Designing and manufacturing microprocessors is normally an onerous, exacting operation for people schooled in electrical engineering (EE). Getting EE Bachelor’s and Master’s degrees is a tough thing to accomplish.
One must acquire a thorough understanding of the underlying solid state physics, discrete mathematics, differential equations, semiconductors, advanced C++ and assembly programming, electromagnetic theory, Very-large-scale integration (VLSI) design; and a smattering of ordinary analog circuit analysis and design that imparts knowledge of the “surround stuff” on a computer’s motherboard.
That said, it’s downright astonishing that researchers of the MESA+ Institute for Nanotechnology and the CTIT Institute for ICT Research at the University of Twente in The Netherlands have managed to coax a tiny disordered pile of gold nano-particles into behaving like a highly efficient network of transistors, using a process similar to Darwinian evolution.
Writing in the Sept. 21, 2015 online edition of the journal Nature Nanotechnology (“Evolution of a Designless Nanoparticle Network into Reconfigurable Boolean Logic”) S. K. Bose, C. P. Lawrence, Z. Liu, K. S. Makarenko, R. M. J. van Damme, H. J. Broersma, and W. G. van der Wiel took a few dozen gold nanoparticles — each just 20 nanometers in diameter — and placed them in a heap with some interconnecting insulating molecules so that each gold particle was about a nanometer from its nearest neighbors.
Eight electrodes were then placed around the clump of particles.
Now, according to physics and logic, by applying just the right amount of voltages at six locations, the pile of gold will act like a network of transistors, owing to the nonlinear behavior switching behavior of the electrified gold particles.
By varying the voltages to the right levels, the pile can act as anyone of six possible “logic gates” that are fundamental building blocks of computer processor chips.
The big question of course is — what voltages should be used? The number and disordered arrangement of gold particles is too complicated to calculate the voltages using conventional techniques.
And it would take too long to try a huge series of random, trial-and-error voltage sets.
That’s where the artificial Darwinian evolution process comes in.
The scientists developed a so-called genetic algorithm to test different sets of six static voltages and register the results. (Yours truly wrote about genetic algorithms and their discoverer, John Holland, in a previous blog posting, “A Software That Evolves Itself.”)
The evolutionary computer program the scientists used treats each control voltage as a virtual “gene” and a complete set of six specific voltages as a “genome,” which would normally be an organisms’ complete set of DNA (that is, all of the genes, in this case a set of six static voltages).
A series (“generation”) of randomly-generated voltage sets (genes) are generated and tested. Some of these sets work better than others, and they are preserved. The classic “survival of the fittest.”
The genome survivors of each generation of tests are copied and then cut into pieces and spliced onto other genomes in a kind of “mating” process to form new offspring, some of which will yield even better results than their parents.
Also, some random mutations are inserted here and there among the genes.
The scientists were searching for six possible types of logic gates, and, sure enough, the genetic algorithm found each one of them after fewer than 200 generations of testing, which took about an hour.
Even more surprising, evolving system came up with a set of voltages that turned the clump of gold particles into a higher-order logic unit, which can add two bits of information.
Unlike conventional human engineer-designed computer chips, such “design-less” self-evolving nano-computers are far more efficient than conventional computer chips, more resembling the configurations of natural neural networks in the human brain.
Researcher Van der Wiel said in the Sept. 23, 2015 edition of the scientific journal New Scientist that while the best microprocessor you can buy in a store uses a few hundred watts, “The human brain can do orders of magnitude more and uses only 10 to 20 watts. That’s a huge gap.”
In their scientific paper, the researchers conclude that the evolutionary approach “ . . . works around device-to-device variations at the nanoscale and the accompanying uncertainties in performance, which is increasingly becoming a bottleneck for the miniaturization of conventional electronic circuits. The results, therefore, also need to be seen in the light of these exciting possibilities."
So, perhaps our future computers will not be designed so much as self-assembled, making them mysterious black boxes of great internal complexity capable of performing wondrous computational feats — with no one really understanding how they do it.
Richard Grigonis is an internationally known technology editor and writer. He was executive editor of Technology Management Corporation’s IP Communications Group of magazines from 2006 to 2009. The author of five books on computers and telecom, including the highly influential Computer Telephony Encyclopedia (2000), he was the chief technical editor of Harry Newton's Computer Telephony magazine (later retitled Communications Convergence after its acquisition by Miller Freeman/CMP Media) from its first year of operation in 1994 until 2003. Read more reports from Richard Grigonis — Click Here Now.
© 2024 Newsmax. All rights reserved.