# Can quantum computers handle 'big' data?

While there are many interesting questions that a computer can solve with barely any data (such as factorization, which requires "only" a single integer), most real-world applications, such as machine learning or AI, will require large amounts of data.

Can quantum computers handle this massive stream of data, in theory or in practice? Is it a good idea to store the data in a "quantum memory", or is it better to store it in a "classical memory"?

• just for clarification, are you simply asking whether it is possible and/or convenient to handle large amounts of data with quantum computers? – glS Apr 1 '18 at 19:56
• @glS Yes, that is what the final paragraph states. – Discrete lizard Apr 1 '18 at 20:45

It's not so much a matter of big data, but that of saving data. Quantum Storage is still (much like the rest of the field) in its infancy.

(Take what I write with a grain of salt. It's likely to change rapidly)
There are a few theories on how quantum computers might be able to hold "memory".

One of these is using nuclear spin. E.g. using long-lived nuclei in a quantum state. Converting an electron qubit (a qubit represented by an electron) to a nuclear qubit is possible.

### Why nuclear qubit/spin?

A nucleus's coherence time - the time for which its phase is constant (when considering its wave function) - is longer than that of an electron. The linked article (same one as before) details how one can increase the coherence time of a nuclear spin (to some extent). The matter is being researched, but there is indication that nuclear qubits can be a form of quantum storage.

### What makes it difficult

The quantum state needs to remain, well, quantum. Additionally, if you entangle two of your "storage" qubits, you are likely to lose data.

Due to no-cloning, one cannot simply "copy" a qubit (who's state is unknown), which is one of the reasons quantum storage is difficult.

As for "big" data, it's just a matter of how much memory you have.

• Interesting. Although I have one remark. "As for "big" data, it's just a matter of how much memory you have.", This, I think, is a bit naive. Be aware that when dealing with $10^5$ TB LIDAR data, (i.e., elevation data for every square meter in whole Europa (that's about a square big footstep for you non-metric folks))), your data doesn't fit into memory, no way! – Discrete lizard Mar 22 '18 at 20:57
• There is an entire sub-field of algorithmics (known as I/O or cache-aware algorithms, of which some experts are at my uni ) that says, wait the common RAM (random access machine, i.e. 'we can get data everywhere in $O(1)$') model doesn't help us here, I/O-latency is what dominates the actual running time, we have to be 'cache aware'. So, I think your final remark is a bit too naive. But feel free to expand on it! – Discrete lizard Mar 22 '18 at 20:59
• @Discretelizard good point. I'll rethink that last bit – ItamarG3 Mar 23 '18 at 7:33
• It may be interesting to think about how one can 'bridge' classical storage with QC, if Quantum storage is hard and possibly not very useful. – Discrete lizard Mar 23 '18 at 7:55