Semi-Autonomous Light Curve Analysis of Transient Nova AT2019tlu

Alexis Tudor and Noah Huerta, advised by Dr. Richard Plotkin

University of Nevada, Reno.

DOI: http://dx.doi.org/10.15629/6.7.8.7.5_6-1_S-2020_2

Citation: A. Tudor and N. Huerta, “Semi-Autonomous Light Curve Analysis of Transient Nova AT2019tlu.” Nevada State Undergraduate Research Journal. V6:I1 Spring-2020. (2020). http://dx.doi.org/10.15629/6.7.8.7.5_6-1_S-2020_2

Abstract

Every so often in the night sky an explosion will occur on the surface of a white dwarf. This is what has come to be known as a nova. This paper follows the eruption of nova AT2019tlu as it grows in brightness and then returns to its former state using the Great Basin Observatory telescope in Reno, Nevada. To analyze the data, we created an algorithm to semi-autonomously derive the brightness of the target star from photometric data. The results were inconclusive on the type of nova this may be, however the methodology used will prove to be useful for future observations with the Great Basin Observatory.

Introduction

The study of stellar evolution results in advanced knowledge not only about stars but about the nature of the universe itself, as stars are the source of all elements heavier than helium. When low-mass stars reach their ends, they shed their outer layers leaving behind their exposed core, a white dwarf. While the white dwarf is affectionately referred to as a stellar corpse, this is not necessarily the end for the star. See Ref. 4 for an overview of stellar evolution including white dwarf creation (1 ). White dwarfs can commonly become novae, explosions of less intensity than their more violent course, the supernova. The observing target this paper will focus on will be transient novae that light up the night sky for a bright and brief moment, allowing a glimpse into stellar evolution before fading into obscurity again. Novae are not as uncommon as might be thought, as many stars undergo them in a year, and some stars are recurrent novae that go off every year, unlike heavy supernovas that signal an end.

White dwarfs are incredibly dense, fitting the mass of the sun into an area similar to the radius of the Earth. If left alone, the white dwarf will stay that way indefinitely, creating and using so little energy that they are near permanent, however, if the star has a binary companion material can be siphoned off of its partner star and accumulate on the surface of the white dwarf. Once enough of the material is accreted on the surface a thermonuclear reaction can occur: a nova.

This report follows a recent transient nova AT2019tlu as it goes nova. The Great Basin Observatory was used to gain photometric data of the star. This data was analyzed in order to learn more about the star itself as well as about novae in general. This paper covers the theory behind nova and photometry, methods for data collection and analysis, and results and conclusions from the data.

Theory

Many white dwarfs end their lives as cold stellar remnants, however, this is not their universal fate. According to the Pauli exclusion principle, no two electrons can inhabit the exact same spin and energy state at the same time. Once a low energy level is filled, more electrons are forced to occupy higher energy states and move progressively faster. This energy causes electron degeneracy pressure, which is the force keeping white dwarfs from collapsing in on itself from the force of gravity (2 ). Because of this pressure, white dwarfs have an inverse relationship between their size and their mass; as more mass is added, the star gets smaller until it reaches the Chandrasekhar limit, 1.4 solar masses, where the white dwarf can go supernova after all (3 ).

The other kind of stellar explosion a white dwarf can undergo is more common. When a white dwarf accumulates mass, usually by stealing from a binary companion star, the mass can accumulate as a non-degenerate layer on the surface of the star. This mass gets hotter and hotter on the surface before experiencing a thermonuclear reaction with less force than a supernova. This event is called, appropriately, a nova (4 ). A nova, while less bright than its cousin the supernova, increases the white dwarf by several orders of magnitude, making a previously unobservable target visible to smaller telescopes. Additionally, unlike the destructive supernova, the nova leaves the white dwarf entirely intact and surrounded by another layer of gas and dust reminiscent of the layers left behind when the star became a white dwarf. This process can continue as long as the white dwarf does not accumulate 1.4 solar masses of mass in its core.

The process involved in creating the explosion is a runaway nuclear fusion reaction. Generally, the stolen matter from the binary star forms an accretion disk around the white dwarf. As more matter accumulates, the heat and pressure build until eventually it achieves nuclear ignition. Like a normal fire ignition is required the fuel is added to keep it burning, in this case the fuel is hydrogen. Technically nothing is burning, as there isn’t even fire which requires oxygen, stars really get their heat from the thermal energy released from nuclear reactions. The nuclear process starts when two Hydrogen atoms fuse together creating Deuterium. One of the protons in this step will lose its charge and become the neutron by ejecting a neutrino and a positron. These subatomic particles add energy to the system. In the second step another simple hydrogen atom fuses with the deuterium and irradiates a gamma ray and forming Helium-3. Gamma Rays are extremely energetic electromagnetic waves that also contribute to the immense energy of the system. In fact, gamma rays are energetic enough to photoionize atoms creating more hot plasmas. The final step occurs when two Helium-3 atoms fuse to Helium-4 and ejects two protons to be sent back to step one as simple hydrogen atoms (5 ). This process can build and build with sufficient fuel as the system of the nova only gets hotter making visible or brighter radiation and ejecting matter in space.

When analyzing changes in stars over time, such as a rotating binary star or a nova, a valuable tool in the astrophysicist’s arsenal is the light curve. A light curve is a graph of the amount of light coming from a star over time as the star grows brighter, dims, or moves around. For novae, this light curve is characterized by a sharp rise in brightness up to the peak of the explosion, followed by a slow decline that can last hundreds of days. From the shape of the light curve, analysis and classification of the nova is possible. Light curves are often the most accurately measured form of diagnostics for novae.

The process of analyzing light in the night sky is called photometry. Photometry uses a CCD that consists of a block of many ‘pixels’, which are designed to count photons. To take photometric data the CCD is pointed at the part of the sky where data needs to be acquired, and allowed to face that sky for a certain amount of time. This exposure time is a balance between capturing as many photons as possible to get a more detailed image of the sky, including fainter targets, and leaving it exposed too long and risking oversaturation as too many photons hit a pixel for it to contain.

Photometry data is not immediately usable. When the CCD collects photons, it also collects background radiation, photons that are ‘stuck’ in the CCD, and dead pixels. To solve this problem the image data is calibrated with three kinds of images called frames. A bias frame is taken with an exposure time of zero and with the shutter on the camera closed. This frame accounts for read-out errors, dead pixels and other errors inherent to the CCD. The second frame is the dark frame, which is taken with the shutter open but the light blocked. This removed thermal noise generated by the detector. The final frame is the flat field that compensates for obstructions in the light path and varying quantum efficiencies of pixels across the detector. While bias and dark frames are subtracted directly from the raw image, the flat field is divided from it as shown in equation (1)

(1)   Calibrated = Raw–Bias–Dark / Flat.

Using this photometry data it is possible to count the number of photons in a star and use that information to obtain the relative magnitude of the star using equation (2), where N is the number of photons captured in the star, and m is the magnitude of the stars.

(2)   (msrc)i = −2.5 log10(Nsrc/(Nref)i) + (mref)i

This calculation is done for three to five reference stars, and then averaged. This can be a lengthy process to do by hand, even with the software available for viewing and analyzing this data. This research seeks to automate the calculation of magnitude in order to capture the light curve of a transient nova without the tedious process of calculation by hand.

Material and methods

AT2019tlu was discovered on October 28th, 2019 (6 ) with a discovery magnitude of 15.4 (7). Using the Great Basin Observatory located in Great Basin National Park in Eastern Nevada, observations of the nova were able to be started on October 31st. Photometry was performed on the star using the observatory’s B, V, R, and Luminance filters for exposure times of 5, 10, 30, 60 or 300 seconds. The plan was to take data for about a month, however this turned out to be impossible as the star faded out of the sight of the observatory telescope on November 8th. Additionally, due to a telescope tracking error, the observations from November 3rd to November 6th were not able to be used. Nonetheless, the peak of the nova and the fading rate of the nova were captured in the usable data.

In order to process all of the data gained from the observatory, the data analysis was done via a semi-autonomous Python script using the astropy library (8 ). The program takes in the data, the dark frame, the bias frame, the flat field, and the location of the star in stellar coordinates. From there the program centers on the true pixel location of the center of the star, and finds the radius of the star using a version of Sobel edge detection (9 ). In order to reduce the effect of background cosmic noise in the calculations the photons in the background are counted over an area (not containing any stars) four times larger than the star itself, and then averaged so that a per-pixel background photon count can be obtained and subtracted from the counts of the star itself. The program will then find up to eight reference stars to use for the magnitude calculation by searching in the four cardinal directions and the four diagonal directions. The program will locate the star, find its stellar coordinate location, and then relay that information to the user to request the magnitude of that star. The user can then copy those coordinates into SIMBAD1, or any other stellar database, and enter the magnitude of the star into the program from there. Then the calculation is performed, the magnitudes are averaged, and the program returns the calculated magnitude of the target star. Figure 1 shows pseudocode for the algorithm used to obtain the magnitudes for all of the data.

Results

We recorded five nights of observations in the B, V, and R filters on October 31st, November 1st, November 2nd, November 7th, and November 8th (due to daylight savings the Great Basin Observatory encountered tracking errors from November 3rd through to the 6th). The measurement of V = 15.4 on October 27th was the discovery magnitude reported on Astronomer’s Telegram (6 ) in an early detection. We tracked the light curve from October 31st through November 8th, wherein the star became too dim to be seen with the GBO telescope. We caught the peak with an obvious jump from 15.18 mag to 14.29 mag on November 1st followed by the beginning of the descent to 14.36 mag (a smaller magnitude corresponds to a brighter object). At 14.29 mag at the peak, AT2019tlu was slightly dimmer than Pluto is from earth.

Figure 1 - For each day of observations.
Figure 1: Pseudocode for the program that calculates the magnitude of the stars semi-autonomously. The formula used in the second line of the algorithm differs from equation 1 in that the bias frame is not included in the code. This was intentional as the bias frame is included in the dark frames from the Great Basin Observatory.
Figure 2 - Star AT2019tlu.
Figure 2: This photometry data shows the star AT2019tlu. This image is made of the data from the B, V, and R filters of the Great Basin Observatory, and AT2019tlu is circled in white and magnified.
Figure 3 - Light Curve of AT2019tlu.
Figure 3: The light curve of AT2019tlu shows the peak of the nova and its rate of decline in the V filter.

Discussion

In summary, novae occur when a white dwarf siphons matter from a binary star until it accumulates enough hydrogen for a runaway nuclear fusion reaction to occur This fusion reaction occurs when atoms form together under immense heat and pressure to form heavier elements and release energy. In order to classify novae it requires tracking the light curve for hundreds of days and more importantly spectroscopic classification. Most novae light curves follow a power scale or exponential decay, however 100 days or even a year later can display a new peak. In fact, some of them are recurring and explode on schedule after years. Without more data, the only thing that can be said about AT2019tlu is that it is not an F-class nova and it is likely not a D-class nova according to the classification system proposed by Strope et al. (10 ). More detail about nova classification is found in the appendix.

Future work in identifying this nova requires hundreds of days of data, and if this project were extended the nova could be identified. In the long term it would be worth observing the nova for recurrence of the nova as recurring nova are a special kind of nova that provide interesting data for understanding the universe. The first author also intends to continue work on the algorithm used to perform photometric analysis on the images in order to make it more generalizable and more autonomous. Currently, user input is required for all neighboring star magnitudes. In the future, the program will work directly with SIMBAD to procure magnitudes for neighboring stars. Making the process of photometry autonomous will have a positive impact on the astronomy community, particularly on those using the Great Basin Observatory telescope, and the work described in this paper is far from done.

Appendix

The classification system defined by Strope consists of seven different types of nova: S, C, P, O, D, J and F. Other classification systems tend to describe the curves somewhat vaguely and could possibly lead to misinterpretation about the traits and physics of the nova. S-class, smooth, novae are the most common making up approximately 38% of novae and follow a “broken power law” decay. Unlike the graph below, on a logarithmic time graph the light curve would decline in a straight line with a steepening slope. C-class, cusp, novae initially behave similarly to S-class but show a secondary peak up to eight months later. P-class, plateau, novae experience a nearly flattened line at a V magnitude three to six below the initial peak after an initial magnitude declination. O-class, oscillating, light curves as a whole create a very similar shape to Sclass novae but at around three magnitudes below the peak begins tracing a sinusoidal-like oscillation above and below the smoothed trendline. D-class, dust dip, curves show an interrupting minima in the curve due to an expanding dust shell that cools and blocks light from the interior but eventually dissipates. J-class, jitters, novae show sporadic spikes in magnitude with no real build up or uniformity. F-class, flat-tops similar to plateaus keep a relatively constant magnitude. This is the only curve with no initial decline in magnitude.

Acknowledgements

The authors would like to acknowledge the support of Dr. Richard Plotkin as a professor and advisor while the work in this paper was being done, and Jeremiah Paul for giving the team access to the Great Basin Observatory. Without them, none of this would have been possible. Additional acknowledgement goes to Levi Ratto, who assisted with some parts of the astronomical analysis. The first author would like to thank Ryan Nunes for his support while completing the photometry algorithm.

References

[1] R. Nave, Hyperphysics, (n.d.). [Retrieved from]: http://hyperphysics.phy-astr.gsu.edu/hbase/Astro/whdwar.html.
[2] Electron Degeneracy Pressure. COSMOS, (n.d.). [Retrieved from]: http://astronomy.swin.edu.au/cosmos/E/ElectronDegeneracyPressure.
[3] S. Chandrasekhar, Journal of Astrophysics and Astronomy 15, 105–109 (1994).
[4] K. J. Shen, D. Kasen, B. J. Miles, D. M. Townsley, The Astrophysical Journal 854(1), 52 (2018).
[5] C. Wood, “The Sun: Fusion at Work.” American Chemical Society – ChemMatters, (February 2007). [Retrieved from]: http://www.acs.org/.
[6] L. Tomasella, S. Benetti, E. Cappellaro, M. Turatto, The Astronomer’s Telogram, (October 2019). [Retrieved from]: http://www.astronomerstelegram.org/?read=13232.
[7] AT 2019tlu. Transient Name Server, (n.d.). [Retrieved from]: http://wis-tns.weizmann.ac.il/object/2019tlu.
[8] T. P. Robitaille, E. J. Tollerud, et al., Astronomy & Astrophysics 558 (2013).
[9] I. Sobel, G. Feldman, (2014). “An Isotropic 3×3 Image Gradient Operator.” Presentation at Stanford A.I. Project 1968.
[10] R. J. Strope, B. E. Schaefer, A. A. Henden, The Astronomical Journal (2010).