David Poulin
2020 sucks.
Yesterday I learned that David Poulin, a creative and widely-beloved quantum computing and information theorist, has died at age 43, of an aggressive brain cancer. After studying under many of the field’s legends—Gilles Brassard, Wojciech Zurek, Ray Laflamme, Gerard Milburn, John Preskill—David became a professor at the University of Sherbrooke in Quebec. There he played a leading role in CIFAR (the Canadian Institute For Advanced Research), eventually co-directing its quantum information science program with Aephraim Steinberg. Just this fall (!), David moved to Microsoft Research to start a new phase of his career. He’s survived by a large family.
While I can’t claim any deep knowledge of David’s work—he and I pursued very different problems—it seems appropriate to mention some of his best-known contributions. With David Kribs, Ray Laflamme, and Maia Lesosky, he introduced the formalism of operator quantum error correction, and made many other contributions to the theory of quantum error-correction and fault-tolerance (including the estimation of thresholds). He and coauthors showed in a Nature paper how to do quantum state tomography on 1D matrix product states efficiently. With Pavithran Iyer, he proved that optimal decoding of stabilizer codes is #P-hard.
And if none of that makes a sufficient impression on Shtetl-Optimized readers: well, back in 2013, when D-Wave was claiming to have achieved huge quantum speedups, David Poulin was one of the few experts willing to take a clear skeptical stance in public (including right in my comment section—see here for example).
I vividly remember being officemates with David back in 2003, at the Perimeter Institute in Waterloo—before Perimeter had its sleek black building, when it still operated out of a converted tavern. (My and David’s office was in the basement, reached via a narrow staircase.) David liked to tease me: for example, if I found him in conversation with someone else and asked what it was about, he’d say, “oh, nothing to do with computational efficiency, no reason for you to care.” (And yet, much of David’s work ultimately would have to do with computational efficiency.)
David was taken way too soon and will be missed by everyone who knew him. Feel free to share David stories in the comments.
Comment #1 June 30th, 2020 at 7:33 am
What a terrible way to die. Poor sod. Hopefully his family has enough saved up to bear with the pandemic.
How much performance do you have to sacrifice to get your decoder into BQP? And are there any particular Hamiltonians where optimal decoding is practical? Also, how much worse is performance in 2D+?
Comment #2 June 30th, 2020 at 12:24 pm
Talking about QC and error correction – how does one go about testing how “perfect” a QC is, as a blackbox? Just like we would test the integrity of a classical processor by running a test suite of operations on it.
Isn’t implementing Shor’s algorithm the best way to do this because it relies on perfect large scale coherence of all the qbits/gates, and the output is very obviously right or wrong?
As opposed to trying to just match against the theoretical probability distribution associated with some random circuit, which becomes vanishingly small and harder to compute (classically) as the number of qbits increases?
Comment #3 June 30th, 2020 at 6:29 pm
Very sad news.
As a Caltech freshman (in 2006), I had the opportunity to do research at IQI, supervised by John Preskill. I was assigned a desk in the same office as David Poulin, who was a postdoc at the time. Even though David had no responsibility to help me out, and I was too much of a novice to do anything useful, David ended up being my mentor and main collaborator. He was extremely generous with his time, and I wouldn’t have had nearly as enjoyable or productive of a research experience without him.
Comment #4 July 7th, 2020 at 5:31 am
Ron Graham, mathematician and juggler, author of Concrete Mathematics, has passed away on 2020-07-06. Rest in peace.