An Azure service that provides quantum computing and optimization solutions.
Hello, @EA !
I have an answer from the Azure Quantum team that expands on the documentation regarding the calculation of the number of physical qubits required for a single T factory:
Edit:
There appears to be a formatting bug in Q&A preventing the second part of this answer from appearing once published:
Answering the first question:
A factory is constructed using multiple rounds of distillation, where each round of distillation can have multiple distillation units. Each distillation unit needs several logical qubits (depends on the type of unit). Depending on the output error rates, all units in a round require some specific code distance (typically increasing over the rounds). The code distance can be used to derive the number of physical qubits per logical qubit and therefore we can compute the total number of physical qubits per unit. Then we can compute the total number of physical qubits for that round by multiplying the counts by the number of units in that round.
Now we know the number of physical qubits per round. We assume that we perform the rounds in order and can reuse the qubits. Therefore, the total number of qubits for the complete factory is the maximum number of qubits over all rounds.
Finally, to produce all required T states for the algorithm execution in time, we may need multiple copies of factories. Therefore, the total amount of factory qubits is the number of qubits for one factory multiplied by the number of copies.
Answering the second question:
We do not consider the routing qubits to transfer distilled magic states into the algorithm qubits using multi-qubit Pauli measurements. However, we do account for layout overhead in lattice surgery to perform the multi-qubit Pauli measurements.
I hope this helps!