Discussion Locked

3 DIMENSIONAL DATA STORAGE

0 helpful
23 replies
Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

Research is currently ongoing into a concept of 3D data storage and is intended to replace current mechanical methods of storing data. The ole spindle and platter may become an artifact in the Smithsonian next to the phonograph.

The entire concept centers on holographic principles. Under this theory data is stored when concentric propagating spherical waveforms collide forming a constructive / destructive wave pattern that is then 'gathered' and imprinted on a recording surface. Typically in this process highly polarized light (laser) is split into a reference beam and a data beam. The data beam is reflected off of some object then joined with the reference beam causing the interference pattern that is recorded.

If developed it may be possible when combined with asynchronous processing that highly intelligent synthetic life may come to fruition.

Currently, I am trying to visualize how this data is stored in the interference pattern. Does anyone have a good explanation?

23 replies

by fluxit

It appears that the data is not stored in the traditional sense of information attributes about the object. Instead the essence of the object is stored. This requires a paradigm shift in thinking.

So where in the inference pattern is image data kept and how can portions of that image be recovered? For example that the apple is red and shaped like a cardoid.

by mrafrohead

And I've tried to ponder it. And I developed a serious headache.

I hate to say it, but as of this point in my life, I'm not ready to absorb this type of knowledge. But I do hope in the future it will make sense to me.

It is a SUPER interesting idea though, don'tcha think?

by fluxit

It is so fascinating and has enormous implications across a vast sector of our world. It impacts our understanding of religion, consciousness, and fundamental understanding of how the mind works. I can talk to this but I am more interested in how data is stored right now.

It requires a quantum shift in thinking away from the current discrete systems. I am attempting to study the mathematics behind it in order to understand how information may be stored in it. The math centers on wave physics.

I understand that it has some limitations. Data is not stored as object attributes as in current database technologies. Abstract information may be difficult to store. However, the essence of a physical object can be stored and recovered somewhat like a memory or visualization in the human mind. Hence, this kind of technology most likely would require a nueral net to process the stored image.

I can envision a nueral net organized in a virtual manner having 100's if not 1000's of processors installed on a backplane and logically organized through software to process the stored images. So if some one was to 'ask' what color is the apple - the neural net would retrieve the image of an apple and process it for color. So a 3D data storage draws its strength from symbolic logic rather than digital logic.

by dhuwrees

Simplest description I have heard to date is that it's like punch cards, but at the atomic level.

by bayani

Multidimensional storage systems are of particular interest to Conceptual Spaces theory. I am currently beginning on the road to my PhD studying this topic. With the current 2 dimensional storage systems, we need to figure out some really nasty and inefficient methods using some sort of search tree system to store multiple dimensions. (e.g. Red, Green, Blue would be 3 dimensions to form a super dimension called colour. Groups of colour in different vectors would then form shapes etc.) Currently, the system that I can see requires about 38 dimensions to get basic functions such as vision based object recognition, and sound based object recognition.

Since all of this information is essentially stored in a one dimensional array, being a hard disk, search trees such as Quad-trees and B-Trees are interesting to us, as this allows for quick searching, but takes a long time to store given a nasty situation.

Further problems also occur when some dimensions are not present. Ie, have X, and Y, but not Z axis information. This takes the function from a Log(MN) function (where N is the number of items stored, and M is the number of dimensions) to a X^2Log(MN) function. (X being the number of dimensions not present)

All of these problems would have a much simpler solution with a 3D storage approach, as you could model a basic Conceptual Space Crystal from the center of the drive, with each dimension being stored at certain distances from the centre of the structure, at a certain angle.

I will be rather interested when we get true storage crystals to be able to store our data systems.

by sleek1

Can you imagine the kind of operating system it would take to govern 3Dimensional data storage? Current windows fault tolerance is somewhere in the neighborhood of 125,000 f/s. I cant imagine anyone ever writing an OS that could handle it. I would compare it to "moletronics", or processors built on the principal of (+-)charged molecules that can be "switched" on and off along their molecular pathways relative to current. It would literally require an OS that could handle MILLIONS of faults per second just to run in a quasi-stable manner. It doesnt matter that we have actually "grown" these processors, 1,000's of which might fit on the head of a pin; the bottleneck is simply fault tolerance. So it would seem for 3D data storage. (Seems like it just might require rethinking the whole OSI Model too, huh?). Paradigm shift is right...

Sleek1~

by fluxit

You seem to point to conventional storage methods applied or adapted to multi-dimensional data storage. I would find it difficult to apply branching or rectangular methods.

I did explore spherical storage approaches myself as you described with the Space Crystal. It would be ideal for scientific storage where complexity is consistent at each order of magnitude. For example, close to the center information would be organized about the sub atomic level. As one moves outward information would be stored on the higher orders of magnitude. Locating the data would be addressed by angular sweep, angular rise, and a linear run (radius).

However, I believe that in order for true multi dimensional storage to be effective it has to use holography coupled with nueral nets and a 'sensory' web. In short, it has to store an image - not attributes - and then use a nueral net to assess and wein data from the image. The sensory web would be used to extract pertinent information from the image.

by fluxit

In holographc storage systems complete information can be recovered from any portion or fragment of the storage surface.

I do not see the need for an user OS being designed for multi-dimensional storage systems. They are 'plug and play'. The internal workings will require some sort of OS and code but that is hidden from the user. The device would be self contained.

by knannest122

In my opinion it's always need for fault tolerance for important data, both scientific or financial.
It seems as if you are referring to data stored
in cubes. If those cubes breaks, how will it be possible to recreate the cubes. I know that it will be possible to read fragments, but what about
the data stored at the areas where the cube breaks. We have the redundance in raid systems, also hidden from the OS or end user.

by fluxit

The human intelligence is fascinating. The brain is the physical operation of thought. There are various types of neurons; sensory, motor, inter-nueral. There is a maintenance feature built in that is a watery solution full of cells known as GLIA. The brain itself has many folds and its surface area is immense. Apart from the physical is the mind where conscious thought occurs.

The thinking process begins with the emission of a calcium ion into the GLIA that causes a ripple. If the holographic principles apply that ripple may collide with others causing an interference pattern that is some how imposed upon the surface of the brain. Sensory devices (eyes, ears, nose, tongue, and skin) influence which nuerons fire to cause the correct interference patterns to form and 'burn' into the brain.

If a 3d storage system can be built to behave in a like manner it would be incredibly awesome.

by bfilmfan

I recall reading an article recently (I want to say it was in Science) where some researchers were speculating that the mind may arise from quantum effects. I will see if I can't dig the issue up and post which one it was.

by fluxit

is something different. What they were attempting to explain is how consciousness is attached to the biological. What that theory suggest is that there are these microtubules made of protein that possess the quantum property of superposition. These tubules become unstable and collapse causing what is termed as a 'stream of consciousness'. The theory is not well accepted among most scientist but is interesting.

by dwiebles

If I understand the basic principle here, Your reference beam is a constant accross all platforms. The interference pattern that is stored is a combination of the actual "data" beam and interference beam. Here it becomes not unlike a math problem with a single variable. If you have the "reference beam" as a constant, and the sum of the beams, or the "interference pattern", you can thereby subtract the constant from the sum for the original data. In some tests storing a single beam with no reference returned a high percentage of loss, but adding the reference beam with the interference/subtract method is yielding much higher results. Isn't the dicovery channel great! :)

-DW

by fluxit

suggesting that the information can be recovered using fourier analysis? In electromagnetics there are a set of mathematical formulas used to model behaviors. Fouriers is one such set.

This suggests that data stored is a function of amplitude. It could also be possibly a function of varying frequency too but amplitude appears to be a more correct solution.

by ivanong

HUH ?

I thought it was a cubed/massed object. say for the sake of this discussion 1 x 1 x 1cm "square". Whereby each "spot" of its mass is mapped in the X , Y , Z cordinates. Therefore each "spot" can be magnetically charged to I or O.

In this scenario the 3.5inche platter + Housing can be scalled down to 1cm or less object. Which would probably be easy to house and does not need any motor spindle or much power to drive it.

A caveman would probably go about producing such an product by : Scrapping the surface layer of all his HDD and then chopping them up to bits (taking care to put the correct surface at the top of course) then supergluing them up to a stack - the eventual product will probably be about the size of a eg. 3mm/Sq cube ?? (Pardon me here , but the technology at this point becomes a blur to me but am sure those guys wo invented the MRI in hospitals can think of something.) - use the freakin MRI (size 8 x 5 x 7 feet) gizmo to charge/retrieve the data. -

Now you know why they call me the Caveman.

by fluxit

Your viewpoint is quite unusual.

Current computers do not actually use a rectangular system to store data. They operate on a frequency having amplitude (radius) and oscillating signal (angular reference). In the end it is a spherical coordinate system that is then computed into rectangular coordinates for conventional data storage. The computational time slows the system.

Gains can be achieved by dropping this and using spherical coordinates directly to access data in the storage device. Changing storage device technology completely can gain even higher results.

by bayani

Yes, using neural networks are great for certain applications, but my research is showing that a layer of simplified abstraction is more likely to produce a better result, with lower processing power.

Quite an interesting way of looking at things. You really need to read Peter Gartenfos's Conceptual Spaces Theory to see what I'm talking about.

by fluxit

on Conceptual Space theories. It appears to me that it really is a sophisticated way of discussing fuzzy logic. The systems methods of discerning and mapping cross spatial correspondence is to use fuzzy logic mathematical algorhythms in the nuerons on a nueral net evaluating the information from receptors contributing to input space (input level of a neural net). Processed information (blended space) is then used to further evaluate information at the higher levels of neural processing.

I have never viewed neural nets as purely discrete processing of a decision tree. In fact, I always viewed them more in the context of this Conceptual Spaces Theory. I guess one could look at neural nets from one of three view points; A discrete system, conceptual methods implemented via software, or a complete unit having hardware backend and software components. I tend to view the neural nets in the latter.

The inputs of a neural net are a constraining point. Not until more sophisticated sensors are available will neural processing be able to achieve appreciable results.

I started a discussion on nueral nets earlier, "NEURAL NETS AND PROCESSING". I would like to get some discussion there.

by fluxit

can be done? The rectangular or stacked memory is currently in use. The spherical coordinates are used in very high end systems where response times are critical.

Lower cost spherical storage systems are in need. But neither are necessary if holographic systems are employed.

by jhansen

holographics are meant to store a 3D image on a 2D surface. To follow this analogy to a 3D surface you would need to store a 4D image to be considered holographic. Since the 4th Dimension is considered to be time, A 3D holographic memory core would have the added advantage of saving the state of memory over the course of time. To carry this out to its logical conclusion, you could litteraly (REWIND, like TIVO) your computer, all the processes, and threads to an early state and then play back in slow motion, or in reverse. I don't know what kind of advantage this could have as i don't see how it could make computers faster.

Jake
---------------------------------
Computer Techs, earn residual income and start your own online backup business today.
http://www.zray.net/AffiliateProgram.html

by fluxit

There are mathematical algorhythms that permit multi-dimensional space to be mapped into fewer dimensions. The process is called conformal mapping. For example, a straight line in 4 space mapped into three space forms an ellipse. 10 dimensional space conformally mapped into 3 space forms spheres and membranes. So the number of dimensions can vary without affecting or attentuating the data.

Your example suggest that the CPU processes are stored in the holographic memory. If do not see a need for that if the data stored is the image and information about an apple pie. However, to assess that information would require a sensory net and nueral net.

This discussion is closed to new replies.