Maurice Wilson's

Astronomy Research and Code


MINERVA Control Software

During the summer of 2016, I returned to Harvard for the third time, but this time not as an intern. After being accepted into their astronomy graduate program, the director of the Banneker Institute asked if I wouldn't mind spending my pre-gradschool summer at Harvard in order to tutor and mentor the new recruits. I was a part of the very first class of Banneker Institute interns and so they wanted my help in solidifying the awesomeness of the program so that it may run smoothly for years to come. In other words, the foundation of this internship was still being built and so it made sense that they asked me and a few other Banneker Institute alumni to help out when the program's second year commenced. I gladly accepted this offer and spent my summer on Harvard's campus once again.

Because I am a workaholic, I was excited to learn that I could begin my graduate school research because my graduate school advisor would be the same advisor I had for the Banneker Institute. This made for a very smooth transition into my first graduate research project. However, because it was the summer immediately before my ~5 year commitment to graduate school, I placed most of my focus on having fun and working a part-time job. Whenever I did get around to research, I would write code for the software that controls four telescopes located on the opposite side of the country.

Control Room, Bedroom

The MINiature Exoplanet Radial Velocity Array (MINERVA) is an array of four autonomous telescopes located in Arizona. I have talked a lot about the MINERVA telescopes in my High Precision Photometry project summary and a few blog posts: Part IV: High Precision Photometry of Transiting Exoplanets as I currently know it and Part VII: High Precision Photometry of Transiting Exoplanets as I currently know it. So if you want details and background information about MINERVA, check out those three links.

Figure 1: Two MINERVA telescopes are located in each enclosure. The fifth telescope known as MINERVA Red is in its own dome.

With the control software, we have the power to tell the telescopes what to do and when to do it. This lets us acquire data from bright stars every night while every astronomer remains asleep in bed. I say "bright stars" because, as you may recall, the MINERVA mission is focused only on observing bright stars. Our control software is optimized for finding the best, bright stars to observe throughout each night. As the program runs throughout the night, it frequently calculates which bright star (currently visible and preferrably near zenith) is the optimal target to observe. This works wonderfully and will prove to be very beneficial to MINERVA's primary mission of conducting a radial velocity (RV) survey of our nearest (i.e., brightest) stars in the night sky. The purpose is to obtain high precision RVs in order to determine whether or not small, Earth-like exoplanets are orbiting the bright stars. Our secondary mission with MINERVA is to gather high precision photometry on these same bright stars. With my code added to the control software, it is now easier for us to perform this secondary mission.

I will now do my best to explain how my code works. I have made a plethora of fancy diagrams so that you can visualize what my code is compelling the telescopes to do throughout the night. My hope is that by the end of this project summary, you will understand the subtle meaning behind the imagery within this diagram ...

This image was used in my poster. If you only care about grasping the knowledge that this specific diagram conveys, skip ahead and jump to my explanation here.

... as well as all of the complexity that is this diagram.


The Diagram

Diagrams have a tendency to look cluttered and confusing when they contain a lot of information, or, on the other hand, look clean and crisp when they leave out a lot of important details. I tried my best to find a good balance, but there is still the need for me to explain the inconspicuous information that my diagrams illustrate.

To start understanding these diagrams, click through this slideshow.


1 / 4
The 4 boxes on the y-axis are the 4 telescopes. The x-axis is time throughout one night of observations.
2 / 4
This is an image of one MINERVA. We have 4 of them.
3 / 4
Let's separate them into 2 groups because 2 telescopes are in each enclosure.
4 / 4
We'll keep the telescopes in this formation for the rest of the slides.


Former Software

Previously, the software could only run one mode of operations at a time. We could only conduct either spectroscopic or photometric observations throughout the night but not both.

Figure 2: All 4 telescopes gather RVs on various targets throughout the entire night.

Figure 3: All 4 telescopes gather photometry on various targets for the entire night.

The software was originally written only to serve the primary mission of MINERVA. The telescopes only needed to gather RVs. Eventually however, it made sense to capitalize on the photometry capability of the telescopes. The photometry feature was then added into the software. Although this worked, it unfortunately only permitted the telescopes to observe in one mode throughout a night, as you see in Figures 2 and 3. Furthermore, it required all four telescopes to be in the same mode at all times. For example, photometry could not be performed on two telescopes while the other two telescopes collected spectra.

This is where I come in. I sought out to write software that would allow any telescope to easily switch to any mode at any time of the night.

New Software

I will now illustrate how I have accomplished that goal. The following slideshow provides an example of how one night of observations could go.


1 / 3
We're at the beginning of the night. All 4 telescopes are getting RVs from various stars from 6 pm to 12 am. Let's imagine that the blue-outlined image symbolizes a variety of stars observed over time.
2 / 3
From 12 am to 1 am, two telescopes get photometry on one star while the other two continue gathering RVs on various stars.
3 / 3
All telescopes return to RV duty.


As I previously mentioned, the robotic telescopes search for optimal RV targets at all times. We can however, force the telescopes to measure RVs on a specific target if an astronomer requests this. In regards to photometry targets, such an observation is always scheduled ahead of time. Astronomers can submit a photometry target schedule to the MINERVA computers and the specified telescope(s) will gather photometry on the target specified in the schedule. This photometry schedule overrides any RV targets that the telescopes have chosen to observe at the time. At the specified time, the telescope(s) stop acquiring RVs, point to the scheduled target, and acquire the photometry. The remaining telescopes continue gathering RVs as if nothing changed at all.

Now that you understand the diagrams and capabilities of the software, let's complicate things a little. The next slideshow provides a complex sequence of events that my code is capable of performing throughout one night of observations.


1 / 6
Night begins with RVs.
2 / 6
T1 is released from MINERVA research duties for the sake of education and outreach. It is being controlled in real time now by some students interested in photometric measurements.
3 / 6
T2 was scheduled to get photometry on the star in the far right. It switches to photometry mode.
4 / 6
T1 is no longer manually controlled by students. It returns to its robotic state. T3 and T4 are still unperturbed by all of the action that's been happening since 7 pm.
5 / 6
T2 continues observing its photometry star. T3 switches to photometry mode and observes a star that's undergoing an exoplanet transit.
6 / 6
T3 switches back to RV mode. T2 continues obtaining photometry on its star for the rest of the night.


Hopefully that slideshow opened your eyes to how much more powerful this telescope array is now. I think the coolest part is the outreach aspect. A group of ... let's say ... high school students could—from a laptop—control T1 and point it to any target(s) they desire and consequently learn how professional-grade data is collected and analyzed by astronomers.

Another example that many, exoplaneteers in particular, may find exciting is Figure 4.

Figure 4: T2, T3, and T4 measure RVs of a bright star. T1 measures photometry on the same star while cycling through its filters for the sake of monitoring stellar activity.

Figure 4 illustrates a wonderful capability because it can be helpful to monitor stellar activity via photometry while simultanously measuring the exoplanet-induced RV signal. If the photometric measurements are sensitive enough, a correlation between the photometry and RV signal may be found. If found, this would help us improve the RV precision.

All in One

Thus far, I've used slideshows to get my point across. If I were to visualize my code all in one image though, that image would be this diagram.

Figure 5: Simplified summary of my code's capabilities. From 6 pm to 10 pm, all 4 telescopes gather RVs on various targets over time. All 4 point at each target simultaneously. From 10 pm to 2 am, T1 is controlled in real time by, perhaps, some high school students on a laptop, all for the purpose of education and outreach. At 2 am, T1 returns to professional research duties. For the rest of the night, T1 and T2 acquire photometry on a star. In the meanwhile, T3 and T4 continue gathering RVs, undisturbed by all of the changes occurring with T1 and T2.

www.astromauricewilson.com is developed and managed by Maurice Wilson.