An analysis of the Betty Hill "star map"

originally posted by: james1947
I thought that it might be more productive to show all of you "how" I got to where I am.

Betty was given the post-hypnotic suggestion that she could sketch a copy of the "star map". Although she said the map had many stars, she drew only those that stood out in her memory. Her map consisted of twelve prominent stars connected by lines and 13 lesser stars that form several distinctive groups. She said she was told the stars connected by solid lines formed "trade routes", and dashed lines were to less-traveled stars.

There are a total of 25 stars in the template with 13 identified by Marjorie Fish around 1968. Much of her efforts were undone by others who knew little about Astronomy, in any case, Ms. Fish’s work has served well to start this analysis.


Computer modeling and Datasets
To create our view of the stars, and make a search for a match possible we created several database tables to use with Microsoft’s SQL server.

DataSets
Hipparcos
The word "Hipparcos" is an acronym for High precision parallax collecting satellite. It was launched in 1989 by the European Space Agency and collected data on over 118,000 stars. There is also an auxiliary table called “Extended Hipparcos” that contains additional astrometric data. These tables are maintained by the University of Strasbourg, Fr.
These two tables were imported into an instance of Microsoft SQL Server Express, and used as the basis for all stellar data.

exoplanet.eu_catalog
A catalog of all known exoplanets. This table was also imported into SQL Server. However the table required some modification to allow it to accurately interface with the Hipparcos tables, and other software.
All of the data elements (columns) imported as text, and there was no record “identity”; so distance and age in a numerical format was added. Software methods were created to populate these new fields with their text counterparts’ conversions to a “double” data type. Additionally the Hipparcos index number was added by creating software methods the access “Simbad” (a web service of the University of Strasbourg, Fr.) to obtain a cross reference where possible.
These three data elements constitute the majority of data used in this query. Programming languages used in this data phase were C#, and TSQL.

Computer Modeling
Two categories of model were generated using a combination of C#, TSQL, and Python.

3D Models
Two major 3D models of local star were created. The first was stars within 33 parsec, the second expanded to 46 parsec.
A C# application was created to apply SQL search criteria (everything < 33 parsec) and create a sub table from which astrometrics of individual stars could be retrieved. Using specialized templates written in Python, the position (X, Y, Z), along with star class, and name were compiled into a Python script for use with “Poser” (a 3D modeling and animating application) to place “Bucky balls” of appropriate color, and name into the 3D space provided by Poser.
As I was using the Fish interpretation as a base of sorts I included additional instructions to highlight the stars in her “list”. This allowed me to position a “camera” somewhere that was logical to view the “Fish Stars”. It was incredibly easy to find such a location, knowing approximately where to look from. As I was refining my view I noticed that I had gone outside my 33 parsec radius. I increased the radius to 46 parsec so that I might know where I was viewing from. After adding the additional stars I found that I was near a star called “HIP-26737”, yet another class “G” star in the mix (actually G2V).

2D Models
Methods were also developed to provide 2D Models of individual Star Systems. These method produce a “PNG” file that expresses the star and the orbits of any planets along with a representation of the Stars Habitable Zone. These methods were written in C# and SQL for the Windows Desktop.

Discoveries
There were some discoveries found in these models:
1. Marjorie Fish was spot on with all stars but Kappa Fonacis. I haven’t built a model based on the Gliese dataset, so it is entirely possible that better “location” of the star moved it, what is actually, a considerable distance. And, based solely on the stars identified by Ms. Fish; her interpretation isn’t really a very good match, but, it did serve as a good starting point.
2. The identity of 13 stars that Ms. Fish didn’t identify were identified.
3. Among the newly identified stars were 5 that were not known in either 1961, or 1969; so neither Betty, nor Ms. Fish, could have “guessed” their existence and location.
4. The doubling and connecting of the two Zetas Reticuli is suggestive of a Graphical User Interface. It appears that if one were to “mouse over” or touch Zeta Reticuli; it splits into the two stars of the overall system. Just as doing the same to other stars has the same result. And of course doing that to a multi-star system might show individual solar systems.

"Alas, it is probably more complicated than that. Based on obsolete data, Marjorie Fish’s interpretation of Betty Hill’s map has been shown to be wrong. In the early 1990s the European Hipparcos (“High precision parallax collecting satellite) mission measured the distances to more than a hundred thousand stars around the Sun more accurately than ever before. Some turned out to be much further away than previously thought. Other research has looked at stars included in Fish’s research. Two, 54 and 107 Piscium, have been revealed to be variable stars, while Gliese 67 and Tau 1 Eridani are in fact close binaries. Then some stars discounted by Fish have turned out to be potential abodes for life after all, for example Epsilon Eridani is not after all a binary star. Using Fish’s own assumptions and more up to date data, six of the fifteen stars chosen by her must be excluded."


more at :
armaghplanet.com...

originally posted by: james1947
While the two performed differently, and used different algorithms, they did agree on the match, and it's quality.

But what about a human match?

After all, humans are relatively good at pattern matching, and a numerical approach to a match, even if 99.99% correct, may be too different when a human looks at it.

I don't know how those two algorithms work, but if they consider, for example, a 99.9999% match on the average of all stars, that only means that the average is a match, not the individual positions, and, in this case, we are looking for several individual matches, not a generic match of several stars.

originally posted by: joelr

"Alas, it is probably more complicated than that. Based on obsolete data, Marjorie Fish’s interpretation of Betty Hill’s map has been shown to be wrong. In the early 1990s the European Hipparcos (“High precision parallax collecting satellite) mission measured the distances to more than a hundred thousand stars around the Sun more accurately than ever before. Some turned out to be much further away than previously thought. Other research has looked at stars included in Fish’s research. Two, 54 and 107 Piscium, have been revealed to be variable stars, while Gliese 67 and Tau 1 Eridani are in fact close binaries. Then some stars discounted by Fish have turned out to be potential abodes for life after all, for example Epsilon Eridani is not after all a binary star. Using Fish’s own assumptions and more up to date data, six of the fifteen stars chosen by her must be excluded."

Actually, I covered this some time ago, but...
54 piscium: Has at least one planet, AND a brown dwarf companion. Back in the day that probably meant "variable". Th is is due to the fact that "variable" really means that on subsequent observations a different luminosity was observed. Did you know that this is one way of detecting planets and companions?

A planet passes in front of it's star and causes a change in luminosity. Much more is required to confirm a an extrasolar object, but that is how it starts.

Officially 54 Piscium is only "suspected" to be variable.

107 Piscium: We know very little about this star, it's "variable" status is: Suspected.

Gliese 67: Yep, It's a binary, so is Zeta Reticuli. Gliese 67 is a spectroscopic binary whose companion is a red dwarf with 38% the mass of Sol. It orbits with a period of just under 20 years, and approaches as close as 4.2 AU (that's 4.2 times the distance of the Earth to Sol). It is also on the HABCAT list of stars likely to have life.

Tau(1) Eridani: This a spectroscopic binary with an orbital period of 958 days. There doesn’t seem to be much on Tau (1)’s companion other than its long orbital period, however with the length of orbit, it must be located a respectable distance from the primary star, and thus may have little effect on any inner planets, should they exist.

Much of this wasn't known when those statements were made, we know better now; none of Ms. Fish's stars should be excluded.

Any way, y'all should have read the White Paper , it addresses those issues and more.

originally posted by: ArMaP

originally posted by: james1947
While the two performed differently, and used different algorithms, they did agree on the match, and it's quality.

But what about a human match?

After all, humans are relatively good at pattern matching, and a numerical approach to a match, even if 99.99% correct, may be too different when a human looks at it.

I don't know how those two algorithms work, but if they consider, for example, a 99.9999% match on the average of all stars, that only means that the average is a match, not the individual positions, and, in this case, we are looking for several individual matches, not a generic match of several stars.

Well, ArMap, Humans can be influenced by emotions and psychology, as well as bias and prejudices, computers fall prey to none of these. Computers are "cold and calculating" and don't care about the outcome.

We've seen people right here in ATS that will allow their personal ideas, notions to affect what they see and how they see it.

But, as to the "Human" match; in my paper there is a whole 'section' dedicated to a "Human visual" approach. I identify image elements called "blobs" and show that those blobs exist in both image samples.

On the algorithms used. There were two, each using a different 'base' algorithm and a different set of metrics. Most of this is "hidden" inside a "black box" that is the "library". Now, it not all hidden and 'black' since the original source code is available. I have to ask this; "Have you ever tried to understand someone else's C++ source?"

I haven't looked at the original source. Mainly because; I don't care "how" it works, only that it does. Both of these libraries are as close to "industry standard" as a method/algorithm can get without an IEEE specification (unfortunately such a real standard doesn't yet exist). Anyway it is these libraries that are being employed by industry, law enforcement, government, and many others for almost exactly the same application as mine...matching part of an image to a smaller image/template.

originally posted by: james1947
Well, ArMap, Humans can be influenced by emotions and psychology, as well as bias and prejudices, computers fall prey to none of these. Computers are "cold and calculating" and don't care about the outcome.

That's true, but it's not a reason to consider it "less" than a computer version, specially if we do not know what the computer is matching.

We've seen people right here in ATS that will allow their personal ideas, notions to affect what they see and how they see it.

And we have seen people that do not allow their personal ideas or notions to affect what they see and how they see it.

But, as to the "Human" match; in my paper there is a whole 'section' dedicated to a "Human visual" approach. I identify image elements called "blobs" and show that those blobs exist in both image samples.

Why not post that part here? You probably know as well as I that most people do not follow links posted on a thread, so just by saying that you are not helping your point of view.

This is discussion forum, presenting reasons to keep the discussion alive is better than just say "read my paper".

I have to ask this; "Have you ever tried to understand someone else's C++ source?"

Yes, I have even tried to understand someone else's assembly.

Anyway it is these libraries that are being employed by industry, law enforcement, government, and many others for almost exactly the same application as mine...matching part of an image to a smaller image/template.

The problem is that "almost": what are the differences? What we need in this case is not matching part of an image, is matching the relative positions and distances between several points. If you are using the wrong tool you will get wrong results.

originally posted by: ArMaP

That's true, but it's not a reason to consider it "less" than a computer version, specially if we do not know what the computer is matching.

I don't Human Vision "less" than Computer Vision. What the computer is matching is an image, a template to a larger image. Not unlike "Find Waldo"; we have our template: an image of Waldo, and a virtual "sea" of other faces "like" his...the algorithms find Waldo...although it can take minutes. Then it reports a binary indicating IF a match is found, a set of coordinates and a size; to indicate where the match is found.

Why not post that part here? You probably know as well as I that most people do not follow links posted on a thread, so just by saying that you are not helping your point of view.

This is discussion forum, presenting reasons to keep the discussion alive is better than just say "read my paper".

Copyright issues; I don't like the copyright agreement ATS uses. IF I keep my work on my own site ALL of my copyrights are preserved, If I post it on ATS it becomes public domain whether I like it or not.

So...it's only a click and a few seconds...

Yes, I have even tried to understand someone else's assembly.

Assembly is easier! (that was my first "favorite" language...for the 8080 family)

The problem is that "almost": what are the differences? What we need in this case is not matching part of an image, is matching the relative positions and distances between several points. If you are using the wrong tool you will get wrong results.

No, all we are given to work with is a "part of an image". Betty's original drawing was 2D, so we are restricted to 2D. A 3D pattern matching engine does not exist to the best of my knowledge, though I suppose a Machine Learning system might be up to the task...but, I can already hear and see the objections; after all Machine "Learning".

originally posted by: james1947
I don't Human Vision "less" than Computer Vision.

I only said that because you act as if the results from the algorithms are more reliable than the results of people that look to the "match" and do not see it as a match.

Copyright issues; I don't like the copyright agreement ATS uses. IF I keep my work on my own site ALL of my copyrights are preserved, If I post it on ATS it becomes public domain whether I like it or not.

I don't think it's like that, but I'm far from an expert on copyright laws.

No, all we are given to work with is a "part of an image". Betty's original drawing was 2D, so we are restricted to 2D.

There's no problem with a 2D representation of a 3D space, after all that's how we see it. Because of the distances involved we are really just seeing something like a 2D projection of a 3D scene, and also because of the distances parallax its not likely to affect the results.