Question about Dwave & Probablity.

Hello D-wave community,

Our team is developing an AI/ML application for the Dwave and are currently setting up test problems/data and getting familiar with the Ocean platform for Covid19 research. We're working on a GPS based contact tracing AI on a blockchain geolocation platform. We have a simple binomial probability calculation that we want to run on our machine in parallel with the Dwave QPU. After running a few dozen of the D-Wave example experiments over the past week, we familiarized ourselves enough with the Ocean platform that we are confident that we can move forward with the knowledge that our initial entanglement hypothesis has been satisfied and we are prepared to proceed.

When scaling up, we can combine this as a predictor variable for contact tracing along with the geolocations, and we envision this being a powerful data tool used for predicative uses. What we need to make that happen is to do a pilot test where we have 24-hour access to run this binomial and chi-square probability calculation for each NED data stream. Basically something like this python code:

 

Zsquare=[]

pvalue=[]

for Ned in Neds:

        EX = NedVal[Ned]-(TotalRuns*NEDspeed*8*0.5)

        snpq = (TotalRuns*NEDspeed*8*0.25)**0.5

        Zval = EX/snpq

        Zsquare.append(Zval**2)

            pvalue.append(scipy.stats.norm.sf(np.abs(Zval)*2))

ChiSq = scipy.stats.chi2.sf(np.sum(Zcum),len(Zcum))

NedVal is how many 1s were produced by the NED, TotalRuns is how many seconds the stream has been running, NEDspeed is how many bytes are produced from the device per second. The two-tailed p-value is then calculated for each NED, and the combined p-value for all NEDs are calculated from the chi square survival function.

David J initially referred us to the Dwave-hybrid for this problem, but it looks like I still need a binary quadratic problem to feed it. We can put the calculation of a z-score in the form of a binary quadratic:

Z^2 n = 1/pq x^2 + 2p/pq xn + p/q n^2

where p and q are constants. Is there a way to have D-wave solve for Z^2 n given x and n? I understand that this is a problem typically done with ease on a classical computer (since quantum computing BQMs usually scan the whole solution space at once), but for our experiment we need to run it both locally and on D-wave.

 

######

If anyone would like to join us on this problem you can post here or DM.

0

Comments

0 comments

Please sign in to leave a comment.

Didn't find what you were looking for?

New post