Is Ternary an option for a tech future?

Question

I have an idea for a story set in the near future, with a character who breaks onto the global scene with a breakthrough advance in technology that makes him billions.

My question is whether or not Ternary computing (as opposed to Binary) could be used as this breakthrough. I don't know much about it except that (apart from a few experimental computers) it never took off despite possibly being more power efficient or something (I don't understand a lot of what smart people do).

Without the need to explain too much about how the technology works in the story, would it be plausible for Ternary logic to be the base of this tech revolution?

Answer 1

Walter Tevis wrote "The Man who fell to Earth" with the alien protagonist turning into a billionaire by releasing advanced technology on Earth, but this was done with a long term purpose: to build a spacecraft and bring his people from their planet to Earth.

Tevis didn't dwell too much on the sorts of technology that were being introduced, since the story wasn't about the technology, per se, but rather the struggles of the being with the various temptations available on Earth, and how they eventually derail his plans.

So focus on the story, unless the nature of the invention is the centerpiece of the plot.

Answer 2

Ternary itself, no.

But some other revolutionary improvement that necessitates a ternary computer, yes.

Possible candidates:

Optical computing - high speed

More accurately; low latency optical computing.

We transmit data via optical cables today, but we still do all our processing electronically. Converting photons to electric signals, doing something clever to make useful electric signals, converting them to photons and sending them on their. This takes time. Not much, but it happens often enough to be an annoyance.

Doing calculations on the photons themselves reduces the processing-delay.

Optical computing - massively parallel

One of the current big changes in computing today is the push from sequentially doing things (very quickly) to doing things simultaneously (even if it's slower).

When you run a silicon circuit really quickly it'll struggle to switch on/off in time(1). Typically, you solve this by turning up the voltage to make them reach switch voltages faster, but this means they get hot and melt.

But designing circuits that do many things at once is hard. We've been working on it for decades and we're still not very good at it. The first person to be good at it is going to be very rich indeed.

A combination of the above - large-scale neural networks

The discrete set {-1, 0, +1} looks a bit like the range [-1, +1]. That's the range of the Sigmoid Function. The Sigmoid is often used in artificial neural networks because of its similarities to the behaviour of real neurons.

This doesn't automatically mean you have Artificial General Intelligences running about the place. This doesn't necessitate Skynet or the Robot Apocalypse. Consider that the software. What your character has developed is the hardware.

Things that are more reasonable uses for a new, powerful, ANN; self-driving cars, market-trading software, better weather prediction (you'd be surprised how much money you could make by accurately predicting something that's assumed to be chaotic).

(1) I'd provide a citation for this, but my Google-Fu is a bit lacking today. I learned this while at university.

Answer 3

No, not really.

While it is possible to speculate a technological advance that would make ternary efficient and a problem domain where ternary is better than binary, the economics would still not make sense. While ternary increases information density, it does it in a way that is fairly easy to emulate in binary logic. Two bits actually have more information than a trit does. Dedicated binary hardware for emulating ternary is quite viable if demand exists. Which it so far has not.

And it is difficult to imagine an implementation where ternary logic is not significantly more complex than the binary logic with same computational capacity. It is not impossible, but it is unlikely as most miniaturized (which you need to be competitive) components would be sensitive to non-linearities caused by the difference between -1 and +1 being larger than difference to 0. This adds overhead to the implementation similar to issues with analog computers with much lower gains than analog solutions gave. You'll note that analog computers are fairly rare as digital computers were easier to program and (usually) more reliable.

People are running actual experiments with quantum computers and artificial neurons. These both have problem domains (or are widely believed to have anyway) where they can outperform binary computers with ease. And may map to some method of implementation directly. Quantum effects are part of nature and artificial neurons can be modelled after result from brain research. Either of those might make somebody rich.

I started writing a list of possible breakthroughs, but that would go into idea generation, which is not what answers are really for. Just use something else. Unless it is story relevant you can just invent a "cool" marketing name for the invention and never explain what it was. All the characters already know what it is and none of them, possibly including the inventor, really understands it well enough to talk about it. Or maybe experience has taught him that if he tries to explain people suddenly have urgent business elsewhere.

Answer 4

I believe Ternary computing has been implemented in 1958 at Moscow State University. You can refer to this article: https://en.wikipedia.org/wiki/Setun

However, this didn't really make a breakthrough I think.

Answer 5

History One early calculating machine, built by Thomas Fowler entirely from wood in 1840, operated in balanced ternary. The first modern, electronic ternary computer Setun was built in 1958 in the Soviet Union at the Moscow State University by Nikolay Brusentsov, and it had notable advantages over the binary computers which eventually replaced it, such as lower electricity consumption and lower production cost. In 1970 Brusentsov built an enhanced version of the computer, which he called Setun-70. In the USA, the ternary computing emulator Ternac working on a binary machine was developed in 1973.

Balanced ternary Main article: Balanced ternary Ternary computing is commonly implemented in terms of balanced ternary, which uses the three digits −1, 0, and +1. The negative value of any balanced ternary digit can be obtained by replacing every + with a − and vice versa. It is easy to subtract a number by inverting the + and − digits and then using normal addition. Balanced ternary can express negative values as easily as positive ones, without the need for a leading negative sign as with decimal numbers. These advantages make some calculations more efficient in ternary than binary.

"I often reflect that had the Ternary instead of the binary Notation been adopted in the Infancy of Society, machines something like the present would long ere this have been common, as the transition from mental to mechanical calculation would have been so very obvious and simple." —Thomas Fowler

The future: With the advent of mass-produced binary components for computers, ternary computers have diminished to a small footnote in the history of computing. However, ternary logic's elegance and efficiency is predicted by Donald Knuth to bring them back into development in the future.[5] One possible way this could happen is by combining an optical computer with the ternary logic system.[6] A ternary computer using fiber optics could use dark as 0 and two orthogonal polarizations of light as 1 and −1. IBM also reports infrequently on ternary computing topics (in its papers), but it is not actively engaged in it.[citation needed]

The Josephson junction has been proposed as a balanced ternary memory cell, using circulating superconducting currents, either clockwise, counterclockwise, or off. "The advantages of the proposed memory circuit are capability of high speed computation, low power consumption and very simple construction with less number of elements due to the ternary operation."

In 2009, a ternary quantum computer was proposed which thus uses qutrits rather than qubits. When the number of basic states of quantum element is d, it is called qudit.

As you can see from the Wikipedia article, qutrits have advantages. It is very plausible for ternary to be used if binary was not adopted. One of the reasons that ternary is not used today is that everything is based on and written in binary. (Wikipedia article on Ternary Computing)

Answer 6

No, the technology to produce ternary computers already exist, but there is no necessity. It is true that a ternary computer could be more fast with same number of transistors, but the performance gain cannot justify the high cost of production.
You should also consider the power consume, if 2 binary transistor can consume less than 1 ternary there is no advantage. Think also that we need to redesign all architectures and rewrite all software, define new standards, migrate data (think to floating point from binary to ternary will be an hassle).
Too expensive and no viable for the little performance gain.

Think that now we continue to use x86_64 architecture, it is a very complicated architecture designed to keep compability with the old 8086 CPU. Intel tried to propose a new clean x64 architecture, called Itanium but it has been a failure, the performance gain was unable to justify the the rewrite of all sotware,

In my opinion the future will be 3D integrated circuits, once we will be able to 3d print processors and memory we will get a great increase of density performance and reduction of power consume, without breaking compabilities.

Another future tech is quantum computing, in that case the performance gain (for some algorithms) is so big to justify the cost.

Answer 7

Ternary is the future. While we already had an example in 1958 Moscow with the "Setun" soviet ternary computer, we have yet to make it mainstream. The breakthrough will happen with new types of transistors that can do this adequately without errors.

actually it may be quite messy to do this, but you can do ternary right this second with common transistors you can buy online or from electronic hardware store. Use npn transistors and flip their polarity. On both the collector and emitter inputs, you'd split each side with diodes so when the opposite N is active, the diode doesn't allow that corresponding output to go through it. When the OTHER side is powered, then the OTHER side's corresponding output's diode does the same thing. Like I said this is messy but it should work. Inaccurate and slow, but it should work.

Transistors already have ternary capability. And because of the way they count, (by threes, not increments of one) they do both addition and subtraction all at once to get your value, and they do it by using far less logic gates to get the work done.

So let's understand how ternary computers of the past counted. They used 0, +1 and -1 to represent their 3 values. They used ferrite blocks and diodes to do the job. However if we were to use actual transistors then 0 is of course the off position. +1 is the transistor in its on state going in one direction, and -1 is the transistor's polarity being switched to the OPPOSITE direction.

To make things easier, we can write these three values as 0, 1 and T. 0 of course is OFF. 1 is +1 and T is -1. Here we go.

Here is an example of how to count in ternary. The left most transistor is +9, the middle one is +3 and the second one is +1. Now when you flip these transistors, of course it's easy to figure out what their values are :) -9, -3 and -1.

000

Ok all 3 transistors are in the off state.

This is the value "ONE"

001

This is "TWO"

01T

Ok so the middle transistor is at "1". That means it's +3. The right most transistor is T or -1. That means it's minus one. 3 minus 1 = 2. 1T for short. (but don't forget their positions. 01T is more obvious but 1T is short)

10T

That's 8! +9 minus 1 equals 8!

Eight in binary is 1000, making it more expensive and energy consuming. Of course this particular number isn't more energy efficient in ternary than it is in binary, but most numbers are in fact less energy efficient. Not to mention the complication of circuitry in binary adding to the cost of the CPU. The amount of transistors in a common GPU is like 5 billion these days. Probably more. It's ridiculous and it's not going to get cheaper or more energy efficient any time soon.