Addressing Landauer's question concerning the influence of static gate defects on quantum information processor performance, we investigate analytically and numerically the case of the quantum Fourier transform (QFT) with defective controlled rotation (CROT) gates. Two types of defects are studied, separately and in combination: systematic and random. Analytical scaling laws of QFT performance are derived with respect to the number of qubits
n, the size
δ of systematic defects, and the size
ϵ of random defects. The analytical results are in excellent agreement with numerical simulations. In addition, we present an unexpected result: The performance of the defective QFT does not deteriorate with increasing
n, but approaches a constant that scales in
ϵ. We derive an analytical formula that accurately reproduces the
ϵ scaling of the performance plateaus. Overall, we observe that the CROT gates may exhibit static and random defects of the order of 30\% and larger, and still result in satisfactory QFT performance. Thus we answer Landauer's question in the case of the QFT: far from being lethal, the QFT can tolerate tremendous static gate defects and still perform its function. The extraordinary robustness of the QFT with respect to static gate defects displayed in our numerical and analytical calculations should be a welcome boon for laboratory and industrial realizations of quantum circuitry.
The quantum Fourier transform (QFT) is one of the most widely used quantum algorithms, ranging from its primary role in finding the periodicity hidden in a quantum state to its use in constructing a quantum adder. Testing how the QFT performs under more realistic conditions, we find that the QFT, when used for period finding, shows extraordinary robustness with respect to static gate defects. For instance, replacing all rotation angles $\ensuremath{\pi}/{2}^{j}$ of the controlled rotation gates in the QFT circuit by $\ensuremath{\pi}(1+r)/{2}^{j}$, where
r is a uniformly distributed random variable taking values in the range $[\ensuremath{-}1,1]$, effectively resulting in a QFT with random gates, the QFT performs well above the expected random result. However, it is important to keep the
2j terms in the denominators of the rotation angles, resulting in random, but hierarchically random, gates. Relaxing this hierarchical structure of the QFT circuit, we find that the performance of the QFT deteriorates significantly. This observation indicates that the hierarchical structure of the quantum circuit of the QFT is more important for the observed robustness in performance than the precise actions of individual gates. In addition to the specific example of the QFT circuit studied here, this observation also corroborates our experience with more general and more complex quantum circuits. Thus, backed by our detailed numerical and analytical results, we may condense the results of our research into the following general principle: The topology of a quantum circuit is more important than the precise actions of its gates.
In excellent agreement with our numerical simulations of Shor's algorithm, equipped with a truncated quantum Fourier transform of bandwidth
b, we find that its performance scales $\ensuremath{\sim}$${2}^{\ensuremath{-}{\ensuremath{\xi}}_{b}n}$, where
n is the number of qubits, ${\ensuremath{\xi}}_{b}=1.1\ifmmode\times\else\texttimes\fi{}{2}^{\ensuremath{-}2b}$, and the bandwidth
b is the number of quantum states coupled by the quantum Fourier transform. Nonexponential behavior is observed for small
n and explained analytically. The large-
n exponential scaling implies that
b=7 is sufficient to operate a 1000-qubit quantum computer running Shor's algorithm on the
95 performance level and implies hardware savings of the order of half a million rotation gates.