|Alexander J. Neal
I’m a physics graduate student at Ball State University. I graduated from BSU in December 2018 with a BS in physics and a minor in chemistry.
|I’m from Muncie and graduated from Burris Laboratory School in 2014.|
|I’ve worked in the Berrington Variable Star research group since August 2017. The research involves the study of eclipsing binary stars. The stars I’ve worked on are:
Along with observing and modeling these systems with programs such as PHOEBE, I’ve recently been doing research on the O’Connell effect and measures of it. The O’Connell effect is the apparent difference in brightness between the two peaks (quadrature) of the light curve. Assuming symmetry, there should be no expectation of a difference in brightness, however, some eclipsing binaries produce this characteristic nonetheless. A more intuitive measure is looking at the difference in shape of the two halves of the light curves mirrored over each other. Plots of this can be seen in the star pages linked above.
I wrote a Python code similar in function to the professional program PERANSO we use for determining the periods of our variable stars. The program samples many periods within a range and determines the most reasonable period in the range. For each of the sampled periods, the program fits piecewise least squares polynomials to the light curve, then produces a Fourier transform (FT) from the polynomial resampling (the FT is not totally necessary, but the difference in the fit is negligible and the FT is much easier to deal with than piecewise polynomials coding-wise). The resulting FT is then computed at the observed phases, and a cost function χ2 (i.e. ~error) is calculated. The period which produces the smallest χ2 value is then chosen to be the “best” period. The program can be continued wherein the period can be refined. Because very poor fits can result in very large χ2s, the “signal strength” is plotted by 1/χ2. The resulting graph clearly highlights the better period(s), and diminishes the noise. The error of the period is estimated using a similar method as PERANSO’s, developed by Schwarzenberg-Czerny (1991), which uses a “post mortem” analysis of the signal plot. The full width at half maximum (FWHM) of the Gaussian spike at the best period determines the error. Here’s a cool animated gif of the program’s output:
Here’s a more detailed look of the period finding process:
The fitting routine struggles mightily to make sense of it all, until it finally catches a break and produces a coherent fit near the “best” period (middle of the gif).
Make your own binary star outline! See bottom of the code to change the parameters (three examples are given).
|I’ve also done research in organic chemistry with the Albiniak group at Ball State from May to December, 2018. So far I’ve been making 2-(Benzyloxy) pyridine which can be used to make other compounds such as the Dudley Salt and possibly arylmethyl ethers and esters.|
Personal Interests and Activities
|I enjoy playing many games, both digital and in real life. I play the trading card game Magic: The Gathering regularly and you can usually find me at the Wizard’s Keep at 8 pm on Fridays. I mostly play Bethesda made video games (Elder Scrolls, Fallout) but also play many others.|
|I enjoy watching football (NFL) and basketball (NBA), but also statistical analysis. I occasionally edit and write on sports Wikipedia articles.|
|Other personal interests include paleontology, anthropology, philosophy and music.|
Other Cool Stuff
Binary equipotential contour plot.
Binary orbit diagram
Lagrange point “hilltops,” M1 fills its Roche lobe.