On quantum computing and IonQ Error Decoder

In my university we were told many times that the day will come and we'll use quantum computers. That was ~20 years ago.

IonQ announced "Industry’s First End-to-End Real-Time Quantum Error Decoder". What does that really mean?

Decoder is part of quantum computer that figures out where errors most likely happened so these could be corrected. Without correction the computer can't function reliably for long enough for real work.

Qubits making errors constantly is normal. The issue there is physics: reading qubit destroys its state so we can't look for errors directly. That's why there are extra "helper" qubits linked with original data ones which are answering "are data qubits 1 & 2 still match" repeatedly (parity check).

Reading helper states produces a stream of "syndromes" (yes/no answers). Rate of the stream is several millions of answers per second. That's the input of the decoder.

Output is exact a set of most likely places for the errors that explains "syndromes".

Now back to what IonQ announced.

1. Realtime decoders are not new. Google did it in 2024 (Willow chip).
2. Running a simulated correction for 408 logical qubits on a single M4 Max CPU is truly new. Data rate is about 1 megabit per second, but workload is quite complicated.
3. It depends heavily on IonQ's architecture and is hardly transferable to other platforms such as superconducting ones.
4. One reason less to doubt quantum computing overall.

What's left for useful quantum computer now is hardware with 12K physical qubits w/ same architecture and accuracy that runs stable for 30 minutes and more. They have 100 qubits today. Optimistic goal is to reach 10K by 2028 but I think that it likely won't be the case.

Also, a proof that it's useful. Real problems where quantum beats CPU/GPU haven't been demonstrated yet.

If these are achieved, some specialized tasks would be possible and achievements would worth the price. These are exponential problems: chemistry, materials, magnets.

What we won't get yet:

1. Breaking classical encryption.
2. Proven gains in finance, logistics, AI.
3. Faster general purpose PC.

So, 20 years later the useful quantum computers aren't there. But overall it advanced and problems are now mostly engineering, not physics. We likely still won't "use" quantum computers directly, but we may use materials and drugs designed with them.

#quantumcomputing