Sunday, March 24, 2024

HyperPhysics

Our usual first stop in research is Google and then the Wikipedia link that they provide if no other link shows up. Wikipedia then branches off. Or, scanning the information allows formation of a better prompt. 

Prompt? Yes, there's a new field with a title of "prompt engineering" which pertains to controlling output from a GenAI process (or session). At first, it seemed rather facetious. However, as the concept took hold, there are studies being done with using this method some of which are technical in nature. 

On the other hand, "prompt' from outside will not overcome deficiencies latent in the learning process. So, that will become more obvious as these studies go on. Essentially, "crap" cannot be trained out. Preprocess, in the old sense of "a priori" is a necessity (was and will be, as well). 

But, we're not tracking down that path. We're looking content and configuration which was our favorite topic for a while and has not gone away. 

So, we were searching on Navier-Stokes and solenoid. The first was prompted (pun) by a paper that discussed Einstein's PhD thesis. It's at a paid site, so we only saw glimmers of the topic and will look further. But, the other prompt was that Einstein and Infeld's look at the evolution of physics touches upon the device in their discussion of the "rise of the mechanical' view. The interest has to do with the fact that this work is 100 years old, plus or minus. As 1905 was the year that people became aware of Einstein's thinking. 

HyperPhysics
Too, AIn't and such plus things relating to ourselves will be understood better with an updated grasp of what Einstein was trying to show. Movies, such as Oppenheimer, bring out some of the notions and concepts and problems related to technology and its advances. 

So, to today's theme. One of the links from Wikipedia went to a site at Georgia State which is using an older format to demonstrate ideas with interactive demonstrations via Java and Javascript. To me, that's a great find. Have seen it before. I have pointed to my favorite site which dealt with findings in the realm of mathematical physics and which still is there in a 1995 format. There has been an update which references back to the classical site. 

Like we're doing. Here is the site and its graphic. The site was set up for teachers of High School Physics and is available as an app. 

Einstein
experiment
At the same time, they have a HyperMath site that is very good.

Staying with the theme, there is a new site that represents changes over the past decade with respect to what I call configuration. 

HyperPhysics was featured in PhysicsWorld which offers current information. Their site has a more modern look and configuration. There isn't the same level of interaction. But, we're considering look and feel. Consider that the topic would not be comparison directly but concerns for managing information going forward. 

As a side note, Geertruida de Haas-Lorentz provided an experiment to a French museum. It had been done by her husband and Einstein in 1915 and showed a connection between magnetism and "angular-momentum of electrons" using a simple device.    

Remarks: Modified: 03/24/2024

03/24/2024 -- Tied to to the TGS, Inc. focus on technology: Geertruid de Hass-Lorentz


Thursday, March 21, 2024

Albert and his boys

As in, Albert Einstein and the guys (and gals) of Physics. His ideas were central to the development of truth engineering. The trouble is, what are his ideas? He, himself, complained that once the mathematicians got hold of his model, he didn't understand it. Too, he covered a lot of bases over his time. So, in truth, we will have to track through all of that, over time. 

And space? Look at GenAI (which is really is AIn't as I have mentioned many times since 2021) and how it represents unwarranted "search" through hyper/multi-dimensional spaces. We will look at that closely. This recent work is dabbling with heterogenous stuff and assuming that transforms will bring in homogeneity. But, the work of Einstein dealt with things that were more amenable to being handled in the "canonical" fashion that homogeneity brings. 

Note: That last paragraph points to the essence of the main issue of the cloud which we can (and will) go into in detail (over time and space). However, when we look at the foundational issues being addressed by Einstein, we will see a close stepping forth (dance like) of mathematics and the operational so as to become problematic to the extreme. GenAI is a mere symptom of a huge problem. Time to start to grapple with that. 

All of this plus more is on the table as we bring truth engineering to bear using knowledge based engineering as an initial framework. Let's look at one book that will be of use for this work. It's his book with Infeld (so, E&I) with the title of The Evolution of Physics using archive.org's copy. The authors briefly start with the briefs and come through the usual trek of Galileo and Newton on down the pike. They look at the rise and fall of the mechanical view after which, of course, we get the introduction of "field and matter" that comes from the advent of relativity. They go through the ideas of (and which contributed to) special and general and finally end with a view of the quantum work that somewhat merges (interlaces) with relativity. 

Okay, we all know of the visual arguments that Einstein used (train, elevator, and others). The book has no mathematics, so the authors grunge out explanations using words. It works. On the other hand, there is something to what Feynemann mentioned with respect to not being able to understand quantumness without mathematics. 

I say, mathematics drove the development of the thoughts. We will spend a lot of time there. BTW, if there seems to be a different tone to these posts, that's quite observant. If not, that's okay, too. But, the work to date was mostly surveying all of the thoughts, as expressed over the past few centuries that relate to the themes associated with deciding what's true. That issue became imperative as the computer evolved, but we did not. 

Want an opinion? We have bent over backwards over the past few decades to make the computer look good. Oh yes. The biggest argument for this? The descent brought by misguided development of the web and its muddy cloud which was then fed into ML (machine learning) in order to become some sort of omni presence or know-it-all or what have you. How about overall jerk (3rd derivative, if you're an engineer)? 

But, back to E&I. I have been through the book several times. Why? I had it for some time and mostly dabbled (as Koestler mentioned with respect to library angels). Of late (blame GenAI's rise and verge toward a fall), I took the time to step through E&I's logic. Plus, having used GenAI to probe somewhat into the workings that were hidden, the imperative nature became apparent. So, we'll get through the whole of it. 

For now, I just realized that E&I mentioned television. The book is about 1938 in emergence to public access. E died in the 1950s. The book was republished. I said that he looked to make changes to bring it up to date and only did a few. So, the book stands as a point in time reference to needed discussion. 

See pgs 189 and 190 for the discussion of an inertial frame and its clock. E&I lay out how to pick some central position and try to synchronize. Remember, this was from the 1930s. But, there were rudimentary sets which showed the concept. Now, jump ahead. We have the internet with its cloud and more. Too, we have sophisticated clocks that can sync across the country. When we watch a game in the U.S., everyone seems to consider that the variability is minor with respect to delays and such. After all, consider the work put into supporting real-time streaming. 

It was good enough for on-line (say Zoom) meetings during the pandemic. People walk with phones to their ears (some drive, too) and seem to think that they're having a real conversation with little delay in the signals. Oh yes, for those who may want to know, we'll include some technical sidelines. 

For instance, the cloud? It's a bunch of services that are provided via distributed (some very much far apart) servers running synchronized  algorithms using high-speed connections. Oh wait? Everyone knows of Nvidia and its $trillions evaluation according to the ca-pital-sino (my neologism - see Fedareated). Right? That whole affair handles the relativity issue quite well, at least ostensibly. You know, we could be picky (and might due to the importance of the topic) and point out issues.  

But, that's later. People need their games even if it's like gambling and pulls money from the pocket and one's family needs. Same goes for GenAI which was done without due concern for things that are important. Don't worry, we'll get there. 

No, remember, E&I are not the only players in truth engineering. We're targeting that due to the interest in things quantum on the part of the computer people who like the ca-pital-sino and seem to want to have control more than is necessary. 

People's rights are being trampled with insidious algorithms. But, then, it's early yet. Will there be improvement or further decline. 

So, with that digress, let's put this off until tomorrow or after. 

Remarks: Modified: 03/21/2024

03/21/2024 -- 

Saturday, February 3, 2024

Assessments, in general

We have done this blog, in concert with 7'oops7, since 2007. We started Fedaerated in 2009. Then, in 2010, we started the TGS blog (last post, Technology assessments ...). For the Thomas Gardner Society, Inc. (TGS), as well, we had some publications (Gardner's Beacon). In 2015, we went over to Quora and have been there since. Various excursions were related to using other modes, such as WordPress. 

But, we're back to Google and its Blogger. Why? Of all of the xnn/LLM systems that we looked at (granted, the survey was not exhaustive), we liked Bard the best. So, we'll work in that environment. At the same time, we'll touch all of the to other resources on a regular basis. It's just that we'll use this for publishing where we want to support mobile devices. 

Of late, we have tried LinkedIn's approach which we look at below. There we published the first of a series. We just completed the third piece what is expected to be five pieces. What we call AIn't is the motive. Truth engineering has been on our table for a long while but was problematic. What did it mean and for what motive is it necessary and why just this? 

2023 brought the answer to the questions. Answers have been embedded within these posts all of this time. So, the effort was right on. Let's just say that I was being exhaustive in a way that can only be understood in the academic sense as being highly multidisciplinary. I was retired and had the time and the interest and the ability. 

There is no hubris in that comment; rather, it's a nod to a continuation of what I was doing anyway - advanced computional systems - without having the time to be thorough. So, it's time to pull all of the factors together so that we can analyze and perhaps predict and propose changes. For me, being successful at the complete survey of the western civilization's influence on the world is enough of a motive. However, the world is large, so we need to bring into the discussion all other cultures. 

Notice that the TGS is based upon the history of the U.S. as an offshoot of the U.K. Be that as it may, we know that New England was not alone. Too, we know the prior history of these regions. Albeit, lessons learned or not over 400 years require attention. BTW, the U.S. started 250 years ago, almost, so that event will get regular mention. 

Okay, after seeing ChapGPT and noting the issues, I began discussion with Larry Walker who had run the Knowledge Systems Center for Sperry Univac for which I worked. The focus on "knowledge" was strong for both of us. That was not what we were seeing with the machine learning systems being pushed out. We can discuss that at any level. For now, these posts will be setting the stage for discussions that are pending and necessary. 

I became aware of ChatGPT in February of 2023 which was late. I wrote several posts on it and its cohorts and other topics that relate. But, in December, I published on LinkedIn the 1st of a series. It was also presented in PDF and at WordPress. The title: Artificial intelligence, not solely machine learning (AI, not solely ML). The 2nd of the series continued the theme: Knowlege and truth. 

Neither of these seemed to be of importance of the work that went into xNN/LLM. Why say that? It's been by career to work with these issues operationally (real time, industrial environments); at the same time, my private work (autodidact, by nature) covered the bases with respect to what's at play and of scope. Essentially, "theoretical" is my middle name. 

I just finished the 3rd of the series: Physicalness and mathematics. There will be one more (4th) that deals with the emergence, and surgence, of machine learning during the past two decades. As well, we have to look at data and decisions as a key topic with regard to computing and technology. That is, people, in general, need to know. Experts? We'll address their issues, too. After the 4th, we'll do a 5th which will deal with "What's next?" with all of this stuff. 

The series is based upon experience in KBE as it functioned as the basis for truth engineering. So, it's more oriented toward those operationally involved with engineering and science. I'll turn around and write this series (condensed) for the general public. 

So, here is the series: 

Notes: 

* "truth engineering" coined by David E. Jakstis in discussion of a white paper by John M. Switlik on their joint work with computational modeling for fabrication of forgings and castings. 

** This is open to public read. Some Linkedin pages may require account.       

--------------------------------

In tems of history, the U.S. D.O.D. supported early work in artificial intelligence. As well, data science was a common program. It is of interest to future discussion to point to the recent guidelines for digital enginering. We have made reference to "digital twin" which is an important concept that must be brought into the discussion related to advanced computing. We will have this document a major piece of our bibliography. 

Digital engineering

Remarks: Modified: 02/04/2024

02/04/2024 --  Linked this to 7'oops7 (AI, not ML solely) which will bear directly on the 777 project underlying KBE which relates to truth engineering as AI when it matures. Notice: Shattered dreams


Monday, January 1, 2024

New Year, 2024

Notice the start of decline in our activity here when we moved to Quora (2015). 

Here is a list of posts. 

We will be active in 2024.  











Remarks: Modified: 01/01/2024

01/01/2024 --