I decided to revive this thread rather than start a new one. It appears that D-Wave is the cover story for Time this week. Since I don’t care to pay for a subscription, I am wondering if anyone has read the article yet, and is it worth reading? I am skeptical of magazines such as Time when it comes to topics like this. I may track it down at the library, but I get the feeling this thread already has provided more information than the article will.
A recent study suggests that the D-Wave computer might not be nearly as useful as once thought. (This excuse that it was tested on the “wrong” type of problem is not taken seriously.)
There are an increasing number of people who are wondering whether the D-Wave computer works as advertised at all. And since the US Government and Google have spent millions on these, this is some matter of concern.
Here’s also Scott Aaronson’s (who’s mentioned in it) critical examination of the Time article.
For those interested I think this is the full text of the article.
I share your skepticism of the mass media, but it’s interesting reading at least with respect to the background of the company and a superficial overview of the technology.
I also found it interesting that they’re not using a gate-model approach (analogous to the logic gates of classical computers) but something called quantum annealing.
My knowledge of this stuff is just about exactly zero, but the sense I get from the article about how this works is that it’s distinctly non-procedural and seems to resemble old analog computers more than classical digital ones, but that may just be the way the article is written.
The other odd thing, as I believe already mentioned by Half Man Half Wit, is that no one is really sure if it’s really doing quantum computing, nor is it clear that there’s any real performance gain – at least yet – over classical computers. But this thing has some impressively high-profile financial backing.
So, if I am reading that right, even if the D-Wave is doing, or could do, any quantum computing, it is not, even in principle, able to do the sort of quantum computing (such as using Shor’s algorithm) that (according to theory) would significantly speed up certain sorts of otherwise impracticable computations, and what it can do, it cannot do significantly better than a regular computer.
Is there some way that D-Wave is not a scam?
If NASA is purchasing a few of these devices, it’s not because they have some calculations which can be performed more efficiently on them. It’s because they want to study the devices themselves. What they can do now is not as interesting as what they can be made to do.
Wolfpup, thanks for the link - though this sentence surprised me if it is from Time: “Another challenge rose and company face is there is a small but nonzero number of academic physicists and computer scientists who think they are partly or completely full of shit.” Italics added. If so, well, hmmm, cool?
Half Man Half Wit, thank you for the link to Aaronson’s response (and all the other info you presented previously).
I hope to see functional quantum computing in my lifetime, but I think Aaronson and Monroe have the most honest assessments of what current tech is. I am sure if something tangible develops, they will be among the first to acknowledge it. Right now, I am comfortable following their skepticism.
Thanks to all for the responses so far.
A silly comparison I once heard was: Think of it like an IBM punch card with ALL holes punched (like this, almost). Such a card clearly has ALL POSSIBLE data punched in it. Now you can operate on that all at once.
But, like Half Man Half Wit points out, although you can quickly find all possible answers, how do you then filter that to select the one you want?
My thanks to Half Man Half Wit for his cogent and clear explanations in this thread.
Yeah, that’s a silly analogy.
It reminds me of a clever trick that allows one to sort a set of index cards O(logN). Assume each card has a row of holes punched in the top. Then on each card, use these holes to encode its sequence number in binary, where a 0 is encoded by removing the paper between the hole and the top of the card, and a 1 is encoded by leaving it in.
To sort the cards:
1 line them up in a stack
2 insert a pencil through the least significant bit-hole
3 raise the pencil and lift the cards (only the ones with paper intact will rise)
4 place those cards behind those that remained
5 insert the pencil through the next hole
6 repeat steps 3-5 until done
I doubt that’s a good analogy for quantum computing, though.
Regarding QC, I don’t understand how we can understand an algorithm well enough to make it work, but not understand why it’s faster than another algorithm. I think this is the source of njtt’s confusion too.