How to Tune a Quantum Computer with Dyspoissometer

Home

In the 20th century, human laborers would use a primitive device called a tuning fork to tune the musical notes on a piano in order to ensure that they complied with musical frequency standards. (Ah, the 20th century -- how mechanical it was!) Similarly, we can now use Dyspoissometer to tune the entropy quality of a quantum computer. You can imagine a robot showing up at your lab in a pair of overalls, Dyspoissometer in hand, ready to politely inform you that your qubits are slightly biased!

Following is the general procedure:

1. Instruct the quantum computer to produce a random mask list where the mask span, Z, is a power of 2. For maximum statistical significance, the mask count, Q, should equal Z. (2^16) 16-bit samples should suffice in practice. Each mask should come from the collapse of an N-qubit superposition (ideally, entangled) to an N-bit (classical) mask. Make sure to round-robin the qubits so we uniformally sample each of them.

2. Compute the logfreedom of the mask list with Dyspoissometer (see related functions on the logfreedom page).

3. Loop back to step 1, accumulating a list of logfreedoms.

4. Sort the list.

5. Compare the median value (in the middle of the list) with the output of dyspoissomater_logfreedom_median_get() after running it for 10 or more times as many iterations.

6. If the median is much less than expected, then the qubits are probably biased. But if the median is much more than expected, then perhaps the random qubit output is being pseudorandomly postprocessed to hide a systematic error caused by manufacturing problems. Either way, we have a problem necessitating qubit recalibration. Double check after running dyspoissomater_logfreedom_median_get() for as long as possible, perhaps in multiple parallel threads. In practice, (10^6) iterations should be considered minimal.