Fujitsu and Osaka University have disclosed some new innovations for fault tolerant quantum computers (FTQC) that are aimed at reducing both the number of qubits as well as the runtime for running useful applications with error correction.
The first innovation would change how a universal gate set is implemented. Quantum scientists have known for a long time that to run any quantum program, one needs to have available a few basic gates, called the universal gate set. One such set as shown in the picture above on the left would consist of a CNOT, Hadamard (H), S (π/2 phase shift) and T (π/4 phase shift) gates. (There are other combinations of universal gates, but this set is the most common set taught.) From this set of gates, any other desired gate such as the Pauli gates (X, Y, and Z), Control-Z, Toffoli gate, or arbitrary phase shift gate can be synthesized using a combination of gates from this universal set. It turns out that in many fault tolerant quantum computing implementations, it is relatively easy to implement the CNOT, H, and S gate on a logical qubit, but the T gate is quite hard. A special process called magic state distillation is used to generate a gate like this and it requires a large number qubits and many gate operations.
What Fujitsu and Osaka University have come up with, is a replacement for this T gate they call a Phase Rotation Gate that can perform a phase shift on a qubit to an arbitrary angle Θ. This can be seen in the gate set shown in the middle of the picture above. They say that their process for creating this Phase Rotation gate is quite accurate and can save a great many qubits and quantum gate operations because it can directly generate an arbitrary phase angle rather than synthesizing a gate that does this out of many T and H gates. Recently, they have been able to improved the accuracy of this phase rotation gate by implementing redundancy in its operation to reduce the effect of a single rotation error on the phase angle. The second innovation is an improvement that parallelize the execution of creating the phase rotation gate so that more qubits can be processed in a shorter period of time.
The team believes that these steps can reduce to the number of physical qubits needed for a useful quantum calculation such as material energy estimates can be performed using only 60,000 qubits, far fewer than the roughly 1 million qubits previously estimated. This innovation can potentially accelerate the availability of early FTQC era processors that would be able to do commercially useful work.
Fujitsu and Osaka University will be continuing their partnership with future steps to demonstrate these innovations operating on real quantum hardware. For more about these developments, you can view a press release posted on Fujitsu’s website here, a presentation covering the developments here, and a YouTube video of a press briefing about it here.
August 28, 2024
