Forums  > Software  > D-Wave, quantum computer application in finance?  
     
Page 1 of 1
Display using:  

zhanghaze


Total Posts: 16
Joined: Jul 2009
 
Posted: 2014-03-09 20:47
D-wave one claims to have 128 qubit and has been used by google. I wonder if this quantum computer is legit, how would it be used in finance/trading world? Speeds up monte-carlo simulation?

curvefitter


Total Posts: 124
Joined: Oct 2007
 
Posted: 2014-03-10 00:11
if you had 2 or more working qubits you'd be fighting off swarms of VC willing to give you millions for some ink, you wouldn't bother actually running excel on it..

FDAXHunter
Founding Member

Total Posts: 8335
Joined: Mar 2004
 
Posted: 2014-03-10 07:02
D-Wave doesn't really have a quantum computer in the sense that we understand quantum computing. Some people would even claim it's a scam.

Quantum algorithms can tackle many problems, but they excel the most at those that are NP-hard (pun not intended). For those algorithms we have no better solution than exponential running time (or more accurately: non-polynomial time).

Most Monte Carlo Simulations in finance being run currently are not NP-hard. Actually, they are easy (in the computational complexity sense). They are relatively trivial to parallelize. I think you would find that if you had a decent real quantum computer, that wouldn't really be on your lists of things to do. The first financial application I could immediately think of would be breaking open every Bitcoin wallet in the world at the same time Smiley. This is in fact the classical application (again pun not intended) for quantum computers.

The Figs Protocol.

Steve Castle


Total Posts: 306
Joined: Sep 2010
 
Posted: 2014-03-10 09:13
I believe they are still trying to prove it's a quantum computer and not just running something else.

[edit]
Earlier I said "quantum annealing" but I think that's what it's supposed to run. I'm confused.

article explaining it

in the words of one such quant ‘were on the whole either less quanted or not quanted at all’.

zhanghaze


Total Posts: 16
Joined: Jul 2009
 
Posted: 2014-03-10 16:10
Thanks for the articles and comments. Suspicious indeed. However, this kind of revolutionary change happens fast...

Are there any NP-complete problems in finance? Would a quadratic speed up help? Could it be used in finding Nash equilibrium, train AI algorithms, etc?

Cheng


Total Posts: 2819
Joined: Feb 2005
 
Posted: 2014-03-10 16:19
A quadratic speed up won't do a lot for a problem that can only be solved in non-polynomial time...

If you consider cryptography a part of finance then, yes, there are some NP-complete problems. Heck, one highly esteemed member here approached me once with a knapsack problem in disguise as part of a finance operation Smiley.

"Flammandus et Contemptus / Seyn Todt in Schwartz "

jslade


Total Posts: 1070
Joined: Feb 2007
 
Posted: 2014-03-10 17:07
"this kind of revolutionary change happens fast... "

I agree with that. The fact that D-wave has been around since 1999 and has yet to do anything that could be construed as quantum computing indicates that nothing much is really happening. The time between von Neumann's paper and improved von Neumann machines, by contrast: a couple of months.

Nobody really knows if quantum computing is a legitimate idea. People in computer science come up with "I can solve NP-hard problems with my imaginary computer architecture" all the time. A protractor can solve NP-hard problems in theory (see, for example, this paper: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.23.1744 ). In practice, it can't, because there's no such thing as a physical real number. I've always suspected that quantum computers are assuming "real numbers" somewhere.

Quantum computers would be good at solving combinatorial optimization problems, which come up sometimes. In fact, that's what D-wave is trying to do. Too bad a giant thing the size of a house can't beat a cell phone on this task.

"Learning, n. The kind of ignorance distinguishing the studious."

rod


Total Posts: 362
Joined: Nov 2006
 
Posted: 2016-10-27 01:58
This thread has not yet mentioned Scott Aaronson. He has written an enormous number of blog posts on D-Wave. So many, in fact, that his blog even has a D-Wave tag. Here's one of those blog posts:

Hopefully my last D-Wave post ever (December 17th, 2009)

Note the blog's header:

"If you take just one piece of information from this blog: quantum computers would not solve hard search problems instantaneously by simply trying all the possible solutions at once."


katastrofa


Total Posts: 359
Joined: Jul 2008
 
Posted: 2016-10-27 10:23
Aaronson has a bit of the envy the academicians have towards industry practitioners.

rod


Total Posts: 362
Joined: Nov 2006
 
Posted: 2016-10-27 10:27
Aaronson was a tenured professor at MIT. He now moved to UT Austin.

He is arguably the most famous computer scientist of his generation.

Why exactly should he envy D-Wave?

For allegedly building an actual quantum computer? People build quantum computers in academia, too. But in physics departments, not CS departments. One of my grad school acquaintances built one. The media all over the world hyped up his creation, like they do with D-Wave's creations.

katastrofa


Total Posts: 359
Joined: Jul 2008
 
Posted: 2016-10-27 11:59
Well...

Quantum computing is one of the overhyped areas of quantum physics. It's basically a way for quantum physics researchers to get grant money for playing around with random matrices and mathematics of finite-dimensional Hilbert spaces. Not much practical results have come out. And the quantum "computers" built at universities are just proofs-of-concept, unusable for any practical purpose. They still haven't resolved the basic problems like controlling decoherence. There is no chance in hell any of the academic machines could have been sold as a commercial product.

D-Wave, however, sold its machines to a real, for-profit company: http://www.dwavesys.com/our-company/customers

Can Mr Aaronson say the same about his machines?

rod


Total Posts: 362
Joined: Nov 2006
 
Posted: 2016-10-27 12:10
I know very well how academic hyping works. Yet, the field of quantum computing is barely 2 decades old. Perhaps it won't ever get anywhere, but something interesting may be discovered.

Ørsted found in 1820 that electric currents create magnetic fields. Only in the late 1880s did Hertz discover electromagnetic waves. Things take time.

Having for-profit companies as clients proves nothing. Google also hired Kurzweil, who is a huckster. It only proves that Page & Brin & Co fall for charismatic snake oil salesmen.




katastrofa


Total Posts: 359
Joined: Jul 2008
 
Posted: 2016-10-27 12:21
"Yet, the field of quantum computing is barely 2 decades old."

Umm, no. It's a bit older than that. Feynman started talking about it in 1982.

"Having for-profit companies as clients proves nothing."

In 2016, neither does publishing papers... unfortunately.

rod


Total Posts: 362
Joined: Nov 2006
 
Posted: 2016-10-27 12:36
"Feynman started talking about it in 1982"

Shor's algorithm was published in 1994, which is considered the field's date of birth.


"In 2016, neither does publishing papers..."

Papers in experimental physics may prove something, unlike the claims of snake oil salesmen.

There are also mathematicians like Gil Kalai publishing papers on why quantum computing may not be feasible. This is healthy. Kalai's debate with Aram Harrow was a good read.

Some people doing mathematical research on quantum computing are advancing fields such as algebraic geometry and optimization. Even if quantum computing never goes anywhere, the funding did produce results for other fields. Quants using correlation matrices may be using mathematical results produced with funding for quantum computing.

katastrofa


Total Posts: 359
Joined: Jul 2008
 
Posted: 2016-10-27 13:19
"Shor's algorithm was published in 1994, which is considered the field's date of birth."

Very convenient! 12 years since the birth of the concept. And when was the *last* new quantum algorithm published? I remember a talk on this subject about 10 years ago or so, in which the speaker - a researcher in quantum computing - said that the field is desperate for new algorithms and very little new work is being done on this.

If you want to say that the non-practical benefit of physicists playing mathematicians is some new mathematics, just wait to hear what real mathematicians think about it ;-)

Playing around with Hamiltonians does not produce a lot of fundamental progress in theoretical physics either. It's mostly rehashing old stuff which is aimed at producing papers which will make to the cover of "Nature".

"Papers in experimental physics may prove something, unlike the claims of snake oil salesmen."

They may prove *something*, but this "something" is often very far from any practical applications. And dimissing large industrial corporations (Lockheed Martin, not Google!) as naive idiots who easily fall prey to snake oil salesmen is quite arrogant. Semiconductor industry research (Intel, IBM) is very solid and adds a lot of value on top of what universities produce. (See e.g. Parkin's research in IBM).

"Some people doing mathematical research on quantum computing are advancing fields such as algebraic geometry and optimization."

I'd be interested in concrete examples.

"Even if quantum computing never goes anywhere, the funding did produce results for other fields."

I'd be interested in concrete examples.

"Quants using correlation matrices may be using mathematical results produced with funding for quantum computing."

I'd be interested in concrete examples.

rod


Total Posts: 362
Joined: Nov 2006
 
Posted: 2016-10-27 13:53
"Very convenient! 12 years since the birth of the concept"

Concept and field are different things. If I am not mistaken, Feynman's paper was a bit of an interesting curiosity until Shor's paper.


"And when was the *last* new quantum algorithm published?"

I don't know. QC is not my field.


"If you want to say that the non-practical benefit of physicists playing mathematicians is some new mathematics, just wait to hear what real mathematicians think about it ;-)"

My claim was the reverse: mathematicians and computer scientists playing physicists. If the field didn't have (academic) experimental physicists at all, I would be much more worried.


"And dismissing large industrial corporations (Lockheed Martin, not Google!) as naive idiots who easily fall prey to snake oil salesmen is quite arrogant"

It's not arrogance, it's caution. If one cannot trust the self-interested experts who want to keep the funding gravy train running, is one supposed to trust non-experts who get their technical knowledge on QC from Nature at best, Wired Magazine at worst? This is why outsiders like Kalai are invaluable. Kalai may have the knowledge, but not the incentive to deceive others (or himself).


"Semiconductor industry research (Intel, IBM) is very solid and adds a lot of value on top of what universities produce"

Semiconductor physics is much more mature and established than quantum computing. The field is around 70 years old. Fairchild Semiconductor was founded around 60 years ago, and Fairchild wasn't living off hype for decades, they were selling chips to NASA and the Pentagon, living off the Red Scare.


"I'd be interested in concrete examples."

How can I give you concrete examples when QC isn't my field?

What I do know is that some people working on semidefinite programming (SDP) have been funded by the QC Mafia. Same for people working on topological insulators and alike. I am closer to SDP than to condensed matter physics, though. Since my field was (more or less) control theory, it's obvious to me why SDP is useful:

Do they use semidefinite programming in industry?

katastrofa


Total Posts: 359
Joined: Jul 2008
 
Posted: 2016-10-27 15:17
"If one cannot trust the self-interested experts who want to keep the funding gravy train running, is one supposed to trust non-experts who get their technical knowledge on QC from Nature at best, Wired Magazine at worst?"

How do you know that LMT did not simply hire 1 or 2 physics PhDs to vet the thing?

rod


Total Posts: 362
Joined: Nov 2006
 
Posted: 2016-10-27 15:28
Because:

1) outsiders don't really know who the good PhDs and bad PhDs in a given field are. In a sense, that is the purpose of conferences: to have an academic community ranking their own. Most papers are, then, advertisements and, thus, "useless" by design.

2) fresh PhDs are overspecialized. They know their thesis topic better than anyone else, but they haven't had the time to acquire breadth of knowledge, which takes a decade or more. This wouldn't be much of a problem if QC had an industry (like, say, semiconductors). However, D-Wave is the QC industry. Once one is out of academia, there is only D-Wave.

Microsoft has Station Q in Santa Barbara. However, I consider them part of academia. Fields Medalist Michael Freedman works there.



rod


Total Posts: 362
Joined: Nov 2006
 
Posted: 2016-10-27 15:57
Interestingly, the guy I met in grad school who built a QC was funded by the Intelligence Advanced Research Projects Activity (IARPA).

A computer science professor once said that QC was a bit of a hedge. The funding agencies allegedly do not expect much success, but what they want to ensure at all costs is that the Russians and the Chinese do not make breakthroughs in QC before the United States. In that regard, QC reminds me of nuclear fusion.

chiral3
Founding Member

Total Posts: 4986
Joined: Mar 2004
 
Posted: 2016-10-27 17:37
I remember hanging out with Freedman when he got that job. Around 2000 at the 100 conference. Dude was way laid back. I am guessing he had / has no responsibilities.

Nonius is Satoshi Nakamoto. 物の哀れ

tbretagn


Total Posts: 245
Joined: Oct 2004
 
Posted: 2016-10-27 17:57
---threadjack---
Got that in my mailbox a couple weeks ago:

Cambridge Quantum Computing (http://www.cambridgequantum.com/), London office, is looking for a Data Scientist expert in Machine Learning (Artificial Neural Networks is a plus).
The job will focus on the development of state of the art Machine Learning models in Financial markets.

I'm amused but not so amused...

---/threadjack---

Et meme si ce n'est pas vrai, il faut croire en l'histoire ancienne

rod


Total Posts: 362
Joined: Nov 2006
 
Posted: 2016-10-27 17:59
tbretagn, that job ad reads like it was composed by an experimental AI from someone's dissertation.

jslade


Total Posts: 1070
Joined: Feb 2007
 
Posted: 2016-10-28 22:55
FWIIW Berkeley has a quantum computing startup; Rigetti quantum computing.
https://www.technologyreview.com/s/600711/the-tiny-startup-racing-google-to-build-a-quantum-computing-chip/

I graduated in 2004, and was thinking of pivoting into the field; had an idea for applying concepts well known in quantum dynamics/chaos to QC field. I think Shepelyanski or one of the Italian guys eventually independently turned one of these ideas into a paper. At the time I knew what I was doing was a punt into a trend, rather than a pivot into something real. It had all the hallmarks of a noodle theory disconnected from experiment. There were/are lots of interesting QC algos and mathemetical developments, and very few interesting experiments. Everyone figured lithography would make the experiments inevitable, but that turned out to not be true.

It's interesting stuff, if only because it digs deep into interesting ideas in quantum mechanics, but betting on it as a technology is something for DARPA types rather than sane investors. There is no "how to get from point A to point B" in the same way as something like, say, sending people to Mars.

"Learning, n. The kind of ignorance distinguishing the studious."
Previous Thread :: Next Thread 
Page 1 of 1