Bwog Science was active yesterday – while new writer Riya covered a film about autism, EIC Betsy Ladyzhets went to a visiting Yale professor’s talk on quantum computing. She has little knowledge of both quantum physics and computing, but was still inspired by Prof. Steven Girvin’s self-described miraculous solution to the problem of quantum computing error.
Quantum physics (i.e. the physics of atoms and subatomic particles) is full of paradoxes. Perhaps the most famous of these is Schrödinger’s cat, a thought experiment devised by Erwin Schrödinger in response to the concept of quantum superpositions. According to this principle, a quantum system such as a photon can exist in multiple states of energy at the same time – until it is observed by the outside world, when it will collapse into one of the possible superimposed states.
Schrödinger demonstrated why he found this principle ridiculous by constructing a feline analogy: imagine a cat is placed in a sealed chamber, along with a measurement device containing a small amount of a radioactive substance and a relay system linking this device to a vial of poisonous acid. An atom of this radioactive substance might decay, which would cause the relay system to shatter the vial, poisoning the cat. But with equal probability, the atom might not decay, in which case the vial would remain intact and the cat would remain alive. However, because the whole thing is inside a sealed container, nobody could know if the cat is dead or alive until they opened the box. In this analogy, the cat is a photon, technically existing in both dead and living states (0 and 1 states) until someone checks on it.
Steven Girvin a professor and vice provost at Yale who studies the quantum mechanics of large collections of atoms, started his talk on quantum computing yesterday by calling attention to paradoxes like that of Schrödinger’s cat. “Is quantum information carried by waves or by particles?” he asked. The audience (of, I gathered, almost entirely physics students), chucked as he announced the answer: “Yes.” Quantum mechanics has come a long way since Schrödinger metaphorically killed (and didn’t kill) a cat, but it hasn’t gotten any less difficult to wrap one’s head around.
Girvin didn’t waste time wrapping his head around these concepts, though, as he launched into the subject of quantum computing. Quantum computing is basically what it sounds like: computing using the properties of quantum particles, such as superposition. Common, digital computing is based on a system of bits using a binary language in which every particle is always in a state of 0 or 1. In quantum computing, however, a quantum bit could use superposition to be both a 0 and a 1 at the same time. (This type of superposition, the quantum superposition of two macroscopically distinct states, is actually called a “cat state,” named after Schrödinger’s cat.) If many of these quantum bits (or qubits) were connected in a system, they would be able to solve massive computational problems, as their capability to be in two states at once allows them to explore an exponentially large number of possibilities at high speed. Girvin said that even a small quantum computer of 50 qubits could do operations too difficult to simulate on conventional supercomputers.
Quantum computing is still in its infancy, as very few quantum computers have been built (the first was constructed in 2009). Yet the subject is a hot one in current physics research, as both scholars such as Girvin and engineers at tech companies such as Google and Microsoft are racing to a practical means of constructing these fabled machines. The reason prototypes have been so impractical is that the huge processing capacity of quantum computers comes with a price – namely, enormous capacity for errors. Quantum computers are highly sensitive to noise disturbance, and the phases of the photons used in qubits can easily get out of sync, causing errors in calculations.
After explaining this problem, Girvin finally reached the main purpose of his talk: he and his team have found a way to correct this error. Their solution relies on simplicity; rather than scaling up the number of qubits in a system until the error is so large, it’s easy to detect (a solution that, Girvin said, some tech companies are working on), he looked for a way to error correct in a system with only one qubit.
Here’s how Girvin’s system works. An empty box stores a small number of microwave photons, or, photons at microwave frequency. Information (the type of computational information that would theoretically go into a quantum computer) is stored in those photons. A single qubit, or ancilla bit, is then attached to this box. The ancilla bit is in a cat state – in other words, it superimposed in phase 0 and phase 1. Attached to the photons, it becomes an oscillator, able to shift the photons between phase 0 and phase 1 depending on the status of its superposition. The system then uses a piece of technology called a Maxwell’s demon, a photonic machine able to measure the energy imbalance between light beams.
If noise or some other type of environmental disturbance displaces the qubit (error), it will oscillate, causing an energy shift in the set of photons that knocks one of them out of the system. This results in a change from an even number of photons to an odd number of photons, also known as a parity shift. Girvin explained that, in his system, the Maxwell’s demon can measure the photon set’s parity up to 500 times without the act of measuring causes error to the state. Thus, by regularly measuring parity, the system can determine if an error has occurred in the qubit. This information is communicated out using code words that are set in a cycle of four, similar to the phase shifts of rotational motion: lose one photon (odd), lose two (even), lose three (odd), lose four (even/back to the beginning). With the knowledge of how many times a photon has been lost, it is possible to restore the original system and correct the error, like returning pool balls to the center of a table after they’ve slipped into the pockets. Girvin explained many more technical details of his system that are not explored in this article; if you’re curious, hit this writer up and she can send you her notes.
Girvin proudly told the crowd that his system has been experimentally demonstrated to reach the break even point (or QEC) that declares a qubit system free of error, although they only just barely got there. In future experiments, he hopes to make the system’s lifetime longer (past 500 parity checks), and control systems with more than one qubit. It’s possible that, someday, quantum computing systems capable of solving immense math problems will be created from Girvin’s simple system. But theoretical physics, like any other field of science, is a long process, and it will take many more dead-and-alive cats to get us there.
Photo via Betsy