General > General Technical Chat
Google claims to have reached quantum supremacy
<< < (3/7) > >>
rfeecs:

--- Quote from: RoGeorge on September 23, 2019, 05:52:09 pm ---Only when I went into details, looking at the hardware implementation of various types of existing quantum computers, I realized that it should be possible to implement the same functionality of qubits and quantum gates without using quantum particles.

--- End quote ---

I read something like that recently:
"'Poor man's qubit' can solve quantum problems without going quantum"
https://phys.org/news/2019-09-poor-qubit-quantum-problems.html
rfeecs:
This video may be what the Google paper is about.  (I found turning on closed captioning was helpful):
https://youtu.be/gylmjTOUfCQ

This sounds something like:
   I will prove that my quantum computer is superior to a conventional computer at solving at least one specific problem... and that problem is simulating a quantum computer.  :o
RoGeorge:
The YouTube video is from March, wasn't the retracted announcement of quantum supremacy from last week this month?

Yes the phys.org you posted is one recent announcement, from Purdue University, the previous one I know is from Linköping University:
https://phys.org/news/2019-09-poor-qubit-quantum-problems.html
https://phys.org/news/2019-09-quantum.html

And then is the retracted Google announcement, that nobody knows what it is about.

It was quite a sour surprise to see them popping all of a sudden, in the same month.   :-\
RoGeorge:

--- Quote from: SilverSolder on September 23, 2019, 08:41:16 pm ---It is in any case always going to be slower than "the real thing" - no?

--- End quote ---

I think it will have the same speed as a quantum computer.

Same speed, and I'm not even considering the unfairness everybody is using right now to compare digital computers with quantum computers.  To detail what I mean by unfairness, when the big-O is calculated for a classic algorithm, they are considering only 1 bit questions, while for quantum algorithms big-O is calculated for questions about all the qubits at once.  I don't know why is this practice, is hard for me to believe that nobody really seen this unfairness.

For example, when the "guess my number" big-O is calculated, usually it is used the half-split method, so the question put to the "oracle" is "is your number in the first or in the second half of the given interval", and the oracle respond with yes/no, so only one bit, hence O(log n), while for the quantum algorithm they ask the oracle "what number did you pick", and the "oracle" answers with all the qubits at once by simply telling the number I chose, so the secret number is always guessed in one step, so O(1).  Very unfair way to question the oracle, if you ask me.  A fair thing would have been to use the parallelism for classic algorithm, too, not only for the quantum algorithm, and then the classic computer would have responded in 1 step, same as the quantum one.



At minute 27:27 there is a drawing of the quantum algorithm "guess my number", where the secret number (hardcoded) is 1101.  If instead of qubits we use normal bits, and instead of quantum CNOT gates we use classical XOR gates, then either the quantum or the classical computer will guess the secret number in one try, with big-O=1.

This suggests that there might be no advantage at all for quantum computers, but this is a software research and at the moment I'm interested in hardware.

Does anybody else noticed the unfairness when calculating big-O for the classical algorithm, or am I losing something obvious?
Marco:
I googled for the paper so you don't have to.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod