We have reported recently on many developments in fault tolerant computer research with many different codes such as the surface code and q-LDPC (Quantum Low Density Parity Check) being researched. But a big concern about many of these error correction codes is the large overhead they will require in order to achieve a low enough error rate for a usable quantum advantage. For example, several proposed implementations of the surface code would require 1000 physical qubits to create one logical qubit with an acceptable error rate.
Recently, we have started see a few efforts to develop codes or technologies that have a much lower Physical-to-Logical ration that don’t provide full error correction, but still greatly reduce or provide detection of errors. These approaches may be enough to enable running a few applications that achieve a usable quantum advantage before full fault tolerant machines are available.
Some examples include the CAT qubits that Alice & Bob are developing that nearly eliminate bit flip errors, the new superconducting dual-rail qubit that Quantum Circuits Inc. is developing that provide built-in error detection, and a new approach that Q-CTRL is working on for creating circuits that provides error detection with only a 10% incremental qubit overhead. Both the Quantum Circuits and the Q-CTRL approaches could enable a new class of software algorithms that provide accurate answers using a technique known as “Try Until You Succeed”.
Now, IonQ has unveiled a new approach that provides partial error correction with a scheme they call Clifford Noise Reduction (CliNR). This scheme greatly reduces the error level of Clifford gates in the circuit and only requires a Physical-to-Logical ratio of about 3:1. This would compare to a different full error correction code that IonQ has studied that would have a ratio of at least 13:1 or more. So for the same number of physical qubits, this partial correction approach would provide an order of magnitude greater logical qubits. The only caveat is that it only works with Clifford gates, but not for non-Clifford gates like the T-gate or the Toffoli gate. Real quantum circuits still require the non-Clifford gates, but the hope is that the overall circuit error rate is low enough to provide an accurate enough output to be usable for some applications. IonQ is planning on making this technology available in their future Tempo processor which is scheduled for next year.
For more information about this development, you can view a news release posted on the IonQ website here. And to see the technical details, you can review a technical paper that IonQ has posted on arXiv here.
August 7, 2024
