Huygens Optics
In this video I look into the idea of using optical interference to construct different kinds of logic gates, both from a conceptual- as well as from the practical perspective.
Contents
0:00 Intro
2:15 Logic gate operation
3:36 Optical logic gates
4:45 Concept of a diffractive logic gate
7:52 Practical aspects (photolithography and etching)
8:46 Wave front observation method
9:20 Results
13:51 Possible applications
Webpage with detailed and complete information on logic gates:
http://www.ee.surrey.ac.uk/Projects/CAL/digital-logic/gatesfunc/index.html
The diffractive patterns were made by using a home-brewed windows Zone plate app and then modifying them further with photoshop. The Zone Plate app can be downloaded from:
http://www.huygensoptics.com/assets/zp_writer.zip
For Windows 10 and for personal use only. Virus scanners and Windows will nag about it being a rare / unknown file and make installation sometimes difficult. Sorry about that. Install this app at your own risk.
An overview article on various known optical logic gates can be found in this paper:
https://www.hindawi.com/journals/aot/2014/275083/
However it does not contain any mention of the configuration described in this video
The video contains references to detail videos on the following subjects:
Photolithography: https://youtu.be/Lf-ev2Fop_k
Chromium photomask etching: https://youtu.be/foMT8gLYxBY
Other diffractive optics (photon sieves): https://youtu.be/TshYfYIxR9E
Microscope viewing method: https://youtu.be/TshYfYIxR9E
i'd say optical gates are very attractive mostly by the energy efficiency considerations. I was hoping to see the gate where one path is half wavelength longer than the other so they end up canceling each other. Not sure how realistic it is though…. in other words if you somehow can etch conduits with 90 degree turns you could manipulate the relative phase of the light paths. no idea how it is even possible on nanometer scale, but, it would be awesome if you could explore this realm.
Great video! An un-picable lock. Brilliant. You should file for a patent on that if you are the first, or at least write a whitepaper and have it published – even on YouTube if no one will pick it up. Optical computers, this is very stimulating (no pun intended). If you could do this with individual photons, quantum computing may merge with this. For those of you not familiar with digital logic gates: In electronics, all you really need is NAND gates because a 2nd NAND gate (with one input tied to a logic '1') on the output of the first acts as a NOT (inverter) gate and converts the NAND gate to an AND gate. So, using NAND gates, you can make: NAND, NOT, AND, OR, NOR, EXCUSIVE OR, EXCLUSIVE NOR, and any other gate with various combinations NOT on the inputs and output… and you have ALL the building blocks of computer logic. With Optical Logic, you start OR and NOT gates, and those too can be combined in various ways to create ALL of the same building blocks gates used in electronics. For example: Demorgan's Therom states you can convert, say an OR gate to an AND gate by putting a NOT on each input and output of the OR gate. Try it: The truth table of a Negitive Logic OR gate (one with NOT on the all the inputs and output) and truth table of an AND gate, both tables give the same results with respect to their inputs (thus the same gate, just inverted "logic/polarity" symbols). A postive logic AND gate is the same as a Negative logic OR gate, they do the same but you look at the logic from positive or negative perspectives, so the symbols are different but do effectively the same.
Diffusing Silica (glass) on other layers, you could make micro fiber optic paths to connect the gates together. And, if diffraction patterns could be introduced in these fiber/optic paths, they could combine at an aperture – that would eliminate need for the diffuser layer as this would be encoded into the paths themselves resolving into logic functions where they come together at various aperatures.
Lockpicking lawyer : yikes ! I am trouble or am i ?
@12:41 any thoughts on using those depth-pattern-influencing-phase-shifting rings in 3D photolithography?
How might we normalize the wave amplitude between connected gates. I assume branching an output into multiple inputs weakens the signal, so does an OR after one of two inputs turns false. Thoughts?
I was working on this process and the application of this process as far back as more than 9 years ago, but less than 10 years ago. I was working on redundancies to allow for full application of this process in the form of hi speed primary memory and processing logic at least as far back as Fall of 2019. I had concepts for mitigating waste heat and recovering waste photons, etc. One of the immediate advantages of this process is that you can make an optically actuated clock, and with negligible difference in waste heat production, run a hypothetical cpu, or bus, or memory at several orders of magnitude faster switching/clock speeds than the viable ceiling in semiconductor processing(mitigating the problem of gates operating in the shorter end of the x-ray band having slightly longer delays than is applicable in the best viable semiconductor gates, and especially applicable if working with uv light – which is more immediately testable ). I had strategies for approaching the prototyping and testing of a drop in replacement for ram in sodimm form factor, but needed to wait until I had more resources for fabrication.
Imagine a massive fpga array that occupies almost no space, has 1/1000 the latency of a conventional fpga array, and consumes 1/10 of a watt?
It is hypothetically possible to create a switchable diffraction array in multiple physical planes of interaction that can be intermittently changed to affect diffraction of indefinitely many poly-degree layers of diffraction/interaction. In other words, there is a straightforward path to making a processor that can physically change it's architecture across extremely short timeframes.
I have always had it in my head that we should be able to implement logical operations with photons, nice to see it actually come to fruition.
I'm no computer scientist, but wouldn't multiple focal points be interesting for more complex logic gates?
Like, since you can have multiple focal points on a single gate, could the output be changed depending on where you are receiving it? 🤔
I mean, I'm probably talking non-sense… 😮💨
In the first minute, the question is asked: where does the energy from a dark fringe go? Strangely, the answer is not discussed. The answer, of course, is that the energy that would have appeared in the dark fringe goes to the nearby bright fringes, which are thereby brighter than the source (non-interfering) light. Light energy can be understood as wave energy, and is conserved just like any other form of energy. There is nothing contrary to the conservation of energy in conventional discussions of wave interference, either destructive or constructive.
Why must a photonic bit be represented by a light level? Why not a frequency or phase? Phase, in particular, is attractive, since it is an aperiodic characteristic of periodic waves, varying from zero through a circle of values back to zero, either in real numbers or in complex numbers. And phase can be combined through M-Z interferometers or polarizers either at full lab scale or in miniature. All this can be done classically, without quantum complications such as a need for refrigeration and without the horrible mechanical complications of generating interference fringes and selecting one fringe. As a toy lab experiment, the author's method works just fine, and could be expected to produce some gates and flip-flops realized on top of a very large bench.
My intuition about optical car locks is that, with the development of the required tools and instruments, such locks would provide equivalent security to today's Internet encryption methods, meaning that they would have to encode similar numbers of bits to provide a similar level of security. But, with the advent of quantum or photonic computers that could try all keys quickly, the bit length in such car locks would have to be increased over time to keep up with the speed increases in computing, just as with Internet encryption. The only technique I know of that would counteract increases in computer speed is the MAC algorithm that uses multiple encryption in the software. In car locks, this would increase the complexity of the lock/unlock hardware without much change needed to car keys.
It's important to be able connect the gates together to perform computation. The performance of the entire processor is dependent on the delay of the signal through the circuit, so for best results you would want the output of one gate directly feeding the input of the next, perhaps with amplification if required.
The method of amplifying a signal must be low latency to have good performance. I know very little about optics, but I imaging a mechanism that charges, pumps, and emits light could have very high latency resulting in a high signal propagation delay.
An inexpensive FPGA kit ($200) can switch up to approx 1GHz. Some of the fastest (sequential) processors have clocks up to 5GHz (note this is the clock rate for 64-bit ALU operations so transistors are switching faster than 5GHz).
How fast can you run a flip flop out of purely optical components with minimal latency? If the signal can propagate up to C without degradation then it could be extremely fast. Visible light frequency is in the 400THz to 700GHz range.
This is a really cool idea, but I think the most difficult problem to deal with would be maintaining the phase of the light as data is transmitted throughout the machine. You would have to be able to guarantee that any time you get two inputs to an optical logic gate that those inputs are in the same relative phase. For example if one input travels a longer distance than the other it would potentially be out of phase when you actually want it to be in phase yielding a 0 when you expected a 1. I think it would probably be more robust to come up with a way to construct these logic gates with the light either on or off. The cool thing about optical computers though is that you can transmit multiple bits on the same line by using different non-interfering wavelengths. Like a byte 0xff would be white light while 0x80 would be pure infrared, 0x40 -> red, 0x20 -> green, 0x01 – uv etc… You could even handle multiple bits through the same logic gate at the same time. Depending on the difference in wavelength necessary for the light not to interfere, you could potentially have some very wide data lines (I don't know the exact resolution but I'm sure its far more than 8 bits). Also given that the speed of light is much faster than the speed of electricity and light won't generate nearly as much heat waste, optical microprocessors could be a game-changer for modern computing if we can get it to work. Is there anything like an optical transistor? That would certainly add a huge level of plausibility to the concept.
I recently got into phase-contrast microscopy, so it's pretty cool to see weirder applications of the concept.
As for the question of viability… Modern silicon transistors are less than 10nm, already a 50th the size of your green laser. The only real benefit you could get out of this is potentially quantum or analogue computing.
Very interesting!
the idea is very interesting. if it's possible to make a functional Full Adder and Register, a light based 6502 would be an amazing goal and Proof of Concept!
13:38 could one maybe display different information to different people, depending on their location? imagining a display which shows a slightly different scene to closer viewers
Dude 😀! if you make next-gen optical chips , you Rule 😀 !
Go Get iT 😀 !
I heard there are photonic computers in development with prototypes ready I think
Literally 1,000 points for the lock picking lawyer reference!
You'd need blue phase LCD to obtain the needed switching time.
Do you know how xanadu optical computing chips work
This is an interesting concept. SEEMS like it could work, but who knows?!
It's worth a shot, I would say!
Maybe it could be easier to build such a device in the radio frequency ranges.
I am a EE from U.C.Berkeley. This is quite interesting. Will research this more.
Nice work , how about introducing Michelson interferometer experiment into the idea of a logical circuit; two reverse wavelength can block the light.
Great video! Around 12:14 you start talking about trying to create no focal point, and you experiment with these paterns that create 2, 4, 8 focal points. I seems like more segments the more focal points. What would happen if you had `psudo infinite` segments, like in a spiral pattern. Would it then be all focal points, or no focal points, or is this stupid :p.
"…the best lockpicking criminal … or lawyer" 😀
In King Lord Master Friend and my Father Jesus`, Name, Amen
EOR and ENOR are so cursed
can I comment as we go along with your lecture, please?
I think it is a mistake to look at positive and negative energy as existing on opposite sides of zero… logically they are both part of what we call energy… I will ask.. what can we see around us that is less than zero… ya can't get more nothing..
we cannot have positive without negative… in electrical and magnetic systems it is the difference between them that creates the electromotive force… it is the foundation of thermodynamics and the motion of particles.. more or less.. but not nothing and less than nothing…
in computers we use negative numbers for subtraction… but the only place I actually find negative numbers… less than zero seems to be in my bank account… go figure
It doesn't have the metastable issue which is the most amazing thing of this tech