A quantum computer manipulates qubits through gates — small unitary transformations, the quantum analogue of logic gates. There are infinitely many possible single-qubit gates: every point on the sphere is one. Yet real hardware can only implement a handful of gates natively, and fault-tolerant architectures typically restrict themselves to an even smaller finite gate set.
So the natural question is: can a finite gate set approximate any single-qubit operation? And if so, how long does the approximating circuit have to be?
The answer is the Solovay-Kitaev theorem, proved independently by Robert Solovay in 1995 and Alexei Kitaev in 1997 (and later written up by Dawson and Nielsen in 2005). It says: yes, any finite gate set whose closure is all of — a set that is dense in the group — can approximate every single-qubit gate to precision in a circuit of depth only for a small constant (roughly 3.97 in the original proof, reduced to near 1 by later work). That polylogarithmic overhead is the heart of the theorem — and the reason quantum computing is practically universal.
Comments
Loading comments...