Google Quantum AI
Today, Google announced a demo of quantum error correction on its next-generation quantum processor, Sycamore. The iterations in Sycamore are less dramatic. Same number of qubits, just better performance. And getting quantum error correction isn’t really news.
Instead, the signs of progress are a little more subtle. In previous generations of processors, qubits are error-prone, so adding more qubits to an error-correction scheme caused more problems than gains in correction. In this new iteration, it is possible to add more qubits and reduce the error rate.
we can fix it
The functional unit of quantum processors is the qubit, which can be used to store and manipulate quantum states, such as atoms, electrons, and masses of superconducting electronics. The more qubits, the more powerful the machine. It is believed that by the time we have access to hundreds, we will be able to perform calculations that are difficult or impossible to perform on conventional computer hardware.
That is, we assume that all qubits behave correctly. Generally they are not. As a result, throwing more qubits into the problem increases the chance of an error before the computation is complete. So even though we now have quantum computers with over 400 qubits, attempts to perform computations that require all 400 qubits will fail.
Creating error-corrected logical qubits is a commonly accepted solution to this problem. This creation process involves distributing quantum states across a series of connected qubits. (In terms of computational logic, all these hardware qubits can be addressed as a single unit, hence the name “logical qubits”). Error correction is enabled by additional qubits adjacent to each member of the logical qubit. We can measure these to infer the state of each qubit that is part of a logical qubit.
Now, if one of the hardware qubits that are part of the logical qubit has an error, the fact that it only holds part of the information of the logical qubit means that the quantum state has not been destroyed. means that Then measuring its neighborhood reveals the error, which can be corrected with a little quantum manipulation.
The more hardware qubits dedicated to logical qubits, the more robust they are. There are currently only two problems. One is the lack of spare hardware qubits. Running a robust error correction scheme on a processor with the highest number of qubits would result in using less than 10 qubits for computation. The second problem is that the hardware qubit error rate is too high for this to work. Adding an existing qubit to a logical qubit does not make it more robust. You’ll likely have many errors at once that you won’t be able to fix.
