It has been a while since I wrote about some really fundamental quantum physics. I feel that you, my dear reader, have not suffered enough during that drought. So, quantum physics it is. Even better, we are going to talk about entanglement and the strange case of one-way EPR steering. One-way EPR steering is an idea that has moved from a purely theoretical suggestion to something that might actually work in practice.
Let’s do some physics
The concept of entanglement in quantum mechanics expresses the idea that seemingly separate quantum particles can have correlations that are larger than would be possible in a purely classical world. When combined with superposition, in which particles have an indeterminate mix of two properties, it becomes pretty mind blowing.
Since the research in question used photons, let’s use photons for our examples. Imagine that I have a device that produces pairs of photons that are entangled in their polarization state. (Polarization describes the orientation of the photon’s electric field.) The polarization could involve oscillating in parallel with the lab table, or it could be oscillating vertically. But, whatever polarization one photon has, the entangled one has the opposite. As soon as I measure one, I know the other. So far, so not special.
I don’t have to measure the polarization along those two directions though. I could measure at 45 degrees. At that angle, the photons are in a superposition state of being at +45 degrees and -45 degrees. If I were to measure 10 photons in identical superposition states, I would get, on average, half of each polarization.
But my photons are part of an entangled pair. When I make a measurement on photon, it randomly chooses a polarization. As soon as it does so, the state of the second photon is locked into place. This is despite the fact that the two photons may be separated by light years and have no way of knowing about the measurement made on either photon. This seemingly nonsensical result was pointed out by Einstein, Podolsky, and Rosen (hence EPR). Experiments have since borne out that reality is, in fact, nonsensical.
If that did not disturb you, try the following. Our two entangled photons are sent in different directions. One is captured by Alice, and the other by Bob. Alice can change the results of Bob’s measurements by making her measurements first. And, if she knows what sequence of measurements Bob is going to make, she can steer the results of those measurements through her own choice of measurements. Of course, Bob can do the same to Alice. This is EPR steering.
A version of this called one-way EPR steering means that Alice can manipulate Bob’s results but Bob cannot manipulate Alice’s. This is what the researchers have demonstrated.
Or sort of demonstrated, as part of a larger collection of results. What the researchers really did was close some of the loopholes that allowed a bit of uncertainty about whether steering was really happening in previous experiments. But the experiments also show that not all quantum states are steerable. The states have to be relatively pure. A photon in an entangled state could also be mixed with a photon that is not in an entangled state, for instance. But one-way steering does not tolerate that very well.
The researchers also showed that the relative detection efficiency matters. The state can be less pure if Bob is less efficient at detecting photons than Alice. Or, to be more specific, the experiment uses a technique called heralding. Every entangled photon is accompanied by a second photon that can be split off. Bob and Alice use the arrival of this second photon as a signal to make their measurements. If Bob has difficulty detecting the herald photon, he can’t easily steer Alice’s measurements, while Alice can still steer his measurements.
Is the case for steering completely closed? Well, maybe not. The researchers were dissatisfied with how to distinguish between steering and non-steering states. So they dove into the theory and derived a new way to distinguish the two cases. Their new limit is more general than the previous result. That is really useful because it gives everyone else something to use. In their own words, though, their expression represents a condition but not a condition. It may be possible to come up with an even stricter limit that is both and .
I’m sure that if we all thought really hard about this, we might think of possible applications. And, yes, EPR steering has implications for quantum key distribution. But really I don’t care. This is quantum physics at its best: exploring the implications of the theory and twisting commonsense into knots.