This memorial was composed by friends and colleagues whose memories of Michael are shared below.
This semester, we had the brief joy of sharing Michael’s company during his stay as a fellow at the Simons Institute. Michael’s passing in September was sudden and tragic, and though the news has already reverberated through the community, we continue to struggle to comprehend the loss. Michael was a brilliant researcher: his creativity and sense of aesthetics were obvious to anyone who spent even ten minutes in his company, and will be remembered through his work.
At the age of 25, Michael was already becoming an iconic figure in the fields of spectral graph theory, linear algebra, and optimization. He made progress on a number of fundamental and notoriously hard problems: designing better algorithms for the k-server problem, computing the stable distribution of a random walk in nearly linear time, solving Laplacian linear equations faster than sorting, and developing new approaches to many sampling problems in linear algebra. In a striking single-author paper, Michael gave the first algorithmic construction of Ramanujan graphs of every degree, solving a problem which had been open since the 80’s. For this work, he won the best student paper award at a top TCS conference. These are just a few highlights; Michael was a coauthor on more than 15 papers, with over 25 collaborators. These numbers will grow as more of his ongoing projects make their way to publication in the coming years.
To us, Michael was more than an exceptional researcher; he was a colleague and a dear friend. We admired Michael’s curiosity, honesty, and buoyancy, which we wish to commemorate here through stories from our interactions with him.
“It is hard to find good words describing Michael’s excitement and energy for theoretical computer science, but it was apparent in every interaction with him. Most research meetings with Michael were a unique experience: as his long-term collaborators Aaron and Yin Tat nicely put it, Michael seemed to have access to “The Book” with exactly the right view on a problem. Working with Michael often felt like gently nudging him to turn the page in “The Book” so that we could read the next paragraph, which he was always eager to share.
I fondly remember one of the many long research meetings that Michael and I had. We started our meeting fairly late and continued working until 2am, at which point I became too tired to think and we decided to stop for the day. I cycled home and got ready to sleep, but when I picked up my phone to set an alarm I noticed that I had received a barrage of chat messages from Michael where he enthusiastically described the progress he had made since we left. I told him that I would get back to work the next morning, but that didn’t keep him from sending me messages full of valuable insights until 4am.”
“Michael's work is always a pleasure to read, but interacting with him online or in person was even more exciting. Michael would literally squeal with joy when encountering a new fact that interested him. I remember him bugging me at a conference last year because he wanted to know about an algorithm for sampling random spanning trees that some colleagues and I had been working on, but which we hadn't written up yet. Michael was relentless, and when I finally told him an outline, he giggled and immediately followed up with a torrent of thoughts as he toyed with the idea.”
“Driven by his curiosity and enthusiasm, Michael rarely had time for social conventions. During talks, he would often interrupt the speaker to ask questions. Eager to learn about the topic, he would try to guess what came next in the talk and suggest approaches for the problem at hand. It would be impossible for the speaker to sweep something under the rug: Michael would interrupt and ask ‘But…’, precisely spotting the minor technical detail missing. Even speakers from other fields would be surprised by the clarity and depth of his questions.
I remember taking an advanced algorithms class, where the professor wanted to skip some technical details of a proof. Michael wouldn’t let him. He kept insisting that ‘This part is really simple if you look at it this way.’, and went on to explain why this was the case with an argument that was elegant, yet too advanced for the class. He felt it was his duty to ensure that his fellow students were exposed to the ‘aesthetically correct’ version of the proof.”
“Michael’s love of knowledge and ideas extended far beyond theoretical computer science. In casual conversations, Michael was a fountain of information on diverse topics, from linguistics, history, and international law to brain chemistry and the Linux kernel. His breadth of interest was remarkable. In one of our first conversations, he asked me if I could account for Harvey Mudd College’s success in attracting women to the computer science major. In one of our last conversations, he speculated about what tides would be like if the Earth’s surface were a perfect sphere.
Michael never held back his excitement when he learned new things, no matter how mundane. I remember his amazement when I took off my shoe to remove a rock, and he discovered “no-show” socks for the first time. When we were at a workshop together in Hong Kong, he avidly gathered information about its history, economy, and municipal water supply. Michael rarely turned up at the office without an interesting riddle, or an unsolved IOI question, eager to involve others in his quest. Sometimes when I was explaining a proof to him at the whiteboard, he would turn and abruptly walk away, muttering to himself, because he didn’t want to spoil the joy of figuring it out on his own.”
“Despite his brilliance, Michael was humble and inclusive. He treated everyone with an open mind and without prejudice. Coming from a different background, I often felt like an outsider to the TCS community, especially at conferences. Talking to Michael changed this. He was non-judgmental and treated me as an equal, with no regard to my background or the fact that I only had a single theory paper at the time. I felt comfortable talking to him, not fearing that my questions might be considered stupid. It was even more encouraging that Michael was interested to hear about a research problem I was working on, and was asking me about it every subsequent time we met. We started working together on that same problem at the Simons Institute.”
“Michael was more than just inclusive: he actively engaged with everyone around him, on any topic, and generously shared his knowledge and observations with others. He loved teaching, and it was a joy to watch him do it.
Ever since we started at the MIT theory group, it was impossible not to feel Michael's presence in the graduate student area; we were effectively sharing an office with him. Our initial reaction to meeting him was “Who is this? And why does he seem to know everything?” We quickly realized that he indeed knew a lot about an incredible range of subjects that he shared with us over many nights in the first year students’ offices.
His thirst for knowledge was genuinely contagious: when we organized a reading group, we unofficially named it “The Michael Cohen Fan Club” with the stated goal of learning enough to be able to understand everything Michael said. While we can’t say we succeeded, we sure learned a lot of optimization along the way.”
Words can only go so far to describe Michael’s vibrant personality and intellect, both in research and in life. In losing Michael, our community has lost a singular talent, and a unique, kind personality. We feel lucky to have known him, and we carry his memory with us as a source of inspiration. He will be dearly missed.
- Letter from the Director, Fall 2017
- In Order Not to Discriminate, We Might Have to Discriminate
- From the Inside: Optimization (1 of 2)
- From the Inside: Optimization (2 of 2)
- Research Vignette: Ramsey Graphs and the Error of Explicit 2-Source Extractors
- Research Vignette: Promise and Limitations of Generative Adversarial Nets (GANs)
- Oral History: Russell Impagliazzo in Conversation with Dick Karp
- Simons Institute YouTube Channel Reaches 1 Million Views
- Looking Ahead: Spring 2018