Time crystals and machine learning
Things have certainly gotten interesting. Between all of the talk just about everywhere about the metaverse and the Facebook/Meta name change you have to mix in what happened with time crystals before that and you have clear evidence that we live in interesting times. You might be wondering if these two topics are related? Why did I end up putting them as back to back topics? The answer to that question is rather simple. The topics were arranged by the order of inspiration. Back on July 28, 2021 researchers from Google, Stanford, and Princeton released a paper on arXiv called, “Observation of Time-Crystalline Eigenstate Order on a Quantum Processor.”[1] The paper is 16 pages and seems to conclude that, “These results establish a scalable approach to study non-equilibrium phases of matter on current quantum processors.” In terms of raw scientific achievement this is a truly interesting paper. We normally think of the states of matter in terms of solid, liquid, gas, and plasma. The math explaining the time crystals within that paper was a little intense. They model and explain the idea of a time crystal where a regular and repeatable cycle is created without burning any energy.[2] This would appear to defy the 2nd law of thermodynamics making it a very interesting discovery.
What does all of this talk about time crystals have to do with machine learning you might be asking? I tried to set the stage in that first paragraph that this discovery is really really cool in terms of theoretical physics and science in general. When the phrase “time crystals” with quotations and machine learning are dropped into Google Scholar a total of 240 papers show up.[3] Some of those papers are part of phenomenon of interdisciplinary crossing where people who have a very clear academic focus elect to take a very popular term and work it into the title of their regular academic research trajectory. In this case that is a very interesting thing to consider doing given the complexity of the mathematics involved in trying to model a time crystal. I could not model one on my home computer or within any of my cloud instances. Nothing that I have available to me can model quantum computing. Note back to that paper title above and the last two words that denote that the research was done on a quantum processor. That directly limits the number of researchers that are going to be able to contribute to this space. Access to quantum computing is rarified air in terms of cost and equipment. You could maybe get access with something like IBM Quantum where you can get access to quantum computing from the cloud.[4] I did spend some time looking at the programs and services that IBM Quantum has online.[5] It is sort of interesting to see which ones are online and how many Qubits they have at each location.
After thinking about the definition of time crystals and looking around to figure out how I could access quantum computing in the cloud I ended up digging into a book from 2020 called, “Machine Learning Meets Quantum Physics.”[6] The book seems to really dig into things people are doing with machine learning within the context of quantum physics. That preview link lets you see roughly 61 pages of the book including the introduction. They footnote in a very similar way to this weekly newsletter.
Links and thoughts:
10 to 20 thousand people who care about machine learning watch Yannic every week. You can see what Yanic is up to this week here, “[ML News] Google introduces Pathways | OpenAI solves Math Problems | Meta goes First Person”
This week while researching and thinking about time crystals I watched and listened to Linus and Luke talking tech during the WAN show, “We Messed Up - WAN Show November 5, 2021”
I watched Bea and Bozhong on the “AI Show Live - Episode 38 - Ignite Recap” edition
For those of you who really like talking about hybrid clouds you are in luck this week, “5 hybrid and multicloud architecture designs for an effective cloud strategy”
Top 5 Tweets of the week:
Footnotes:
[1] https://arxiv.org/abs/2107.13571
[2] https://www.quantamagazine.org/first-time-crystal-built-using-googles-quantum-computer-20210730/ This was the best article I read on the topic that seemed generally to be written to be read by people other than physics officiantos
[3] https://scholar.google.com/scholar?hl=en&as_sdt=0%2C6&q=machine+learning+%22time+crystals%22&btnG= These papers are a real mix of things that are useful to read and things I’m curious about how they got published
[4]
https://quantum-computing.ibm.com/
[5] https://quantum-computing.ibm.com/services?services=systems
[6] https://www.google.com/books/edition/Machine_Learning_Meets_Quantum_Physics/gzfpDwAAQBAJ?hl=en&gbpv=1&printsec=frontcover or here if you wanted to buy it https://link.springer.com/book/10.1007/978-3-030-40245-7
What’s next for The Lindahl Letter?
Week 43: Practical machine learning
Week 44: Machine learning salaries
Week 45: Prompt engineering and machine learning
Week 46: Machine learning and deep learning
Week 47: Anomaly detection and machine learning
I’ll try to keep the what’s next list forward looking with at least five weeks of posts in planning or review. If you enjoyed reading this content, then please take a moment and share it with a friend.