Random Number Generator - F-J's Physics - Video167

Building and using a 'Blinkenlights Supercomputer' to generate random numbers. It's a rather beautiful looking piece of electronics too!
Please consider supporting my work by buying me a coffee at
www.buymeacoff...
Very many thanks, F-J

[@AnthonyFrancisJones]

Please consider supporting my work by buying me a coffee at www.buymeacoffee.com/francisjonesa Very many thanks, F-J

[@Jawst]

❤ big clive!! These supercomputers are brilliant for soldering practice! I love making my own basic led string garden lights with 0.4mm enamelled wire and groups of 3 leds dipped in ABS/acetone slurry! Usually, last about 8 months - a year before they begin to fail from corrosion 😂 I've seen that lava lamp code generator for security, and it is a brilliant idea!!! Super randomness ❤

[@AnthonyFrancisJones]

Excellent - yes Big Clive is brilliant - I don't think he did anything with his 'supercomputer' though he did show the fail rate of cheap LEDs - I wanted to take things a bit further! Tom Scott of course did the video on Cloudflare's lava lamps. Garden light idea sounds brilliant! You will have to post a KZbin short showing them to us and post the link here!

[@radekhn]

Hello Anthony, I have two points to your video. First point, the behaviour of the blinken lights supercomputer is not random. I know it sounds strange, but it is not random. The correct word to describe it is "chaos" the LEDs behave according to the chaos theory. The second point goes back to the question why not use one resistor instead of many. Actually that is great, great question. All the LEDs are coupled via power. When you have many resistors, the coupling is very weak. Using one resistor instead of many will open two questions. 1. when majority LEDs are off, the few shining got all the current. You can "overcurrent" them and burn. 2. Having only one resistor means, that now the LEDs are more closely and strongly coupled. So there should be visible tendency to resynchronise.

[@AnthonyFrancisJones]

Radek, thanks for this. Interesting what you say. I guess it is chaotic as the there is a limited) though massive number of outcomes that are bounded within limits rather than being capable of any outcome. I am working on a chaos video at the moment but I cannot get my analogue electronics to perform the correct differential equation integrations to show the Lorenz Attractor or similar. I will get there at some point I hope! LEDs, I did not realise it would affect coupling and therefore synchronisation. That's interesting! I realise that with multiple resistors the current through any LED should be fairly fixed when it is on obviously assuming a much lower internal resistance of the power supply! Thanks for your interesting comments and for watching. Hope you will enjoy some of the other videos too!

[@radekhn]

@@AnthonyFrancisJones , yes, the Lorenz butterfly comes to my mind when I was writing about chaos. There is schematic he used, and I found it is also in The Art of Electronics the X Chapters, page 358. However you need an MPY634 or similar, and two of them. It is analog multiplier. With that re-synchronisation of blinking LEDs, it is my prediction based on the fact that blinking LED is an oscillator. When you find a way how to couple them strongly enough they must synchronise by definition. There is a nice video on YT somewhere, I guess by Veritasium or someone, and there an mechanical motronomes are used as oscillators. They behave exactly like the blinking LEDs and shortly desynchronised. Then same situation but the metronomes were on hanging desk, which could move, and by the coupling of that desk they synchronised.

[@radekhn]

@@AnthonyFrancisJones , a note to randomness a chaos. By my amateur understanding. Random system is a such system, where, even if you know all previous states, you can't predict next state of the system. Chaotic system is a such, when with a little knowledge about past you can very successfully predict next state. However the more to the future you are trying to predict, the less success and bigger error you get. Typical chaotic system is 3 body problem. Or on steroids, movement of bodies in Solar system. And chaotic system do not need to have attractor.

[@AnthonyFrancisJones]

Thanks Radek. Yes my chaos circuit will be like Horowitz and Hills. The book of theirs I bought when I was at school predated the chaos chapter addition! I am sure I have made a simple mistake or solder bridge in my circuit but it is sensitive to initial conditions by not working! I remember the issue of clocks synchronising on a wall and also remember the paper on synchronising metronomes. I think it has been much copied on KZbin. I think you may well be right about the LEDs - I will have to try a circuit. I do have one that has lots of LEDs in a cube with only one current limiting resistor and have not yet noticed synchronisation. I think they are so cheaply made that they may well synchronise and then immediately go out of sync again due to differences in their manufacture. Al interesting stuff!

[@AnthonyFrancisJones]

I think you have got it just right there so thanks for this. I am glad that the video has resulted in such interesting discussions.

[@simongarlinge3669]

Hi, Love this idea! Could you tell me where you got the LEDs. Thanks!

[@AnthonyFrancisJones]

Simon, you can get LEDs in packs of 100 or even 1000 really cheaply from China. They may not be the very best quality but fine for projects like this. Same with the resistors too. Think I got mine from Ali Express or similar. They were only a few pounds. I still have plenty left over for other projects. Remember to buy the flashing ones! Let us know how you get on and thanks for watching!

[@xxqq5719]

Is it random - read the values into a computer like a Pi and run rngtest or Dieharder on them. If you sampled the LEDs very quickly the values would be close together - not random. If you sampled them once a year they would be random because the variations between LEDs would have an effect. Happy days watching a PDP8, a proper computer with switches and lights on the front (and magnetic core memory poking out the back). Presumably the lights indicated data and address bus values, and the operators could read binary.

[@AnthonyFrancisJones]

Thanks for that and a great story! I remember using a mainframe and a dir command would enable you to hear the hard disk heads moving in the room next door! As you have picked up I am really suggesting the idea of randomness here - I have not statistically tested it! BTW I must do a video on magnetic core memory soon. I have quite a few examples including a full cubic core. I guess you have seen this video of mine? kzbin.info/www/bejne/jXOrlYGmdqhqjpY Thanks again for watching and an interesting comment.

[@fabianmerki4222]

rnd() { return 42; } is a random number, but it's not cryptographically secure pseudorandom number generator... same with your light, the patterns are clearly repeating. maybe when you snapshot it only every hour, then you may have a good distribution...

[@AnthonyFrancisJones]

Interesting. Each LED is wholly independent of each other one and will never have its timer made in exactly the same way so I struggle to see what the pattern repetition could be driven by. Granted snapshots that are temporally close together will look the same but you have got me thinking if I am missing something here. It reminds me of the rack of lava lamps that are all on at the same time but very divergent over a shortish period of time which are used to generate random numbers. Thanks for watching and taking the time to comment too.

[@fabianmerki4222]

they flash with a similar but not identical frequency which breaks your randomness over short time frames. you probably can predict the next number, using the averaged frequency.

[@AnthonyFrancisJones]

Yes, I am sure you are right there. A bit like my lava lamp example (Tom Scott did a video on that I think) one does have to wait a bit and in the case of mine give a reasonable time between each sample. However, all of this said the idea of it generating random numbers wasa only a thought! I guess if it really did someone would have built one by now!

[@pdrg]

Fun! But is that actually random? ;-)

[@AnthonyFrancisJones]

Very good question - not easy to justify but I guess if left for long enough it is not going to repeat exactly the same sequence in order for a very very long time! More a case of me finding a use for this 'toy' as it were!

[@gilbertojunqueira314]

Not only the phase difference but also the temperature will also add some randomness.

[@AnthonyFrancisJones]

Yes it will, you are right!

[@MichaelKingsfordGray]

Thay are NOT random.

[@AnthonyFrancisJones]

Michael, you may well be right but I put the challenge out there to anyone to calculate what the pattern might be after, say, 24 hours of run time and in similar steps of time thereafter. I think there are just so many variables that come into play. Let me know if you come up with anything to disprove the possible random nature of this over larger time spacings. I would be really interested to see. As an aside the LEDs often fail off or on (no longer flashing) and I am not sure I can predict that either! Thanks for watching and taking the time to comment.

[@whig01]

If you could know the exact position and momentum of everything then nothing is random.

[@AnthonyFrancisJones]

Yes, shame heisenberg get in the way!

[@whig01]

@@AnthonyFrancisJonesRight. Only God can know.