Controlling Quantum Chaos with Randomness
Randomly controlling a quantum system can lead to control, but the uncertainty principle gets in the way
I’m only putting out a small Quantum Bite today because I’m on the road. Here’s a brief rundown of our new preprint “Universality of stochastic control of quantum chaos with measurement and feedback”, it hasn’t been peer-reviewed yet, but I’m excited about the result.
The idea came from classical chaos and control, and is pretty simple1. Imagine standing in a chaotic wilderness dotted with a few tiny, perfectly calm mountaintops. Place a boulder on one of those peaks and it can sit there forever—but miss the summit by a hair and it tumbles into the valley below. We want to escape the chaos of avalanches and rock slides, so our task is to sit a boulder up on the top of a mountain.
Now suppose you randomly stop paying attention. Every so often (flip a coin, roll a die), you look up and find the boulder, and nudge it some fraction of the way back up the mountain. This sounds thoughtless—counterintuitive even—but leads us to an interesting realization: There is a critical “nudge-rate” that will get our boulder up the mountain. However, any less often than this value, the boulder is lost to the valleys below. This is a classical phase transition in disguise.
We extended the idea to a quantum-mechanical boulder, and quantum fuzziness changes everything. These types of dynamics have little points of stability classically, but when you look at them quantum mechanically, there is no way to park a boulder at these points. Quantum mechanics has this bad habit of blurring those otherwise stable points.
This is really a battle with quantum uncertainty. If we put our quantum boulder at the top of the hill it is stopped (zero velocity) and we know where it is, but this violates the uncertainty principle! If we know where it is, we should have no idea how fast it’s moving! Even if you get a quantum boulder to the top of the mountain, it will always fall back down.
This alters our whole conception of “control” from before2, but miraculously, there is still a phase transition, but it behaves differently. Whenever we push our quantum boulder up the hill at just the right rate, it still never reaches the top, but it starts to be up there a lot more often than normal. Formally, the probability P(h) that the boulder sits at a height h takes the schematic form
(This looks like it diverges at the summit but that’s the secret sauce of quantum uncertainty, this denominator never actually gets to zero.)
The result is the following plot from the preprint:
The horizontal access is the “nudge-rate” p and the y-axis represents (roughly) how often the boulder is at the top of the hill. Classically, you’d get perfect control for p > 0.5, but quantumly, at p = 0.5, you can see that above that, it’s not at the summit a lot, but we’ve reached a threshold. Even at p = 0.8, you “only” hold the peak about 80% of the time.
So we trade absolute control to something a bit more “uncertain.” We now get to the top more often than not since quantum mechanics tends to get in the way and push us down the side of the mountain again.
Behind the scenes are simulations, analytic calculations, and connections to random walks, turbulence, and even market dynamics, all of which justify the “Universality” in the title. If you’d like the technical story—including how weak measurements implement the quantum mechanical “nudge”—check out our preprint!
The original classical paper was titled The Probabilistic Control of Chaos.
I’m neglecting here how to even implement control quantum mechanically which involves quantum weak measurements. That’s an interesting story and if you want to know details, please read our preprint!