Alex Bronstein

Alex Bronstein

I am Russian by birth, Israeli by nationality and by belonging, Sardinian by adoption, and these days I spend most of my time in Vienna. I am a scientist, an engineer, an entrepreneur and an investor. I am a professor at the Institute of Science and Technology Austria and at the Technion, where I hold the Dan Broida Chair in the Department of Computer Science.

By training I am an electrical engineer, and by temperament someone who wants ideas to leave the page. I love doing research, building and disassembling things, teaching, and sharing ideas with other people. For a little over two decades my scientific work has wandered through geometry, inverse problems, optimal transport, machine learning and, more recently, chemistry and structural biology. The subjects changed, sometimes quite violently, but the game was often surprisingly similar: something in reality is hidden, we see only an incomplete and noisy trace of it, and we try to work backwards — or, better still, redesign the experiment so that reality reveals the thing we actually wanted to know. Along the way some of these ideas escaped from papers and became products and companies: depth cameras, video search, medical devices, quantitative finance, and several other attempts of varying degrees of success and educational value. The research page tells that story less as a list of papers than as a sequence of games we played with reality and what I think we learned from them; the talks page contains some of the more conventional version.

I like making things in general, not only scientific ones. Sometimes the thing is an equation, an algorithm, a camera, a scientific instrument or a company; sometimes it is a paragraph, a translation, a photograph, a piece of furniture, or a room. I have built or rebuilt homes in three different countries, and my current and probably most entertaining engineering project is an old Viennese Altbau villa that I am slowly trying to turn into an AI-driven dwelling without destroying everything that made it beautiful in the first place. The ambition is not to cover a pre-WWI house with screens, gadgets and things that beep, but almost the opposite: to make the intelligence disappear into it — light, temperature, air, energy, sound, and the small rituals of everyday life. It may eventually become a home of the future or merely the world’s most over-engineered light switch; at present both outcomes remain possible.

The rest of this site is what does not fit comfortably in a paper. I write essays about whatever I fail to leave alone — agency and physical law, markets, memory, art, and rather more Chopin than I would once have predicted. I write short stories. Once in a while I am asked to speak about something other than equations. I am not a poet, obviously, but I like playing with language, and whenever I encounter a translation of a poem or song I love that irritates me sufficiently, I make the completely unjustified decision that I can do better; the translations wander between Russian, Italian and English. I also play the piano without being particularly good at it, which has never stopped me from enjoying playing my favorite music unseriously. In whatever qualifies as spare time I read, run, travel, take pictures, sail, listen to music, collect Michelin stars and watch the real ones.

I am easily distracted — or perhaps more accurately, attracted — by beauty in all its manifestations, and I have never been particularly successful at keeping them in separate drawers. A theorem can be beautiful; so can a physical mechanism, a Chopin phrase, a line of poetry, a building, a landscape, a well-made machine, or an unexpectedly elegant solution to an engineering problem that had looked ugly for months. I do not mean that these are the same kinds of beauty, only that the sensation of suddenly seeing why something could not really have been otherwise feels remarkably similar to me. Perhaps this is also why I have never understood very well the boundary between “work” and everything else: I can spend an afternoon looking at a proof, an optical system, a façade or a musical score with essentially the same kind of curiosity.

When I was a first-year student I spent long hours in the library looking up whatever term, theorem, name or physical effect I had overheard in a lecture, and every now and then something would suddenly click. I called that sensation the joy of the first discovery. It did not matter that what I had just “discovered” was known to the ancient Greeks, or to everyone else in the room, or had been sitting patiently in the textbook for a hundred years. For a few minutes it was new because it had become clear inside my own head. I still find that feeling one of the two driving forces in my life. The other, of course, is love.

Perhaps this site is partly an attempt to keep making room for the first one by venturing into subjects in which I have no professional obligation to know what I am talking about. There is a particular pleasure in approaching something without having accumulated enough expertise to know in advance which questions are respectable, which connections are supposedly irrelevant, and which mistakes one has been warned not to make. The price is that one gets things wrong rather often; the reward is the occasional first discovery, several centuries late.

The rules here are correspondingly relaxed. I borrow ideas shamelessly, follow associations too far, digress whenever the digression looks more interesting than the road, and reserve the right to change my mind. My castle, my rules. If that offends you, there is always Facebook. Otherwise, follow me, reader.

Ways in

There is no sensible order to this site, and I have no wish to manufacture one. These are simply a few possible entrances.

Recently

The official version

If you are looking for affiliations, papers, grants, titles, companies and other evidence that I also hold relatively respectable employment, the less subjective versions live at ISTA, Technion, Google Scholar and LinkedIn. They are better places for that sort of thing; this site is for the rest.