# Quantum Bayesian Networks

## October 30, 2015

### Watch Out For the Academics on Halloween Night

Filed under: Uncategorized — rrtucci @ 6:40 pm

You are now entering the IQIM_Caltech Twilight Zone @iqim_caltech @preskill #NSF

John Preskill: Hmm…so Princeton is competing against IQIM for that quantum computing grant…

Saddle up my horse, Tonto the postdoc. I’ve got some physics business to take care of.

Stalin: Good job with arXiv, comrade Paul Ginsparg.

Donald Trump scaring Captain Kirk in an episode of the Twilight Zone.

Max Tegmark: Give me research funding or I will haunt you in Many Worlds of 3 types.

Advertisements

## October 25, 2015

### Quantum Gravity Photo

Filed under: Uncategorized — rrtucci @ 4:45 pm

This POV (Point of View) photo reminds me of quantum gravity (in a poetic sense). I found it in the Blog of Francesco Mugnai.

It’s beginning to look like Quantum Computing will elucidate Quantum Gravity, both by allowing us to perform simulations of theories of it (as Feynman predicted) and by enriching the theory itself (for instance, quantum information, error correction and complexity theory inspired elucidations of the black hole information paradox and Maldacena’s gauge/gravity duality).

Our company http://www.artiste-qb.net has a unique POV regarding quantum computing, that of quantum bayesian networks.

## October 21, 2015

### Snow White Bayesian Network For a Collection of Sets

Filed under: Uncategorized — rrtucci @ 5:21 pm

CB net= Classical Bayesian Network
DAG= directed acyclic graph

In supervised learning, you are given the graph (aka structure) of a CB net, and some data, and you evaluate the node probability matrices from the data. In unsupervised learning, you are given only data, and you are expected to come up with the structure and node probability matrices of the CB net from that data. Nowadays there are computer programs that do both supervised and unsupervised learning of a CB net on classical computers. I believe a quantum computer can do unsupervised learning of a CB net at least quadratically faster (due to Grover’s algo) than a classical computer. In fact, I have a patent for doing unsupervised learning of a CB net on a gate model QC. (The Quail group at NASA has proposed doing this also with a D-Wave annealer QC).

And yet, many of the examples of CB nets that show up in the literature (See, for example, the wonderful work of Andrew Gelman) were obtained “by hand”—their structure was derived without the help of a computer, arrived at partly by logic and partly by hunch. The quality and value of these CB net models depends on how well they fit the data.

So can I provide some guidance on how to find the structure of a CB net by hand? I don’t know how the experts do it, but I’ll tell you how I think about it.

There is one situation that I like to call the Snow White CB net (I call it Snow White because it’s the fairest CB net of them all). It concerns finding a CB net that connects a collection of sets.

Snow White DAG
Suppose you have $n$ sets $A_1, A_2, \ldots A_n$ which are not necessarily mutually disjoint.

1. Merge all sets that are equal into a single set.
2. Write an undirected line connecting those pairs of sets that overlap (but not if they don’t overlap).
3. If $A_i \subset A_j$, then replace the undirected line joining them by an arrow pointing from $A_i$ to $A_j$. Thus

$A_i \subset A_j$
$A_i \rightarrow A_j$
$x\in A_i \implies x\in A_j$

all mean the same thing.

4. If $A_i$ and $A_j$ overlap, but neither is a proper subset of the other, then replace the undirected line between $A_i$ and $A_j$ by

$A_i \leftarrow A_i\cap A_j \rightarrow A_j$

5. Go back to step 1. Exit loop when last two repetitions yield the same DAG.

At the end, you will have a DAG in which the arrows all indicate a subset relationship. Also, by the end, all the root nodes (the ones with no incomming, only outgoing arrows) will be mutually disjoint sets. This is all very trivial and I’m sure a lot of people have come up with the Snow White DAG on their own. I just mention it in case you haven’t yet.

PS. In the above convention, a typical operating system with folders is represented by a tree, with the arrows pointing away from the multiple innermost folders towards the single outermost folder. The outermost folder is often called the root directory in operating system parlance, but here I am calling the innermost folders the root nodes.

## October 12, 2015

### Caltech Abhors MOOCs

Filed under: Uncategorized — rrtucci @ 4:12 am

I’ve expressed my very favorable opinion of MOOCs many times before in this blog.

Today I visited the Coursera and EdX websites and learned that they are currently offering 1,465 courses and 233 courses, respectively. So MOOCs are alive and well, at least today.

I was curious to see how many MOOCs Caltech is currently offering so I went to a website called “MOOC List”. According to that site, the grand total of Caltech MOOCs since the beginning of time is 5. 😸 Let me copy and paste the full list here:

1. Machine Learning (Caltech) Self Paced
2. The Science of the Solar System (Coursera) Mar 30th 2015
3. Galaxies and Cosmology (Coursera) Jan 6th 2015
4. Drugs and the Brain (Coursera) Jan 4th 2014
5. Principles of Economics for Scientists (Early 2013) Jan 7th 2013

Looks like the Caltech Evil Empire abhors MOOCs…

In a small galaxy called Caltech far, far away from Stanford University, Darth Vader Preskill is informed by Emperor Palpatine that the two co-founders of Coursera, Daphne Koller (aka Princess Leia) and Andy Ng (aka Luke Skywalker), are his offsprings, and that they are threatening the Empire of traditional Universities that Lord Vader has sworn to defend.

The following is a quote from the movie “The Empire Strikes Back”, with some minor modifications. My omissions from the quote are crossed out. My additions to the quote are placed in parenthesis.

Darth Vader Preskill: [kneeling before Emperor Palpatine’s hologram] What is thy bidding, my master?

Emperor Palpatine: There is a great disturbance in the (Educational) Force.

Darth Vader Preskill: I have felt it.

Palpatine: We have a new enemy. The young (Coursera) Rebel(s) who destroyed the Death Star. I have no doubt this boy (and girl are) is the offspring of Anakin Skywalker.

Darth Vader Preskill: How is that possible?

Palpatine: Search your feelings, Lord Vader. You will know it to be true. He(They) could destroy us.

Darth Vader Preskill: He’s just a boy. (They are just children). Obi-Wan can no longer help him (them).

Palpatine: The (Educational) Force is strong with him(them). The son (and daughter) of Skywalker must not become a Jedi.(MOOC Jedis).

Darth Vader Preskill: If he (they) could be turned, he(they) would become a powerful ally.(powerful allies)

Palpatine: [intrigued] Yes… He (They) would be a great asset. Can it be done?

Darth Vader Preskill: He (They) will join us or die, master.

## October 6, 2015

### Teaching Quantum Mechanics to Children, the Caltech and Waterloo Univ. Method

Filed under: Uncategorized — rrtucci @ 7:29 pm

I’ve heard some uncouth people, naysayers and sour grapes most of them, voice the extreme, malicious opinion that video games are junk food for the mind.

And what about the recent video games qCraft (by Caltech) and QuantumCats (by University of Waterloo in Canada) which promise to teach quantum mechanics to children? To the naysayers, those video games are poison too, quantum junk food, a way of wasting, piddling away the precious, jam-packed, fleeting years of youth.

It occurred to me that such opinions could be put to the test scientifically. So I was very happy when, while poring over the Lancet, a journal which I read faithful every Sunday, I came across the following article about a study conducted by the Mayo Clinic on this very subject.

Clinical Study of the Effects on Children of Playing Video Games qCraft and QuantumCats
by Mayo Clinic, Oct 1, 2015

Synopsis

We conducted a 1 year study on a group of 20 school children, ages 10 to 18, who showed an early interest in math and science.

10 of the children were our control group A, and 10 were our video gamers group X.

The children from group A were given classroom courses by really good high school teachers in Algebra, Geometry, Calculus, Biology, Physics, Chemistry. They were encouraged to consult Kahn Academy, Wikipedia articles, take MOOC courses and read books on science and math. Those yearning for hands on experience were encouraged to join a Ham radio club or local Hackers club and build their own electrical devices or else join an open source programmer’s group and start writing computer code at an early age.

Group A children were also encouraged to do some physical activity by going to a court or gym and practicing a sport, and joining a youth sport team if possible. Bicycling, swimming, jogging, dancing, etc. were all encouraged

The children in Group X were told that before taking a math or science course, it would be better if they first learned the basics of Science by playing some video games. By practicing how to build a quantum computer out of imaginary Lego blocks or throwing quantum cats with a catapult, they could learn the basics of quantum mechanics first, and then, if after a year or two of that they showed any promise, they would be permitted to take courses in science and math, and consult Kahn Academy, Wikipedia and all those other old-fashioned, boring resources.

If the children from Group X wanted to do some physical activity by playing a particular sport, they were told that it would be better if they first learned the basics by playing a video game about that sport. If after a year or two of that they showed any promise, they would be taken to a court or gym to learn the physical part of the sport.

We found that 95% of the children from group A went to good colleges. 5% never made it to college because they had already started their own high-tech businesses in high school and saw no need to go to college.

We found that 95% of the children from Group X never went to college, because they were recruited by the Army right out of high school as drone plane operators. 5% did make it to college, mostly MIT, where they eventually became professors.

The figure above shows a typical child from Group X, after 0,1,2,3,4 months into the clinical study. His cranial capacity diminished by 15cc after 4 months but we were told by Caltech and Waterloo that if we had run the study for a longer period of time, we would have seen that cranial capacity reaches a minimum after 4 months and then begins to increase. They also pointed out that our study was flawed in methodology and too low in number of children to be statistically significant.

Create a free website or blog at WordPress.com.