Tuesday, January 24, 2012

Predictive analogy, proportional analogy, and Asa H

There are different sorts of analogy.  Predictive analogy "is the process of inferring further similarities between two given situations based on some existing similarities," Bipin Indurkhya , Predictive Analogy and Cognition in Analogical and Inductive Inference, K. P. Jantke, Ed., Springer-Verlag, 1992. In Asa H predictive analogy is performed by dot product similarity measure between cases. Input vector components are compared between cases and output vector components are inferred.

Proportional analogies are "of the form: A is to B as C is to D" Scott O'Hara, A Model of the 'Redescription' Process in the Context of Geometric Proportional Analogy Problems, also in Analogical and Inductive Inference. In Asa H proportional analogy is performed by vector case extrapolation like: vector V4 = vector V3 - C*(vector V1 - vector V2).

Saturday, January 21, 2012

Command hierarchy

I am opposed to command hierarchies (a pyramid of bosses with employees below) wherever they may be found. It's been known for a hundred years that groups can make better decisions than individuals can.  A better system (than command hierarchy) would be something like that described in Maverick by Ricardo Semler, Warner books, 1993. Democracy is both more ethical and more efficient than dictatorship. The workplace needs to be democratic. A country is not free if the workplace is not free. It's sad what we have to pass on to our children.
I suspect that even wars and armies would be better run by a general staff rather than a single field marshal.
Or, perhaps, a group might do better at avoiding war.

Friday, January 20, 2012


I have several dozen "ebooks" (about half nook books and about half kindle) and a much larger number of dissertations, lab reports, journal articles, "white papers", etc. (as pdfs)  I have only read three ebooks from "cover to cover" and I did that on large screen devices.  On large screens reading of ebooks is comparable to reading conventional printed books.  I use ereaders when their small size is a virtue.  A family friend who is an avid reader recently traveled to europe with her daughter for a few weeks.  She bought a kindle so she wouldn't have to carry a large number of paper books with her.  ereaders are getting better all the time. (The local Radioshack salesgirl acted like I should buy a new one even though my kindle was less than a year old.)  With my bad vision, however, I still need a large screen in order to make the experience pleasant. Screen size might be less of an issue for people with normal vision. I also have the problem that (as of a couple of months ago) Emporia State's WiFi would not work with kindles at all and only worked with the newer color nooks. (Or so the IT people told me.  I don't have a nook ereader though I have downloaded the Barnes and Noble software and can read nook books on several of my computers.)

The COTS experiment

Since in the US rockets and spacecraft have all typically been built by industry anyway what NASA's COTS program really is is an experiment to see how small an organization can perform spaceflight and how cheaply, reliably/safely. (Clearly there are tradeoffs.  A cheap rocket may be possible but too unreliable/unsafe.)  Since most companies fail (go bankrupt and close) in 3 to 5 years and 90% of companies fail in 10 years this is a very real question and the answer is uncertain.  The two companies that are trying with Antares/Cygnus and Falcon 9/Dragon are currently well behind schedule. It will take some time and some number of flights to determine the reliabilty that results.  COTS is an engineering experiment.

My first computers

In the 1960s I bought 2 Geniacs(which used switches only) and a Minivac 601 (which used relays!), all SMALL logic circuit machines.  I then learned a little FORTRAN (from a book) and bought a Heathkit EC-1 ANALOG computer like the one shown below (it used tubes!). (I still have it today!)  Low cost digital machines didn't exist at that time and we were still using slide rules. (I'm told that Korolyov stood in the russian mission control using his slide rule to calculate orbital parameters.)

Thursday, January 12, 2012

Children and roomba

Only 2 small children have seen my roomba in operation; a 2 year old boy and a 3 year old girl.  Both were afraid of the roomba.  I wonder if by making the device in the shape of a dog, let's say, and covering it with fur, one could make it more child friendly.

Saturday, January 7, 2012

Some history

Me, 25 years ago, computer interfacing our plasma experiment. The code produced at that time is in my 2 Nov. 2011 post. (and Imad El-Jead's thesis of 1988 from Emporia State University)

Working with my AI, Asa F 2.0, about 10 years ago. Some results were published as Trans. Kan. Acad. Sci., vol. 107, # 1/2, pg 32, 2004.                                   

Today.  Downloading a dissertation to one of my ereaders.

Sunday, January 1, 2012

Project-based learning

Project-based courses are popular with students.  I have used projects as 1/3 or more of my digital electronics lab for 25 years.  But projects have their limitations.  I can think of no (small number of) projects that would cover ALL of the topics and details I need to cover in first year physics, for instance.

My book "Twelve Papers"

My book "Twelve Papers" was published last October.  You can download a copy free from my website http://www.robert-w-jones.com/ , Book, or you can buy a copy for the kindle reader from amazon for $1.99 (ASIN: B005SVVEYC) . (I would have made it free if they had let me.) The book covers more than 5 years of work and contains:

1. Experiments with Asa H  (The latest work on my artificial intelligence "Asa H" including initial efforts toward machine consciousness and natural language processing.)

2. Capitalism is wrong ("Wrong" both morally and in the technical sense of being incorrect.)

3. Chaining case-based reasoners

4. Objective analysis of student data  (A paper I have used for a number of years in the student laboratory.)

5. Experiments with machine creativity  (A paper I wrote for, and used as a part of, an honors course in creativity run a few years ago at Emporia State University.)

6. Experiments with a hierarchical ensemble classifier

7. Neural network categorization of experimental data  (A more advanced software analysis of Langmuir probe data.  See my Nov. 2, 2011 blog for much simpler software assuming a Maxwell-Boltzmann distribution.)

8. Quiet plasma in gas mixtures

9. Plasma optimization using data mining

10. The origin of large scale fluctuations in a Roth (bumpy) torus

11. Plasma diagnostics (A handout used in some of my plasma physics courses.)

12. Turbulent thermal insulation with clump regeneration and wall confinement  (Dated, but part of an argument that more effort/resources should be spent on wall confinement of very high beta plasmas.)

The Big Questions

I am interested in trying to answer the big questions.  Here are some of my current attempts:

Q 1. Why are we here?  A. Evolution acting on the animal kingdom.

Q2. What is our purpose?  A. The purpose of all life, to survive (reproduce and spread).

Q3. Why does anything exist?  A. An empty universe may exist but the question then won't arise.

Q4. Did the universe have a beginning?  A. Yes, at the "big bang."

Q5. What is thought?  A. See my blog of 29 Sept. 2010 and papers on my website http://www.robert-w-jones.com/ Asa H reduces MENTAL processes to a series of standard PHYSICAL processes.

Q6. What should humans value?  A. See my blogs of 11 Nov. 2011, 19 Feb. 2011, 21 Sept. 2010 and my website http://www.robert-w-jones.com/

Q7. What is consciousness?  A.  See my blog of 29 June 2011 and chapter 1 of my book Twelve Papers (available to download free on my website http://www.robert-w-jones.com/, Book).

Q8. Do we have free will?  A.  Depends upon your definition.  See my blog of 8 July 2011.

Q9. Does god exist?  A.  Such extraordinary beliefs would require extraordinary evidence.  No such evidence has been found.

Q10. What are human beings?  A.  Great apes

Q11. Does everything have a cause?  A.  Quantum mechanics says no.

A first course in plasma physics

I have taught a 1 credit introduction to plasma physics a few times and a 3 credit course a few times.  I have not found an introductory text book that I like so I have always taught the course from my own notes.  I have typically supplied printed handouts which cover perhaps as much as half of the course material.
It is not possible to cover everything that I might like in the 1 credit course.  My syllabus was typically:

1. particle balance, ambipolarity

2. energy balance and confinement time

3. the Debye length

4. the plasma frequency

5. Larmor orbits, cyclotron frequency, and cyclotron resonance heating

6. mean free path, crossections, collision frequency, diffusion, and diffusion across magnetic fields

7. fusion reactions, the Lawson conditions, breakeven, and ignition

8. particle drifts, EXB drift, grad B drift

9. magnetic mirrors (and the ionosphere)

10. closed magnetic systems, rotational transform, and equilibrium

11. plasma waves

12. minimum B and stability

Some of this follows my fusion energy review paper, J. Sing. Nat. Acad. Sci., 9, 71 (1980).
I am less sure what should be the coverage for a 3 credit course.  I have typically added things like:

13. A summary of the various plasma confinement systems and resulting fusion reactors. (Startng from something like J. R. Roth's review paper in IEEE Trans. on Plasma Science)

14. A summary of the various sorts of plasma sources. (Starting from something like my review paper in Physics Reports, Vol. 61, #5, June 1980)

15. Some discussion of plasmas in astrophysics.

16. A summary of plasma wave phenomena, instabilities, and their effect on plasma behavior and properties.

17. A discussion of various plasma diagnostics. (Starting from something like the chapter/paper on plasma diagnostics in my book "Twelve Papers" amazon # ASIN: B005SVVEYC)

18. (nonfusion) plasma applications and devices

You can't manage to cover all of these in a 3 credit course but I have picked from this list.

Doing (some) calculus without knowing calculus

It is common to have 2 "first year" physics courses.  One uses calculus, a "university physics" course, and one only uses algebra, a "college physics" course.  With physics enrollments lower than adminstrators would like there has been pressure to combine these into 1 single first year course.

Numerical methods allow at least some of the calculus to be done by students who only know algebra (plus some computer programming skills). The velocity component in the x direction is then the change in x devided by the time change, vx = dx/dt. The acceleration component in the x direction is then the change in vx devided by the time change, ax = d(vx)/dt.  One can then do integration on the computer with the simple assignment statements:

LET vx = vx + ax*dt

LET x = x + vx*dt

A simple planetary or satellite orbit program is then something like:

We can then explore elliptical orbits by increasing vy from 1 toward 1.414.
Beyond that we get parabolic and hyperbolic trajectories.            

Of course there are other issues with trying to make a single "size" course that would "fit" all students.  The lab component for one thing.                                                             

More people read your paper if it's rejected

I recall reading some years ago that scientific journal articles were read by such a small number of people that having your paper rejected (and resubmitted? once? twice?) actually resulted in more people reading it (as opposed to getting it accepted outright).  I have been unable to track down a reference to this but think it was based on the sort of analysis that de Solla Price used to do ("Little Science, Big Science")

Coding ideas

From studying neural networks (or logic circuits) we know that instead of training a single neural network with m inputs and n outputs one can train n separate networks each with only 1 output.  So, if we have a programming problem that requires n outputs and we don't know how to solve it we could start by trying to solve the coding task for just 1 of the required outputs.  If successful we could then try to work on other outputs. (Perhaps sticking to parallel processing?!)  We could also eventually try to connect the separate solutions if needed ("term sharing").

With the kind of vector representations I use in Asa H generalization can be accomplished by deleting less important (smaller?) vector components and reducing the dot product similarity measure required for categorization.  Specialization can be accomplished by adding vector components (during learning) and raising the similarity level needed for categorization.

(A category can be defined by specification of ranges over which each vector component (attribute) can vary.  It need not be defined by specification of the dot product alone. i.e., one can use other similarity measures, etc.)

Human values and gender

Human values aren't what they should be. (Being composed mainly of a small set of simple drives and aversions.)  But such as they are they appear to serve men a bit better than they do women.

For example: A woman should want the fittest man she can find.  Model Daisy Lowe, on the other hand, thinks that "You can always judge a man by his shoes."