We’ve discussed many times how important it is to achieve the best possible qubit quality levels, as measured with such parameters as coherence time and gate fidelity, in order to get good results from a quantum computer. This level is determined by many factors including the basic design, environment controls, and the microwave frequency electrical signals that are used to control the operation of superconducting, topological and spin qubits. (Note that other technologies such as ion traps may use lasers to control the qubits.) Although you might think that a designer will establish the best pattern for these signals when initially building the machine and never change them, this is not the case. In actual operation, the qubits can tend to drift slightly and their responses to the electrical control signals may vary from day to day. This will cause a degradation in qubit fidelity.

To combat this problem, providers of these quantum computers will perform frequent calibrations, sometimes on a daily basis, of their machines to determine the optimal frequency, pulse shape, and timing in order to keep the qubits in tip-top shape. Up until recently, these calibrations were implemented manually by a technician who would would run tests, measure, and update the parameters for the control electronics. As might be expected this method has drawbacks because it will take longer, requires special training, uses valuable technician time, and may not result in the best possible result due to the complexity of this calibration process.

To solve this problem, Q-CTRL has introduced a new feature in their Boulder Opal software called Automated Closed-Loop Optimization which uses AI techniques to automate this calibration process and achieve superior results to what could be achieved manually without needing to understand the physics of the qubits. The software can perform optimizations at the circuit level, but it can also be used with software modules such as IBM’s Qiskit Pulse or Rigetti’s Quil-T which allow a user to use pulse-level control on the supplier’s quantum computers and create a custom set of gates or achieve faster gate operation. The end result is that this can allow users to achieve better performance for their specific quantum algorithm than could be achieved if they just stuck to the standard gates. Q-CTRL asserts that use of this feature can result in an improvement in error rate by 2X or more.

Description of Q-CTRL’s Hardware Optimization Feature. Credit: Q-CTRL

Other companies have explored similar approaches. Google is using automated calibration software to optimize the performance of their qubits and Chinese company Origin Quantum just announced their own quantum operating system that has this capability built-in. However, it is not clear if these companies will open up these capabilities to end users so that end users can create their own set of control pulses specific to their quantum program.

For additional information about Q-CTRL’s announcement, you can view a press release posted on EurekaAlert here and a technical paper posted on Q-CTRL’s website here. A webinar explaining the details of the BOULDER OPAL Automated Closed-Loop Hardware Optimization tool will be held on February 18, 2021. Registration and additional information can be found here

February 10, 2021