The idea began with a frustration. Richard Feynman observed in 1981 that simulating even a modest quantum system on an ordinary computer demands resources growing exponentially with its size, and drew the natural conclusion, which was to build a machine that is itself quantum. David Deutsch gave the notion rigorous form a few years later. The advantage on offer is genuine but narrower than popular accounts suggest. Shor's algorithm would factor large numbers far faster than any classical method known, undermining much of the cryptography in use today, Grover's search offers a more modest quadratic gain, and simulating quantum matter itself remains the most natural application of all. For a great many ordinary tasks there is no known quantum advantage at all.
The hardware comes in several quite different flavours, from superconducting circuits chilled to near absolute zero at Google and IBM to trapped ions at Quantinuum and IonQ, neutral atoms held in optical tweezers at QuEra and Pasqal, and photonic and silicon spin approaches elsewhere. Each trades differently between fidelity, connectivity, and the prospect of scaling, and which will prevail remains unsettled. Our students work across this stack, descending to the machine level where pulses drive a physical qubit directly and rising to frameworks such as Qiskit.
Progress here has been extraordinary, and machines that lived only on paper a generation ago now run in laboratories and over the cloud. The question that presses now is what these machines will actually be good for, and answering it means devising algorithms, which is where the interest of researchers at FARII lies. A quantum computer can hold a superposition spanning an enormous number of candidate answers, but this alone buys nothing, since measuring such a state returns just one of them at random. The art lies in arranging the computation so that the paths to wrong answers cancel one another out while those to the right one reinforce, much as waves do when they meet. Only problems with enough hidden structure to permit such an interference pattern will yield, which is why the list of them is short and why extending it is worthwhile work.
None of this arrives without error correction. Qubits are exquisitely fragile, and the no-cloning theorem forbids the simple redundancy classical computers rely upon, so a logical qubit must be spread across many physical ones in a way that lets errors be caught and reversed without ever reading, and thereby destroying, the information within. Only recently has this been shown to work as the theory always promised, with error rates that fall rather than climb as more physical qubits are added, a threshold the field chased for decades. What makes the subject beautiful rather than merely necessary is where it leads. In the holographic correspondence between a gravitating spacetime and a quantum field theory on its boundary, the bulk appears encoded in the boundary much as a logical qubit is encoded in its physical ones, a parallel the HaPPY code makes concrete. Researchers here draw insight from these holographic models, which suggest new ways of thinking about how quantum information can be protected and, in turn, about how spacetime itself might emerge.