Academic Correspondences
The other day, I was browsing the ArXiv when I found an interesting paper:
Probing the quantum to classical transition via simulated trap merging
I don't really care to delve too deeply into the details here, but suffice it to say that there is a huge difference between the quantum and classical worlds, and the transition between them is incredibly tricky to study [1]. At the quantum scale, each individual atom is incredibly important, but there are at least 10^23 atoms in every mol, which is one of the smallest macroscopic units other fields (like chemistry or biology) consider. For this reason, most researchers ignore individual atoms in favor of mean-field approaches, which merge all the particles into a big glob of matter. The problem is that this approach loses all of the interesting "quantumness" that people love.
If we could figure out how to directly study individual atoms at the macroscopic scale, it would become possible to study quantum effects in real-life systems and potentially learn how the effects really affect the classical world [2]. The paper proposed using a parallelized approach to symbolic computation to allow for better analytical calculations and thereby probe the transition via directly adding one particle at a time to the existing analytical system. It would use modern supercomputers to provide a better equation in a way that could be directly tested experimentally.
Honestly, it was a Nobel-prize-worthy proposal if the experiments worked. Without a second thought, I e-mailed the last author:
Hello Dr. G. Orcman
I am Dr. Qedem, currently a postdoctoral researcher at MIT. I recently found your paper entitled, Probing the quantum to classical transition via simulated trap merging and find it incredibly interesting. I have access to Summit, which is one of the fastest supercomputers available and have deep knowledge of General-Purpose GPU computing. I also have connections with a few experimental labs that might be interested in this proposal.
I can make myself available any time this week, so if you are willing to work together, please let me know your timezone so we can arrange a meeting.
Looking forward to talking soon! Dr. Qedem
I was rather nervous after sending the e-mail. If I'm honest, I don't usually send messages to random people on the ArXiv, but I thought I had enough academic clout to at least get the ball rolling.
After about a week, I still did not receive a response and started thumbing through the paper in my spare time while working on a new symbolic computation engine. After a month, I had everything ready to do some basic integration tests and launch the code on a distributed system, but still no response.
Honestly, Academics are like that sometimes. The beauty of research was that each paper should provide the necessary details to replicate results and build into future work, so I decided to use the paper as a test case and continue on. Once those tests passed, I decided to send one more e-mail:
Hello Dr. G. Orcman,
I sent you an e-mail a few weeks ago asking if we could collaborate on your most recent paper, Probing the quantum to classical transition via simulated trap merging. Since then, I have replicated the results of your paper and am ready to launch a much larger simulation on Summit. The code can be found here: [insert github url]
I would still love to talk to you, if you have a moment. If you are interested, please let me know your timezone so we can arrange a meeting.
Thanks again, Dr. Qedem
After a week, I still received no response and decided to launch the simulation anyway. It would take about a month to finish, so I could work on other projects in the meantime.
At some point while the simulation was running, I received the following e-mail:
I no Idea what u say. Grog hav degree in quantum, NOT ENGLISH. Code Plese rite code in FORTRAN77.
Honestly, it was relatively well-structured for an e-mail from a tenured professor.
[1] For those actually interested, here are the reasons the quantum to classical transition is so difficult to study:
- Analytical (purely mathematical) methods need to account for each individual atom. With every new atom added into the system, the interactions with every other object must also be taken into account. The tedious bookkeeping here is intense. Currently, we can keep up with maybe 3 or 4 particles before it is no longer within the realm of human capability.
- Numerically (purely in code) things are a bit easier in some sense. N body simulations are well-known and used all the time in galaxy simulations, but unfortunately we just don't have the computing power to go beyond maybe a few dozen particles here. Again, the scaling is absolutely insane.
- Experimentally (via using literal atoms), one would need to trap a single atom and then slowly scale it up particle by particle until it is a macroscopic object. The issue here is that trapping single atoms is super hard. Right now, it requires near-0 Kelvin temperature, which can only be found in a vacuum (even less dense than outer space). From there, lasers must be used to guide atoms together one-by-one.
[2] Note that the mean-field approaches actually work quite well for most "real" systems and thus the quantum-classical transition is not too important for most engineering purposes. I personally find it interesting as it allows for people to better understand scaling laws numerically, but there are some interesting fundamental questions that are on the boundary between the two worlds.