The following R&D paper is a major part of the research into the Physics of Thought. ©Copyright 1978 - 2002  Advanced Research Consultants, Inc.  P1-3, P4-6, P7-10, P11-16, P17-20

We can select many more read directions.  We were working in the X-Y plane which required 3 bits to specify any of the 8 (000 thru 111) read directions we used.  Now we add a bit for the X-Z plane and one for the Y-Z plane.  If we add one bit for several cubic directions, not in the planes mentioned above, we have 3 more bits for the Read Vector.  This brings the total to 6 bits for the Read vector (2⁶ = 64 read directions).  Now we see that the address of a 256-bit word is specified by a total of 30 bits.

Suppose we have a computer generated ideal distribution of “1”s and “0”s throughout our cube (256 X  256  X  256) as we had in our 4  X 4 model above.  (Later when we move to n-dimensions, we will see that this is unnecessary because our computer could as easily generate the reduced logic equations directly from the data being stored.) We could have a computer search our cube for a specific 256 bit word and give us the address of that word.  Using the result of such searches we can perform TAG (address linked) storage.

Consider the following method of TAG storage.  Given a large amount of data to be stored we take the last 226 bits (256-30 bits for the address) of the data, add a 30-bit end marker, search our virtual cube until we find it.     

Word 1

The search produces the address of word 1

This address is attached to the next 226 bits of data to be stored.  Our word format is:

Repeating this search and attach process we have the following picture.

Word 1

©Copyright 2002 Advanced Research Consultants, Inc.                                             Page 17 of 20 pages

Word 2

Find address of word 2 and attach the next 226 bits of data to compose

Word 3

Repeat until all data is stored.

.

.

.

Word n

Leaving us with

This Word n Address is in fact the address of all of the (TAG stored or linked) data.  Feed the logic Word n Address to retrieve Word n.  Strip off the 30-bit address of word n-1 and feed it into the logic circuit to retrieve word n-1.  Send the data to the output cache.  Repeat till end marker is reached. This whole retrieval process suggests the metaphor of pulling a little thread (address of word n) of a sweater and unraveling the entire sweater.  Except that in our case the sweater (data) remains intact since we can repeat this data retrieval as often as we please.

Skepticism jumps in here.  Is this like some perpetual motion machine schemes in that it can’t work because something fundamental has been overlooked? ---  This can’t be an infinite capacity memory. ---  First, this thing can’t be bottomless since there can only be 2³⁰ = 1,073,741,824 (over 10⁹ ) unique addresses using a 30 bit address.  What is our fancy computer supposed to do when it can’t find a particular 256-bit word? Or finds an address that has already been used (produces computer retrieval execution loop)?  Create another dimension with a different distribution of imaginary “1”s and “0”s that specifically includes the word being sought?  Execute twists and turns through our cube?

First, the creation of another dimension (4^th) adds another 8 bits to the Start Coordinate part of our address.  This would bring us to a 38 bit address thus increasing the number of unique addresses to 2³⁸ = 274,877,906,944 (over 274 X 10⁹).  Another dimension (5^th) would bring it to 2⁴⁶ = 70,368,744,177,664 (over 70 X 10¹²).  The progression follows:

©Copyright 2002 Advanced Research Consultants, Inc.                                             Page 18 of 20 pages

6^th dimension = 2⁵⁴ > 1 X 10¹6; 7^th 2⁶² > 4 X 10¹⁸; 8^th = 2⁷⁰ > 1 X 10²¹ etc.  Remember, all our computer is doing is calculating the logic equations required to produce the pattern of “1”s and “0”s it has to store.  The data retrievable by execution of the logic (directly in a computer or via ‘burned into’ ROMs or via manufactured chips containing the logic circuits).

Secondly, our computer can be programmed to create logic (address) that includes turns every 16, 32, 64 or 128 digits.  Using 16 (groups of 16 bits in 256 bit word) times a 6-bit read vector (64 read directions - read 16 bits, change to 1 of 64 read directions to read the next 16 bits) uses 96 more bits in an address.  This would reduce the data part of our 256 word from 226 to 130 bits.  It would also create 2⁹⁶ = 79,228,162,514,264,337,593,543,950,336 > 79 X 10²⁷ additional unique addresses.

Together we have 2³⁰ X 2⁹⁶ = 2¹²⁶  > 79 X 10²⁷ X 10⁹ = 79 X 10³⁶ unique addresses in 3 dimensions.  If we now extend to the 8 dimensions, we have:

 This would bring our unique addresses to 2⁷⁰  X  2⁹⁶  = 2¹⁶⁶ > 10²¹ X 79 X 10²⁷ = 79 X 10⁴⁸.     

There is only increased logic complexity associated with our increasing the word size to 512 or 1028 bits.  If we used a 1028-bit word with a 256-bit address we would have 2²⁵⁶  = 2¹⁶⁶  X 2⁹⁰ > 79 X 10⁴⁸ X 10²⁷ = 79 X 10⁷⁵ unique addresses.  Similarly we are not confined to the model I’ve used to introduce this concept of folding information into Information Space via logic circuits leaving only a single address.

©Copyright 2002 Advanced Research Consultants, Inc.                                             Page 19 of 20 pages

We are approximating infinity here and we can get as close as we please.  What’s the catch?  How do we write a computer program to develop these “ideal” distributions of “imaginary” “1”s and “0”s in hypercubes of n-dimensions?  How do we write another computer program to search these hypercubes for specific series of “1”s and “0”s that doesn’t take forever to execute? 

Skipping all the proprietary work going on in Genetic Programming (the computer program development via evolutionary schemes technology).  We can already see such super memory systems in the real world of living animals.  Some animals can remember pictures and sequences of pictures. 

Then there is the information stored in the living cell.  Not only the “program” for the growth of cells into specific structures (Cats don’t give birth to dogs, etc.).  But there is also the “program” in the brains of all animals (Cats act like cats, not dogs).  A chick pecks its way out of its egg, says “peep peep”, walks off, pecking and eating food.  None of the neural activity producing those actions is random or accidental.  (In humans we refer to these programs as “instinct”.  The crying/sucking/yawning programs in babies; babies crawling, then walking.  All the while the babies “minds” are spring-loaded in the record position.)  If we were designing robots, we would refer to these programs that come with the robot animal as its Operating System. 

=====================================================

temporary end

=====================================================

Notes:

. . . requires time to store (create logic) . . . while we’re sleep? . . . why we sleep?

Further, d.

.

.

.

©Copyright 2002 Advanced Research Consultants, Inc.                                             Page 20 of 20 pages

UNDER CONSTRUCTION --- TO BE COMPLETED