Searching the SN 1987A SETI Ellipsoid with TESS

The TESS mission succeeded Kepler in the search for exoplanets, focusing on stars 100× brighter and 10× closer to Earth—on average—than the earlier mission. As a result, TESS's coverage is mostly constrained to stars that lie within 200–300 lt-yr of Earth as opposed to Kepler's average of ∼3000 lt-yr (Jenkins et al. 2016). Moreover, while the Kepler prime mission field of view surveyed a fixed 100 deg² area near the constellation Cygnus, the TESS field of view shifts every 27.4 days, allowing it to complete an all-sky survey in 2 yr (Koch et al. 2010; Ricker et al. 2015).

One 27 day continuous coverage of TESS's 24° ×96° field of view is known as an observing sector; 1 yr of coverage–hereafter referred to as a cycle–contains 13 sectors and observes most of one hemisphere (Ricker et al. 2015). In this work, we use the 2 minute cadence data from the first three cycles, completed during 2018 through 2021. As shown in Figure 1, Cycles 1 and 3, which ran during 2018–2019 and 2020–2021, respectively, provided coverage of the southern hemisphere sky, whereas Cycle 2, active during 2019–2020, surveyed the northern hemisphere.

Zoom In Zoom Out Reset image size

During these three cycles, 329,176 unique stars were observed, but most of these sources were only present in one sector per cycle. This is apparent in Figure 1. We are mainly interested in targets that appear in multiple sectors, and therefore have long spans of continuous photometric coverage, since this helps us account for the uncertainties in the expected arrival time of a presumed signal, and provides multiple month-long, high-cadence light curves to use for anomaly detection. Consequently, we focused our search on stars that were present in at least 11 sectors in a given cycle, such that there are between 300 and 356 days worth of data for each target. This criterion shrinks our sample by 96%, resulting in 14,614 stars to select SETI Ellipsoid targets from, and guarantees that we search the entire CVZ, as well as nearby fields with very good coverage. The distribution of the reduced data sample on the celestial plane is shown in bright orange in Figure 1; we will refer to this sample simply as the "CVZ" in the following text.

Since Cycles 1 and 3 surveyed the same hemisphere, 48,744 southern-sky sources were observed during both time periods, 2 yr apart, 2493 of which were in the CVZ. Note that although roughly the same area was surveyed, only about 15% of targets overlap. This is because we are using the TESS 2 minute targets. These targets were selected through proposals, which saw a 500% increase in the first extension of the mission (Cycles 3 and 4). ⁶ The Full Frame Image (FFI) data from TESS include far more targets than those considered in this work. We further discuss FFIs in Section 5. As will be described in Section 3, timing is essential for this technique; thus, a source that might be trivial in 2019 could be a target of interest in 2021. Accordingly, we decided to independently consider the 2493 overlapping targets in both cycles, increasing our database of CVZ targets to 17,107 rows.

The SETI Ellipsoid technique of target selection can only be applied when the distances to the surveyed stars are precise and accurate. Gaia's EDR3 provides a significant improvement in parallax measurement precision over the previous Data Release 2 (DR2), leading to median parallax uncertainties of 0.02–0.03 mas for bright stars (G ≤14 mag), a 40% decrease from Gaia DR2, which had median uncertainties of 0.04 mas for stars of the same magnitude (Lindegren et al. 2018, 2021; Gaia Collaboration et al. 2021). With this unprecedented parallax precision, Bailer-Jones et al. (2021) estimated geometric and photogeometric distances for 1.5 billion stars. As shown by Davenport et al. (2022), the catalog of Gaia EDR3 geometric-probabilistic distances from Bailer-Jones et al. (2021) can be effectively used to make precise estimates for the expected arrival time of a presupposed synchronized signal using the SETI Ellipsoid method for a well-documented source event. Here, we use the same procedure of target selection but for TESS objects, focusing on the 14,614 CVZ targets chosen in Section 2.1. We did not limit these targets by any particular parameter (e.g., stellar type, binarity, or habitability). A system could be used for the SETI Ellipsoid technique regardless of whether it is habitable—in fact, it might be preferable for the ETA to place the signaling technology in an uninhabited system.

To assign each target its corresponding Gaia ERD3 probabilistic distances, we used the crossmatch service provided by CDS, Strasbourg (Pineau et al. 2020).

We merged our TESS data with the Gaia EDR3 distances database using the targets' celestial coordinates within a radius of 1''. During the crossmatching process of the CVZ sample, 725 sources were dropped, likely due to a disagreement greater than the 1'' threshold between the TESS and Gaia R.A. and decl. values. We did not attempt to recover these missing targets, which correspond to 4% of the initial CVZ list. Thus, the final data sample contained 16,382 rows of TESS CVZ sources with Gaia EDR3 distance estimates.

As noted in Section 1, the SETI Ellipsoid is a method of target selection that works under the assumption that an extraterrestrial civilization or agent would actively try to draw attention to themselves by broadcasting a signal soon after first witnessing a notable astronomical event (Makovetskii 1977; Lemarchand 1994).

The speed of light, c, constrains our targets of interest to a particular region of space at a particular point in time. We consider stars located at the precise distances from Earth that allow for an electromagnetic signal—sent in coordination with the astronomical event—to reach Earth at any given time, with the intent of aligning our monitoring efforts accordingly.

The region of interest is the Ellipsoid whose foci are at the source event and at Earth. The Ellipsoid grows with time as new stars observe the event, and is defined by

$$\begin{eqnarray}&&{d}_{1}+{d}_{2}=2\,{\mathscr{A}}=2\,{\mathscr{C}}+T,\end{eqnarray} \tag{ 1 }$$

where (d₁) is the distance from the Earth to a target laying on the Ellipsoid, (d₂) is the distance from the event to the target, ${\mathscr{A}}$ is the semimajor axis of the Ellipsoid, ${\mathscr{C}}$ is the distance from each foci to the center of the Ellipsoid, and T = c t is the time since Earth observed the event times the speed of light. Therefore, given a synchronizing event with a known distance, and a sample of field stars with distances, we can compute their Ellipsoid crossing times as a function of time using Equation (1).

Ideal events for the Ellipsoid technique are fixed in space and time, relatively uncommon in our Galactic neighborhood, and bright enough to be used as a synchronizing beacon on a galactic scale (Lemarchand 1994).

Although other events are worthy of examination as well (e.g., galactic novae, as explored by Davenport et al. 2022), supernova outbursts are particularly good source events to place at one of the foci within the Ellipsoid framework, and SN 1987A is an ideal Schelling point for our purposes. First, its distance to Earth is very well-documented with a low uncertainty. We used the distance estimate computed by Panagia (1999) of 51.4 ± 1.2 kpc, where the 1.2 kpc uncertainty corresponds to a 2.3% error. Second, SN 1987A is the closest supernova to Earth since SN 1604. Given that it is relatively nearby in spacetime, SN 1987A has a thin associated Ellipsoid that closely surrounds Earth, ensuring that many targets of interest will lie within relatively short distances and therefore have low distance uncertainties, resulting in low timing uncertainties for the arrival time of a synchronized signal. Lastly, the location of SN 1987A in the Large Magellanic Cloud is optimal for this work due to the cloud's proximity to the south ecliptic pole, which is the area surveyed by TESS's CVZ in Cycles 1 and 3. The polar location, along with the current thin shape of the Ellipsoid, allows for the intersection of a significant portion of SN 1987A's Ellipsoid with the southern TESS CVZ cone, making it highly likely to find several targets of interest within our previously selected data sample for two of the three cycles considered. In contrast, we expect little to no Ellipsoid candidates on the northern hemisphere. This can be appreciated in Figure 2.

Zoom In Zoom Out Reset image size

The list of TESS CVZ sources found in Section 2.1 contains 14,614 individual objects and 17,107 rows of data due to sources that were observed in two different cycles. After the crossmatch with Gaia distances performed in Section 2.3, we know both where these targets are located in space, and when they were observed by TESS. Solving Equation (1) for a desired time allows us to calculate the distance of our CVZ targets to the Ellipsoid at that time.

The algorithm we developed uses the start and end dates of observation for each target's first and last sectors, respectively, obtained from NASA's TESS observation times site. ⁷ The mid-time of observation is then calculated, and since most of the targets in our sample were observed in all TESS sectors, this mid-time is the same for every 13-sector target in each cycle as shown in Table 1.

We then use the right-hand side of Equation (1) to find the length of the semimajor axis of the Ellipsoid, ${\mathscr{A}}$, at these times. According to Equation (1), a target is on the Ellipsoid if

$$\begin{eqnarray}&&{d}_{1}+{d}_{2}-2\,{\mathscr{A}}=0.\end{eqnarray} \tag{ 2 }$$

However, we have decided to select targets that are within reasonable distance to the Ellipsoid as opposed to exactly on it, given the uncertainties involved in the process of finding a target strictly on an infinitely thin surface. Other uncertainties might also play a role, such as the error in the distances to the sources, explored in Section 3.4, or the potential delay in the broadcast of a burst from an ETA, discussed in Section 5. We allowed for some leeway in the Ellipsoid distance by creating a tolerance parameter, etol, a positive value in light years, which determines the distance-to-the-Ellipsoid threshold for target selection. This adds thickness to the Ellipsoid, essentially turning it into an Ellipsoidal shell defined by

$$\begin{eqnarray}&&| {d}_{1}+{d}_{2}-2\,{\mathscr{A}}| \lt {\mathtt{etol}}.\end{eqnarray} \tag{ 3 }$$

The same steps were followed by Davenport et al. (2022), where the selected tolerance was etol = 0.1 lt-yr, to make use of Gaia's highest timing precisions. However, as Davenport et al. state, the value of this threshold can and should be picked based on the specific monitoring campaign's properties. For this work, we have selected etol = 0.5 lt-yr to match the approximate duration of one TESS cycle (1 yr). Choosing targets 0.5 lt-yr away from the mid-cycle Ellipsoid helps account for targets that would have been on it at the beginning of said cycle—0.5 yr before mid-time—or at the end—0.5 yr after mid-time. This is not a perfectly precise threshold since the Ellipsoid grows slightly slower than the speed of light near the foci, resulting in a decreasing eccentricity over time such that the Ellipsoid becomes a spheroid at t = ∞ . This widening effect is illustrated in Figure 5 of Davenport et al. (2022). Due to this Ellipsoid rate of growth differential, a 0.5 lt-yr tolerance potentially includes targets near Earth or near SN 1987A that did not intersect the Ellipsoid exactly during the TESS's observing cycle. Note that this widening effect occurs over centuries, and is likely negligible over timescales of individual missions.

After running our algorithm for the 17,107 row list of CVZ sources, we obtained 12 targets of interest laying within 0.5 lt-yr of the SN 1987A Ellipsoid as of 2019 January 20 mid-time of Cycle 1 and 24 targets of interest as of 2020 December 28 mid-time of Cycle 3, resulting in 36 selected TESS objects. As expected per the discussion in Section 3.2, no targets of interest were found in the northern hemisphere, surveyed during Cycle 2. Accordingly, we did not extend our search to TESS's Cycle 4, which also observed this region of the sky.

The selection of the 36 targets of interest from our list of 17,107 CVZ sources described in Section 3.3 was computed considering only their 50% probability geometric distances, known as r_med, obtained from Bailer-Jones et al. (2021), as well as the reported distance to SN 1987A from Panagia (1999), and the celestial coordinates from the TESS 2 minute data, not accounting for errors for any of these quantities. However, these errors directly impact the uncertainties in the intersection time between the Ellipsoid and any target of interest, which, consequently, establishes a window for the expected time of arrival of a synchronized signal.

We investigated how much the error of each of these quantities affects the distance of our targets to the Ellipsoid to (a) assess which one of these uncertainties will have the greatest impact on our calculations, (b) to eliminate targets with very large uncertainties, i.e., uncertainties larger than TESS's yearlong coverage, and (c) to find the previously mentioned "window of interest" to monitor, which corresponds to the range of time where we expect to see an anomaly in the light curves of our targets of interest coordinated with their first detection of SN 1987A. To accomplish this, we incorporated the R.A., decl., SN 1987A distance, and target distance uncertainties into our original selection algorithm by using a Monte Carlo simulation. We recalculated the distance to the Ellipsoid for each selected target by drawing 1000 random Gaussian-distributed samples using the standard deviations for the parameters mentioned above. For the target distances, we used Bailer-Jones et al.'s (2021) 16% and 84% distance probabilities, r_lo and r_hi , respectively, such that σ_r = (r_hi − r_lo )/2. For the SN 1987A distance, we used the uncertainty from Panagia (1999), ${\sigma }_{\mathrm{SN}}$=1.2 kpc. For the celestial coordinates, R.A. and decl., we used the errors from Bailer-Jones et al. (2021) and Astropy (Astropy Collaboration et al. 2022) to assign celestial coordinates to each draw using the angle standard deviation ⁸ we defined as ${\sigma }_{\angle }={({\sigma }_{\mathrm{RA}}^{2}+{\sigma }_{\mathrm{Dec}}^{2})}^{1/2}$. We first ran the Ellipsoid calculation including the uncertainty for each parameter in isolation in order to determine which error has a greater effect on the final results. We found that the average error that propagates from the uncertainty of the celestial coordinates to the distance-to-the-Ellipsoid uncertainty corresponds to 0.073 lt-yr, while the uncertainty in the distance to SN 1987A only contributed 0.006 lt-yr to the mean error. As expected, the dominant contribution to the distance-to-the-Ellipsoid uncertainties propagates from the Bailer-Jones et al. (2021) error of the distance to the targets, which works out to 0.193 lt-yr on average. The final simulation did, however, consider the error in all the evaluated parameters.

A histogram illustrating the results of the Monte Carlo simulation for our Ellipsoid targets per cycle is shown in Figure 3. We used the standard deviation of these results to assign a distance-to-the-Ellipsoid uncertainty to each target in light years, which is equivalent to finding their Ellipsoid timing uncertainty, σ_et, in years. This uncertainty will indicate the length of the window of interest, defined earlier as the best time to look for abnormalities in the light curves of a selected target.

Zoom In Zoom Out Reset image size

All targets in Cycle 1 had timing uncertainties below 0.3 yr due to their proximity to Earth. However, some targets in Cycle 3 happened to lay farther along the Ellipsoid toward SN 1987A. The most distant targets in Cycle 3 had uncertainties of up to 13 yr, and with TESS's continuous coverage lasting about 1 yr, we decided to set an Ellipsoid timing uncertainty threshold and only consider targets below σ_et = 0.5 yr, which matches the tolerance parameter we established earlier and guarantees that most of the window of interest for any given target will occur during TESS's coverage, even when considering a distance-to-the-Ellipsoid value close to 0.5 lt-yr. Only four targets were dropped using this criterion, reducing the number of Ellipsoid targets to 32. The histogram in Figure 3 shows the distribution of distances to the Ellipsoid for the final 32 targets. In Table 2 we list all 32 of our final TESS targets, their coordinates, and the estimated SN 1987A Ellipsoid crossing times.

Table 2. All 32 Targets of Interest Found within 0.5 lt-yr of the SN 1987A Ellipsoid at the Mid-time of Their TESS Observing Cycle, Arranged Chronologically by Crossing Time with the Ellipsoid

TICID	R.A. (deg)	Decl. (deg)	d 1 (pc)	t cross (yr)
279055252	102.35 ± 0.02	−57.57 ± 0.02	${313.32}_{-1.81}^{+1.47}$	2018.58 ± 0.18
382187278	81.26 ± 0.01	−57.24 ± 0.01	${431.33}_{-2.13}^{+1.98}$	2018.60 ± 0.16
349374407	110.09 ± 0.01	−61.65 ± 0.01	${368.79}_{-1.72}^{+1.66}$	2018.78 ± 0.15
350434130	85.04 ± 0.01	−55.99 ± 0.01	${359.73}_{-1.69}^{+1.33}$	2018.78 ± 0.14
38711529	67.05 ± 0.01	−60.36 ± 0.01	${493.05}_{-1.93}^{+2.64}$	2018.89 ± 0.15
350580827	86.89 ± 0.02	−55.04 ± 0.02	${313.21}_{-1.91}^{+2.22}$	2018.98 ± 0.23
306672432	120.59 ± 0.02	−67.79 ± 0.02	${360.68}_{-2.74}^{+2.62}$	2018.99 ± 0.25
260368525	95.24 ± 0.02	−58.35 ± 0.02	${445.65}_{-3.35}^{+4.58}$	2019.03 ± 0.32
382067226	79.61 ± 0.01	−55.79 ± 0.01	${346.06}_{-1.18}^{+1.09}$	2019.08 ± 0.11
279614617	106.30 ± 0.01	−57.57 ± 0.01	${277.66}_{-0.88}^{+0.97}$	2019.14 ± 0.11
350711056	88.26 ± 0.01	−57.73 ± 0.01	${469.82}_{-3.03}^{+2.62}$	2019.24 ± 0.22
381976956	77.55 ± 0.01	−57.59 ± 0.01	${446.51}_{-2.93}^{+2.20}$	2019.28 ± 0.19
382028301	78.59 ± 0.01	−56.03 ± 0.01	${371.06}_{-1.47}^{+1.45}$	2020.64 ± 0.14
260416034	95.65 ± 0.04	−56.81 ± 0.04	${368.27}_{-4.47}^{+3.91}$	2020.65 ± 0.44
272316124	118.76 ± 0.01	−74.05 ± 0.01	${485.10}_{-3.10}^{+3.32}$	2020.69 ± 0.24
382028819	78.50 ± 0.02	−57.58 ± 0.02	${472.45}_{-3.24}^{+3.18}$	2020.75 ± 0.24
349681799	112.53 ± 0.02	−61.23 ± 0.02	${333.39}_{-1.77}^{+2.19}$	2020.83 ± 0.21
220404759	69.86 ± 0.01	−59.91 ± 0.01	${550.05}_{-3.26}^{+2.96}$	2020.87 ± 0.21
349571548	111.68 ± 0.01	−62.16 ± 0.01	${382.82}_{-1.58}^{+1.70}$	2020.89 ± 0.15
260266066	93.75 ± 0.01	−59.73 ± 0.01	${620.03}_{-5.94}^{+7.12}$	2020.95 ± 0.38
348900215	106.21 ± 0.02	−60.52 ± 0.02	${414.98}_{-3.23}^{+3.46}$	2020.96 ± 0.29
350715741	88.41 ± 0.02	−57.09 ± 0.02	${444.37}_{-2.65}^{+2.47}$	2021.01 ± 0.22
453100169	116.22 ± 0.02	−68.76 ± 0.02	${516.53}_{-4.82}^{+5.34}$	2021.10 ± 0.35
260542038	96.99 ± 0.01	−58.14 ± 0.01	${437.02}_{-2.25}^{+2.00}$	2021.17 ± 0.18
300869262	117.18 ± 0.01	−70.14 ± 0.01	${520.21}_{-2.97}^{+2.90}$	2021.27 ± 0.21
150431816	97.36 ± 0.01	−61.04 ± 0.01	${686.37}_{-4.98}^{+4.84}$	2021.29 ± 0.26
382159540	81.02 ± 0.01	−56.50 ± 0.01	${415.38}_{-1.74}^{+2.31}$	2021.32 ± 0.18
278731442	99.56 ± 0.01	−55.79 ± 0.01	${297.63}_{-1.11}^{+0.91}$	2021.37 ± 0.13
278684349	99.23 ± 0.01	−57.01 ± 0.01	${350.63}_{-1.33}^{+1.32}$	2021.40 ± 0.14
220424430	72.28 ± 0.01	−57.46 ± 0.01	${415.09}_{-2.21}^{+2.05}$	2021.42 ± 0.19
349972412	114.47 ± 0.01	−62.08 ± 0.01	${336.72}_{-1.25}^{+1.39}$	2021.43 ± 0.15
350954265	90.81 ± 0.01	−55.48 ± 0.01	${344.91}_{-1.00}^{+1.02}$	2021.50 ± 0.11

Note. The twelve targets on the top section of the table were on the Ellipsoid in Cycle 1. The twenty targets on the bottom section of the table belong to Cycle 3. The TICID column shows the TESS ID for each object. R.A. and decl. values and uncertainties were obtained from TESS and Gaia data. The distances, d₁, and their uncertainties were retrieved from Bailer-Jones et al. (2021). The Ellipsoid crossing times, t_cross, and their uncertainties were computed using the methods described in Section 3. All values were rounded to the hundredth decimal place for visual simplicity and consistency.

Download table as:  ASCIITypeset image

While using Equation (3) with etol = 0.5 lt-yr is a computationally efficient method of target selection and finding the value of σ_et is equivalent to calculating the length of the window of interest, we still need to find exactly when this window occurs for each target. To do this, we set out to compute the "Ellipsoid crossing times" for each selected source, that is, the time when the Ellipsoid and a given target are expected to meet.

To do this, we used the following rearrangement of Equation (2),

$$\begin{eqnarray}&&{d}_{1}+{d}_{2}-2\,{\mathscr{A}}={E}_{\mathrm{dist}},\end{eqnarray} \tag{ 4 }$$

where the left-hand side is now equal to E_dist, the distance of the target to the Ellipsoid at a given time in parsecs. We used Gaia's r_med distance as our d₁, and the date of each candidate's first TESS observation to get ${\mathscr{A}}$. We then calculated the distance to the Ellipsoid 130 subsequent times in steps of 2.8 days for a total of 364 days, the approximate duration of one TESS cycle. This process is illustrated in Figure 5, where we find the Ellipsoid crossing time for each of the 32 targets of interest. As previously mentioned, the uncertainty of this value is equal to σ_et, calculated in Section 3.

Additionally, in Section 3.3 we found that depending on the location of the targets in space, while their crossing times might fall within the etol = 0.5 lt-yr tolerance, the actual Ellipsoid crossing times may not be within the TESS observing cycle. As we can see in Figure 5, this was not an issue and all the targets did intersect the Ellipsoid within their respective TESS cycles. Furthermore, the slopes of the lines (i.e., the Ellipsoid distance over time) are ∼0.999c, showing that, as expected, any apparent superluminal projection effects are negligible at the scale of ∼1 yr.

Using the lightkurve package (Lightkurve Collaboration et al. 2018), we obtained all the 2 minute cadence Pre-search Data Conditioning Simple Aperture Photometry (PDCSAP) TESS light curves for each candidate. PDCSAP fluxes are produced in the TESS science pipeline by identifying and removing long-term systematic trends from the Simple Aperture Photometry flux, which is generated by adding all pixel values in a predetermined aperture (Morris et al. 2017). After retrieving the PDCSAP data, we normalized the flux in each sector, subsequently stitching all the light curves together for each candidate. This allowed us to properly characterize the yearlong photometric behavior of each target of interest to establish a benchmark from which to search for deviations or anomalies. We then performed additional flux-outlier removal using the built-in sigma-clipping function in the lightkurve package.

The previously computed Ellipsoid crossing times allow us to identify the range of time where we would expect to see a technosignature appear if an extraterrestrial civilization had sent a signal immediately after observing SN 1987A. Consequently, we looked for anomalies in all the retrieved light curves, prioritizing the window of interest given by the uncertainties in the crossing times discussed in Section 3.4.

We first visually inspected the light curves of all 32 candidates, from which we reported no unexplained anomalies. As in Nilipour et al. (2023), we looked for changes in the stellar variability coincident with the Ellipsoid crossing times as a possible sign of anomalous behavior. However, no changes in the underlying stellar variability properties were apparent in the sample.

Changes that would have warranted further inspection include unusual brightening or dimming signals in the wide TESS band light curves, which may stand out due to amplitude or morphology. Assuming intentionality, we expect magnitude changes of at least 2%–10% in order to stand out from noise. This would indicate either the presence of megastructures of sizes comparable to several Jupiter radii, or very high-energy emissions, which are possible for lasers (Kipping & Teachey 2016). However, it is key to consider these amplitude changes in the context of the usual light-curve behavior of the source; a brightening event in a star that has flares is less likely to stand out unless it exceeds the typical magnitude change due to stellar activity by a significant amount or has an anomalous shape. This example is particularly relevant as a flare star is further discussed later in this section.

As for light-curve changes in shape, a signal that mimics the light curve of the supernova is one possible morphology to look for, as noted by Nilipour et al. (2023). Unfortunately, the long timescale PDCSAP calibration of TESS 2 minute data is not currently reliable for recovering low-amplitude slow signals, though work is underway with the TESS FFI data (Hattori et al. 2022). Therefore, we focus on shorter-term flares and dips in this work.

As expected from the range of stellar evolutionary stage of our sample shown in Figure 4, the TESS light curves exhibit a variety of variability amplitudes and timescales. This includes rotational modulation, pulsation, flares, and eclipses. In Figure 6 we present a sample of six of our Ellipsoid crossing candidates from Table 2 that demonstrate a range of stellar variability behaviors. As noted earlier, all of these targets show no discernible change in their normal variability patterns coincident with the Ellipsoid-crossing times. TIC 279614617, also known as CPD-57 1131, is an RS CVn Variable with high amplitude (9.5%) modulations from the 7.4 day rotational period (Martínez et al. 2022). This system also exhibits several large flares throughout its TESS light curve, one of which occurs during the Ellipsoid window of interest. However, these instances are consistent with both the magnitude changes and the "fast rise, slow fall" behavior of known stellar flares, which are typical for these magnetically active binary systems (Tovar Mendoza et al. 2022). TIC 453100169 (HD 63669) is an MIII giant star with long-period quasiperiodic modulations. The shapes of these modulations are visually compelling, but stay consistent throughout the yearlong TESS light curve with no detectable changes during the window of interest. TIC 349972412 and TIC 350954265 are eclipsing binary systems. We calculated the orbital periods of these systems using the box least squares periodogram (Kovács et al. 2002), from which we report a 84.5 day period and 3.3 day period, respectively. A detailed characterization of these binary star systems—which show no contextual anomalies—is beyond the scope of this work. TIC 260266066 (HD 43902) and TIC 382187278 are nearby giant stars (d∼621 pc and 432 pc, respectively), with apparent asteroseismic oscillations in their light curves. No unusual behavior, coordinated with SN 1987A or otherwise, is detected.

Zoom In Zoom Out Reset image size

We also used an outlier detection algorithm, conceptually based on DBSCAN (Ester et al. 1996) and built with Scikit-Learn (Pedregosa et al. 2011), which was developed specifically for detecting anomalous stellar light curves (Giles & Walkowicz 2019, 2020) and applied to TESS light curves derived from the FFIs in Cycles 1 and 2 (D. Giles et al. 2024, in preparation). This generated a "weirdness score" for each target in every sector of data. This analysis was motivated by a search for unusual dimming patterns that could be indicative of the transit of an artificial structure. While inspired by systems like Boyajian's star, this outlier detection method is signal-agnostic, specifically with the purpose of avoiding assumptions related to what a technosignature might look like. Therefore, both intentional and unintentional technosignatures may be flagged as "anomalous" using this classifier.

The outlier detection algorithm was run on the 12 Ellipsoid crossing targets from Table 2 with crossing times in Cycle 1. None of these targets stood out as candidates needing further investigation related to anomalous light-curve behaviors associated with possible technosignatures. The algorithm flagged TIC 279614617 as a target with strong sinusoidal flux over time, ranking it within the top 2.3% most anomalous TESS objects for Sectors 2–5 and 7–13. TIC 382187278 also stood out in Sectors 4 and 8 as being in the top 4.2% most anomalous targets in these sectors. No changes in the stellar variability occurred across these data.