Who would've thought only a decade ago that quantum computers would become real in the upcoming future? Those of us without such hindsight need to rely on what's been reported by scientists, and recently all kinds of developments lend us to think that a quantum computing future isn't that far off. Take the latest qubit experimental set-up made at University of Oxford which demonstrates how trapped calcium ions can be manipulated and prepared to store qubits with record high fidelity. The researchers report extremely low error rates for both two-qubit logic gate operations and qubit preparation/manipulation. The most impressive reporting is an average error probability of 1×10-6 (one in a million) for a single-qubit gate operation, which is more than an order of magnitude better than previous demonstrations.
A more accurate qubit solver
Where traditional computers perform their calculations in binary – using 1s and 0s – quantum computers exploit the odd characteristics of the quantum state of particles at the atomic scale. Like Schrödinger’s cat, the value of a qubit isn't definitely 1 nor 0, but both at the same time. To “solve” a calculation, the quantum state is ended, so that the qubits take a classic 1 or 0 value. Setting the quantum states and superposition up correctly should mean a quantum computer will reach the same answer as a normal one. Unlike a conventional computer that needs to reach an answer sequentially, a qubit will instantly collapse to find the answer. This means that brute force operations that can take months for even today's fastest supercomputers could be quickly and effortlessly solved by a quantum computer, or so the theory goes.
[ALSO SEE] This video explains quantum computing
Before a quantum computer made up of enough qubits to perform meaningful computations can be scaled, however, scientists need to improve the quality of qubit manipulation. If the qubits aren't perfectly controlled, then errors can creep through that add up over the computational process. Techniques like quantum error correction and fault-tolerant designs are just a few tools that can help solve these issues, yet an error probability of less than 1% is required for this to work.
Thomas Harty at the University of Oxford, UK, and colleagues report using trapped calcium ions to prepare and readout qubits with an average error rate of only 0.07% after 150,000 repetitions. Additionally, the ions were subjected to a 3.2-gigahertz microwave pulse that toggles the qubit between its two levels an average error probability of 1×10−6 for a single-qubit gate operation was reported or more than an order of magnitude better than previous demonstrations. Here's how the system works, as described by
"In the authors’ scheme (see Fig. 1), a 43Ca+ ion is confined by radiofrequency electric fields (a so-called “Paul trap”) on the surface of a sapphire substrate. Electrodes connected to the structure provide the signals necessary for trapping the ions and driving changes to the qubit states. The choice of the 43Ca+ was crucial to the authors’ results: With a modest applied magnetic field (146 gauss), the energy-level separation between the two hyperfine ground-state sublevels of the ion is sufficiently large to become insensitive to small magnetic field fluctuations, abundant in a laboratory environment, that affected the performance of previous trapped-ion schemes. The two hyperfine states of 43Ca+ are thus ideal for representing the |0⟩ and |1⟩ qubit states (analogous to “0” and “1” in classical computing systems). A crucial property of a qubit is its coherence time (how long a quantum superposition of |0⟩ and |1⟩ states can be maintained). Unlike previous experiments with 40Ca+ ions, here the hyperfine levels’ stability led to a measured coherence time of about 50 seconds, marking a record for an atomic-ion qubit unshielded from fluctuating background magnetic fields."
"To “prepare” the ion’s electron in a well-defined initial state, the authors use laser-pumping techniques that are well established for atomic systems. They first drive the electron to one of the hyperfine ground states of 43Ca+ by shining a laser beam resonant with the atomic transition. From such a ground state, either the |0⟩ or the |1⟩ qubit state can be prepared using adequate microwave pulses applied to the ion through the on-chip microwave electrodes. In order to “read” the qubit states with high accuracy, the researchers exploit the fact that, when excited by a sequence of optical pulses, only one of the two qubit states would fluoresce. If the qubit was a |0⟩, optical excitation would bring it to a metastable state, which would not emit fluorescence upon application of a resonant laser beam."
While the findings represent a big leap forward for quantum computing, we're still a long way off. A classical computer has a much lower error rate, for instance. A quantum computer performing operations with an error rate of 10-6 on a 1 GHz processor on 32bits would still generate 32,000 bit errors per second. Practically, unusable at this current stage. Even so, the findings, joined by other recent updates like experiments made with superconducting qubits, show that scientists aren't sitting idle and are making progress.