environment, the companion’s magnetized plasma could affect both visibility and polarization through absorption and polarization transfer [43]. The resulting DOL would depend on how propagation modifies the polarization of the detected emission and which burst populations remain detectable. The sign and phase of an activity–DOL association would depend on these physical responses. An activity maximum would therefore not necessarily coincide with a DOL maximum.
A separate question is whether DOL and burst central frequency vary on comparable timescales. Li et al. (2026) reported a 112.91-day modulation in individual Parkes burst central frequencies using the Lomb–Scargle method. Their phase-folding tests with 5–20 phase bins gave a period range of 104–123 days. For our comparison, a combined daily-median central-frequency series was constructed from Parkes [38], Effelsberg [19], and GBT measurements [44]. The series contains 110 daily values. All entries with both an observing time and a measured central frequency were included, without additional burst-level cuts. Bursts observed on the same day were pooled, and the median central frequency and median burst MJD were adopted for that day. An equal-day-weight, floating-mean Lomb–Scargle search was performed over periods from 2 days to one-third of the total time span , with a frequency spacing of . The strongest peak occurs at 112.90 days, which was adopted for the central-frequency sinusoidal fit.
Figure 7 compares this fit with the 119.91-day sinusoid fitted to all 95 daily mean DOL measurements. The DOL curve is shifted and scaled vertically to match the fitted mean and semi-amplitude of the central-frequency curve. This rescaling preserves its peak times. The periods and phases of the two curves were determined separately. During the earlier cycles, the DOL peaks precede the central-frequency peaks. The peak times become closer later in the observing span. The different fitted periods produce a gradual change in their relative peak times, so the apparent alignment varies across the observations.
Figure 7. Comparison of the DOL and central-frequency sinusoids of FRB 20240114A on the same MJD axis. Blue, coral-orange, and purple circles show individual-burst central frequencies measured with Parkes (Uttarkar et al. [38]), Effelsberg (Limaye et al. [19]), and GBT (Xie et al. [44]), respectively. Dark-blue diamonds show the daily medians of the combined three-telescope sample. The teal solid curve is a sinusoidal fit to this series at a period of 112.90 days. The red dashed curve shows the 119.91-day sinusoidal fit to the daily mean DOL series. The DOL curve is shifted and scaled vertically to match the fitted mean and semi-amplitude of the central-frequency curve. This rescaling preserves its peak times for comparison. Gray crosses mark the uGMRT receiver-band centers at the observing epochs (Kumar et al. [15]), which are shown for reference but are excluded from the central-frequency fit.
The local similarities between the DOL and central-frequency trends can be considered within the geometric and propagation scenarios discussed above. Different viewing directions may sample regions with different emission frequencies and linear polarization fractions. A changing viewing geometry could then affect both observables (Zanazzi & Lai [47]). Frequency-dependent propagation could reshape the spectrum and change the relative weights of frequency components with different polarization properties. This would also change the band-integrated DOL (Cordes et al. [7]; Feng et al. [10]). A change in the detected burst population could affect both quantities if the selected populations differ in their central-frequency and polarization distributions (Uttarkar et al. [38]).
Differences in telescope bandpasses, sensitivities, and observing cadence affect which bursts enter the combined central-frequency series. The DOL and central-frequency measurements also generally come from different bursts. Figure 7 therefore provides a qualitative comparison of their long-term timing. The local similarities motivate considering a possible connection, while the changing peak offsets and differences between the samples must be accounted for in any interpretation of a shared modulation.
Summary
We analyzed the FAST polarization catalog of FRB 20240114A for long-term variation in DOL. The sample contains 6,099 bursts detected on 95 observing days over 487.66 days. The daily mean DOL series was analyzed using phase distance correlation and Lomb–Scargle periodograms. The strongest peaks occur at and days, respectively. The quoted ranges correspond to the period intervals above the global thresholds. Their global false-alarm probabilities, estimated from random permutations of the daily values, are and . These correspond to equivalent Gaussian significances of and , respectively. As a supplementary check, the same analysis was repeated using daily median DOL, yielding similar peaks near 120 days. Folding the daily mean series at 119.91 days shows a phase-dependent variation. A descriptive sinusoidal fit to all 95 daily means has a semi-amplitude of 3.23%. The time-aligned DOL and central-frequency fits show local similarities, with peak offsets that change across the observing span. Geometric and environmental modulation should be involved in the approximately 120-day DOL variation.
Acknowledgments
This work is supported by the National Natural Science Foundation of China (Grant Nos. 12233002, 12273113, and 12403046), by the China Postdoctoral Science Foundation (Grant No. 2025M783225), and by the Specialized Research Fund for State Key Laboratory of Radio Astronomy and Technology (No. SDQZ-KF-A2610).
References
- [1]Baluev, R. V. 2008, MNRAS, 385, 1279, doi: 10.1111/j.1365-2966.2008.12689.xDOI
- [2]Beniamini, P., Wadiasingh, Z., & Metzger, B. D. 2020, MNRAS, 496, 3390, doi: 10.1093/mnras/staa1783DOI
- [3]Burn, B. J. 1966, MNRAS, 133, 67, doi: 10.1093/mnras/133.1.67DOI
- [4]Chen, W.-C. 2020, PASJ, 72, L8, doi: 10.1093/pasj/psaa060DOI
- [5]Chime/Frb Collaboration, Amiri, M., Andersen, B. C., et al. 2020, Nature, 582, 351, doi: 10.1038/s41586-020-2398-2
- [6]Chime/Frb Collaboration, Abbott, T., Andersen, B. C., et al. 2026, ApJS, 283, 34, doi: 10.3847/1538-4365/ae3828
- [7]Cordes, J. M., Wasserman, I., Hessels, J. W. T., et al. 2017, ApJ, 842, 35, doi: 10.3847/1538-4357/aa74daDOI
- [8]Cutler, C., Ushomirsky, G., & Link, B. 2003, ApJ, 588, 975, doi: 10.1086/368308DOI
- [9]Desvignes, G., Weltrevede, P., Gao, Y., et al. 2024, Nature Astronomy, 8, 617, doi: 10.1038/s41550-024-02226-7DOI
- [10]Feng, Y., Li, D., Yang, Y.-P., et al. 2022, Science, 375, 1266, doi: 10.1126/science.abl7759DOI
- [11]Frescura, F. A. M., Engelbrecht, C. A., & Frank, B. S. 2008, MNRAS, 388, 1693, doi: 10.1111/j.1365-2966.2008.13499.xDOI
- [12]Glampedakis, K., & Jones, D. I. 2010, MNRAS, 405, L6, doi: 10.1111/j.1745-3933.2010.00846.xDOI
- [13]Ioka, K., & Zhang, B. 2020, ApJL, 893, L26, doi: 10.3847/2041-8213/ab83fbDOI
- [14]Jones, D. I., & Andersson, N. 2001, MNRAS, 324, 811, doi: 10.1046/j.1365-8711.2001.04251.xDOI
- [15]Kumar, A., Maan, Y., & Bhusare, Y. 2024, ApJ, 977, 177, doi: 10.3847/1538-4357/ad84deDOI
- [16]Levin, Y., Beloborodov, A. M., & Bransgrove, A. 2020, ApJL, 895, L30, doi: 10.3847/2041-8213/ab8c4cDOI
- [19]Limaye, P., Spitler, L. G., Manaswini, N., et al. 2026, A&A, 709, A47, doi: 10.1051/0004-6361/202557626DOI
- [20]Link, B., & Cutler, C. 2002, MNRAS, 336, 211, doi: 10.1046/j.1365-8711.2002.05726.xDOI
- [21]Lomb, N. R. 1976, Ap&SS, 39, 447, doi: 10.1007/BF00648343DOI
- [22]Lorimer, D. R., Bailes, M., McLaughlin, M. A., Narkevic, D. J., & Crawford, F. 2007, Science, 318, 777, doi: 10.1126/science.1147532DOI
- [23]Luo, R., Wang, B. J., Men, Y. P., et al. 2020, Nature, 586, 693, doi: 10.1038/s41586-020-2827-2DOI
- [24]Main, R., Yang, L.-S., Chan, V., et al. 2018, Nature, 557, 522, doi: 10.1038/s41586-018-0133-zDOI
- [25]Manaswini, N., Hewitt, D. M., Spitler, L. G., et al. 2026, arXiv e-prints, arXiv:2604.28012, doi: 10.48550/arXiv.2604.28012DOI
- [26]McKinnon, M. M., & Stinebring, D. R. 1998, ApJ, 502, 883, doi: 10.1086/305924DOI
- [27]Michilli, D., Seymour, A., Hessels, J. W. T., et al. 2018, Nature, 553, 182, doi: 10.1038/nature25149DOI
- [28]Nan, R., Li, D., Jin, C., et al. 2011, International Journal of Modern Physics D, 20, 989, doi: 10.1142/S0218271811019335DOI
- [29]Niu, J. R., Wang, W. Y., Jiang, J. C., et al. 2024, ApJL, 972, L20, doi: 10.3847/2041-8213/ad7023DOI
- [30]Phipson, B., & Smyth, G. K. 2010, Statistical Applications in Genetics and Molecular Biology, 9, Article 39, doi: 10.2202/1544-6115.1585DOI
- [31]Scargle, J. D. 1982, ApJ, 263, 835, doi: 10.1086/160554DOI
- [32]Shin, K., Curtin, A., Fine, M., et al. 2026, ApJ, 997, 334, doi: 10.3847/1538-4357/ae2fc4DOI
- [33]Sob’yanin, D. N. 2020, MNRAS, 497, 1001, doi: 10.1093/mnras/staa1976DOI
- [34]Spitler, L. G., Scholz, P., Hessels, J. W. T., et al. 2016, Nature, 531, 202, doi: 10.1038/nature17168DOI
- [35]Sridhar, N., Metzger, B. D., Beniamini, P., et al. 2021, ApJ, 917, 13, doi: 10.3847/1538-4357/ac0140DOI
- [36]Székely, G. J., Rizzo, M. L., & Bakirov, N. K. 2007, The Annals of Statistics, 35, 2769, doi: 10.1214/009053607000000505DOI
- [37]Tong, H., Wang, W., & Wang, H.-G. 2020, Research in Astronomy and Astrophysics, 20, 142, doi: 10.1088/1674-4527/20/9/142DOI
- [38]Uttarkar, P. A., Shannon, R. M., Gourdji, K., et al. 2026, arXiv e-prints, arXiv:2602.16409, doi: 10.48550/arXiv.2602.16409DOI
- [39]VanderPlas, J. T. 2018, ApJS, 236, 16, doi: 10.3847/1538-4365/aab766DOI
- [40]Wang, F. Y., Zhang, G. Q., Dai, Z. G., & Cheng, K. S. 2022, Nature Communications, 13, 4382, doi: 10.1038/s41467-022-31923-yDOI
- [41]Wang, T.-C., Zhang, J.-S., Liu, X.-H., et al. 2026, ApJS, 284, 77, doi: 10.3847/1538-4365/ae632fDOI
- [42]Wang, X.-W., Yan, Z., Shen, Z.-Q., et al. 2025, ApJ, 992, 185, doi: 10.3847/1538-4357/adfeceDOI
- [43]Xia, Z.-Y., Yang, Y.-P., Li, Q.-C., et al. 2023, ApJ, 957, 1, doi: 10.3847/1538-4357/acf5eb
- [44]Xie, J.-T., Feng, Y., Li, D., et al. 2025, ApJS, 278, 49, doi: 10.3847/1538-4365/add2ebDOI
- [46]Yang, Y.-P., Xu, S., & Zhang, B. 2023, MNRAS, 520, 2039, doi: 10.1093/mnras/stad168DOI
- [47]Zanazzi, J. J., & Lai, D. 2020, ApJL, 892, L15, doi: 10.3847/2041-8213/ab7cddDOI
- [48]Zechmeister, M., & Kürster, M. 2009, A&A, 496, 577, doi: 10.1051/0004-6361:200811296DOI
- [49]Zhao, Z. Y., Zhang, G. Q., Wang, F. Y., & Dai, Z. G. 2023, ApJ, 942, 102, doi: 10.3847/1538-4357/aca66bDOI
- [50]Zhou, D., Wang, P., Fang, J., et al. 2026, Science China Physics, Mechanics, and Astronomy, 69, 249512, doi: 10.1007/s11433-025-2866-2DOI
- [51]Zucker, S. 2018, MNRAS, 474, L86, doi: 10.1093/mnrasl/slx198DOI






