/sci/ - Science & Math

>>12301437
trinary

Binary exists because it's simple to read off and on states.
That's one of the reasons why QC is so good, it expands computers greatly.

>>12301443
QC is not about "expanding to more states" in the way that you think. The short of it is that qudit systems can express certain systems that would be exponentially large in a classical setting, and that pure qudits states are expressed as a d-linear combination of orthornormal vectors. What this means that that our computers end up doing math directly on the Hilbert space structure of [math]\mathbb{C}^d[/math] rather than "muh more states." But being able to open up computation in this way means necessarily opening up to a setting much more sensitive to interaction, ie, no cloning.

>>12301443
THIS. It's closer to e.

>>12301437
Analog computing. INFINITE STATES.

>>12301482
Dope but how?

>>12301486
Retvrn to Tradition
What was Old shall become New again

>>12301496
That thing looks REALLY slow.
and I thought compiling in java was slow.

>>12301452
If only it weren’t impossible to implement

>>12301512
If it's running too slow for you just turn the crank faster.

>>12301564
what if we made the crank out of light, nothing is faster than that

>>12301437
non-binary digital values _are_ used in modern comuting. flash memory is a good example; different amounts of charge on a single memory is interpreted as a range of digital numbers, not just 0 or 1

>>12301564
lost.

Unary is the future of computing. There is only 1 fundamental unit in the physical universe. Why should the computational representation be any different

Having discrete states is the nigger of computing paradigms

>>12301437
Actually that's non-binary.

HAHA HAAAAHAAHAA

>>12301437
>Binary is outdated and gay.
>gay.

Actually that's non-binary.

HAHA HAAAAHAAHAA

>>12301948
in all seriousness, i don't think unary exists. it fundamentally assumes the existence of an empty character equivalent to 0. moreover, it doesn't fit the pattern all other positional numeric systems use to compute numeric values
[eqn]
\sum_{n=-\infty}^{\infty} a_n b^n
[/eqn]
because the value [math]b^n[/math] is always 1 when b=1

>>12301437
Vector machines, parallel graph reducers, in-hardware garbage collection.
Nothing new, it has all been done in the 80s and thrown away.

I propose two solutions:
>Unary
It's all ones, no zeroes. Can't get any more economic and efficient than that!
>Non-binary
Use a different number/symbol for each gender, keep the machine updated for new genders that arise, and be careful not to assume a variable's gender. If you can develop a sort of "grammar" for making up new genders from existing ones it'll be more efficient. If you can make the symbols be sets of pronouns that'd be great.

>>12301496
Unironically this. Think I figured something out

>>12301961
kek

>>12301954
Kek

>>12301437
It's not about the encoding, it's about processing. Synchronous computers are cancer, we need to move towards completely asynchronous hardware and software, which is literally the next big thing.

>>12301452
>>12301442
>>12301450
I feel like this sums up /sci/ quite nicely, two idiots talking boldly about something they know jack shit about and horribly misunderstand, and one smart guy giving a scientific explanation that is highly accurate that nobody bothers reading.

>>12301437
Monary, and with luck, monarchy

>>12301437
..26.

>>12301442
"No".

>>12301437
Hex

>>12301437
..26..

>>12301442
"No".

>>12304265
>plebbitor tourist is also a tripfag
Consider suicide then act on your consideration.

>>12304265
is that a render? if so, holy fuck

>>12304231
Oh my apologies for not worshipping the SCIgen post.

>>12304346
Also my post was meant for >>12301442
Sorry if you didn't catch that, noob. It's a classic result from information theory by yours truly: Claude Shannon.

>>12304346
>being coherent on a board about math and science is SCIgen
you don't understand quantum computing and should feel bad using it as an example about "muh expansion of computers with states"
this shit is literally a wikipedia search away

>>12301437
Why is binary outdated? I think removing a few layers of abstraction from current systems would have a greater benefit.

>>12304306
It's non-obvious what you're on about.

>>12304346
Yeah thinking quantum computing is trinary is quite embarassing, honestly. Be a little less defensive...

Normal computer:
binary
Quantum computer:
>>12301442
DNA:
Quaternary

they are close to being equal to people.

>>12304378
>>12304348
All I said was "this, it's closer to e" in response to a 3 state computer. I don't give a fuck about your QC drivel. You can circle jerk your processors that can't handle vibration and only work up to 10 qubits all you want. I'm not even acknowledging a 3 state transistor is worth it.

>>12304350
It's not outdated, but that's what we're trying to do here. There's still innovation left in binary systems too, and also need to remove bad practices as you say.

>>12301437
Blum–Shub–Smale machine

>>12304770
The BSS model is way too powerful. You can compute polynomials of exponential degree in polytime because arbitrary multiplication takes unit time.
The weak BSS model that assigns higher costs to repeated operations is more reasonable, especially since (as far as I remember), the BSS machine restricted to the {0, 1} elements of ring R can solve exactly the same problems in P as a classic TM can do P/poly. So the weak BSS is a substantial increase in power that doesn’t risk being too powerful to the point of impracticality.

Computation over arbitrary rings and computable analysis are really fucking cool subjects I wish got more attention, but the US has a lot of selective attention when it comes to Martin-Löf’s work.

>>12304350
What do you recommend on a high level?

>>12304350
Binaries are based on the fact that transistors can be either on or off. This allows for 2 bits of information per transistors. However if we had 3 or more bits of info per transistors, then we could reduce cpu cycle complexity. For example, if a transistor has 3 states, then if we want to count to 3 in binary, it would take 2 transistor space but only 1 on a trinary system.

The problem is three state systems are hard to do. Quantum bits are the next logical step with their 4 state transistors.

>>12305175
>quantum bits have 4 state transistors
You don’t understand anything about coherent superposition at all.

>>12305175
>Quantum bits are the next logical step with their 4 state transistors.
Pure qubits have [math]\mathbb{C}^2[/math] possible states. The [math]\vert 0 \rangle[/math] and [math]\vert 1 \rangle[/math] represent standard basis vectors, that are identified as classical bits, but their possible states lie in coherent superposition, ie, they can represent any such state:
[math]\vert \psi \rangle = \alpha \vert 0 \rangle + \beta \vert 1 \rangle[/eqn]

>>12305415
meant to say: ie, they can represent any such state:
[eqn]\vert \psi \rangle = \alpha \vert 0 \rangle + \beta \vert 1 \rangle[/eqn]

>>12301437
Feel free to build a machine that works better with something else. Analog? Are you for real, grandpa? As for QC, wake me up when you can implement the one algorithm that got people hyped in the first place.

And if you don't care about the machine, why bother with encoding details? Work in the real RAM model, for all I care.

>>12301486
>Dope but how?

A very large brick of "programmable" magnetic material. You input magnetic fields from the outside, and the the magnetic fields interact with each other in predictable and "intelligent" ways to do the work. You then take measurements from the outside to record the output.

>>12305446
>As for QC, wake me up when you can implement the one algorithm that got people hyped in the first place.
there's actually good research for near term quantum computers. The research has been fruitful already both theoretically and experimentally - the initial models don't have the full UQTM power, but they're not restricted to the point where they have less computational power for the things we actually care about.
It's just that the exciting stuff related to QC relates to CS and physics together - the general population probably doesn't actually care about this.

Do multiple states affect logic?

>>12305555
Nice quads. Yes, multiple states affect logic. Elaborate on your question though, please.

can we build subatomic particle transistors or is everything fucked at this level?

>>12301442
>>12304231
technically, trinary is indeed more efficient than binary, although by a small margin
t. an idiot talking boldly

>>12301437
Yes, analog.

>>12305589
So do logic gates ask some kind of operator on two variables with a boolean return like A or B where is one of the state is on the it returns true and carries the state?
how the fuck do you get a computer from that?

>>12305801
I honestly gave my best attempt at understanding your question. You can create as many inputs and outputs as you'd like to your theoretical black box and only use 2-input 1-output NAND gates to do whatever you wanted, if that's your question.

>>12304941
Can you explain the BSS model? Everything I can find is 50 pages of math my brain can’t handle

>>12305615
I’ve read that it’s already getting fuckey. Transistors have been made small enough that electrons can teleport through them

>>12301437
>programs in quibits
OH GAWD
I'M THIRD POSITIONING
AAAAAAAAAAAAAAAAAAAAAA

>>12306946
I've drank so much tonight, my autorrect is sparing me the embaraassment. Somehow anon, I relate to you.

>>12304455
The post you replied to never mentioned trinary, only QC.

RNA-based biocomputers

>>12307010
retard

>>12301964
Unary is generally accepted to be tallies, e.g.
=0
1=1
11=2
111=3
etc.

>>12307653
my point is that inherent in unary is the empty character which acts like a zero. to be clear, imagine you wanted to encode a unary number on the tape of a turing machine; to indicate you're number is over you'll need an empty character to pad out numbers (assuming a fixed width). in that case, you suppose you have two characters already. a more efficient encoding is binary, and it takes the same number of characters, unlike higher order bases. at the very least, unary is a very strange special case that is fundamentally unlike the others.

>>12301437
>next paradigm in computing
Ethnocomputing

Every time I see people posting about quantum computing and saying it doesn't work and nothing above 10 qubits exist, I cringe a little. But I'm also worried, since these old minds won't be able to handle the coming revolution in computing. Neural compute chips and quantum accelerators.

>>12307742
sure

>>12307742
you seem to know stuff about quantum computers. My current idea of what will happen is that it will be used in scientific servers and stuff and then maybe over the next decade a small component of a cpu will have a 'quantum core' or whatever on it for doing tasks related to quantum computing in the same way cpus have an aGPU. It will only impact some specific tasks and other than that doesn't change much for average joe.

Is that correct, or will it actually allow faster general use computers in some way (even if it's much further off)

>>12307769
No local qcores in the near future, you'll have hybrid q-classical algos with cloud qc. All 'quantum' processes require a classical co-processor anyway, at the minimum for error correction. It's not coming to home users in a traditional sense soon. If ever - it's unclear what problems it would usefulky address in that context.

Here's a digital-to-analogue converter with a resolution of 5 "trits", that is trinary bits. Each input bit can have three states. Assuming ideal components, it approaches the resolution of an 8-bit binary DAC.
25 = 32
35 = 243

>>12308051
2^5 = 32
3^5 = 243

For anyone who thinks >2 state system is superior to binary, please write explain to me how a nand gate can be written with a tit (it's like "bit" but "t"->"tri", cuz "b"-> "binary", get it?)

If nand gate can't be constructed, then please explain to me how absolutely anything will make sense. Write an or gate. Write literally anything. 71 posts so far, noone even attempting to explain a system to how a 2+ state system should function. Or should it work more like qc where completely new gates are forced to be invented, a new mathematical framework developed? Then the questions is: can this be proven to be more effective than binary?

And I'm not even touching on the enormous ee problems with milking more than 2 states out of a 6nm wide transistor considering the very low voltages it requires. So the transistors would probably scale sort of linearly with the span of voltage required to differentiate different states from each other, only that it would introduce higher error rates

>>12307769
The other Anon who responded
>>12307846 wasn't me, but I agree 100% with him. Although I do think there may be some breakthroughs in error correction waiting for discovery.

>>12308105
https://en.wikipedia.org/wiki/Three-valued_logic

>>12308105
aren't nands just a subset of trinary?
if trinary is {-1, 0, 1}, or {0, 1, 2}, or {0, 1/2, 1}, whatever. Just fish out the 0 and 1 and define as usual.
IRL computers have like 9 states anyways

>>12305634
There's only "true" and "false", nothing in between, unlike what retarded-ass Poos believed.
>muh non-Dualism
I wish autistic teenagers on 4chan weren't so susceptible to the influence of exotic pseudo-philosophies.

>>12305762
>analog
>literally a mash of anal+log
Who the hell thought this was a good name?

>>12308815
weak

>>12308826
Pretty much, when you start binary you meet assEmbler, CS is huge shithole, where people from top shit on people on bottom. It's like that for decades, and people who discovered transistors got shit all over the place.

Whole computing feels like fekal fetish.

>>12301437
Binary is simple and divine. Only niggers try to move away from it.

>>12306079
It's not. Learn English shit skin.

>>12309689
>learn english
>>12305801
>posts that
Are you on drugs?

>>12301496
I wonder what would spit out if we applied modern non linear optics to analogue computing. Stick some optical phase conjugate mirrors inside there

>>12301437
Easy, we just declare all math to have an answer of "1" and throw our computers in the trash.

>2020
>uaing integer bases
ngmi
Using fractal bases, especially transcendental numbers, gives a MUCH more stable system than using integers.

>>12309758
It's ok pablo, but if you want to stay here maybe take an english class, homie.

What if we did addition and multiplication by summing the energies of packets of photons

>>12308815
>There's only "true" and "false", nothing in between
Wrong

>>12301437
Array of numbers-nary

>>12307643
You replied to the QC post instead of the trinary post without even realizing and you're calling him a retard for pointing it out? If he's a retard I don't even know what you are

>>12310550
Hmmmmm

>>12301954
t. terry

>>12310381
Start with your phone.

I have to admit guys I’m a little disappointed. I imagined a form of long term storage using a tape system, where the maximum number of states would be width of tape/minimum head write size.
The math is disappointing though. A 1920 x 1080 image is about 2 million bytes uncompressed meaning any 1920x1080p image can be represented by 16 million 2 state transistors.
Now, if that information were to be represented by a 1000 state storage unit (tape), you would still need 1.6 million bytes. :( I imagined that because each byte you add can now hold 1000x more information as opposed to 2x that it would scale much faster.

It’s still some cool math. This means that there are about 2^16,000,000 possible 1920x1080p images (however the vast majority would not be indistinguishable from one another)

>>12301437
We add 2 bits together and call it a quadary

>>12310539
Just write your question properly, moron.

>>12308815
there are more bases to represent numbers other than binary. decimal is an example. (you could even use more exotic rational or complex bases).
among the 'normal' integer ones, trinary is more efficient when you balance that it is more costly (3 states) but a fixed register size may contain many more numbers
t. same bold idiot, though starting to feel not the worst offender here

>>12306079
You tried, Anon.

>>12305801
>>12309689
>>12310539
You should focus on literacy before asking questions about computers.

>>12301437
qubits ?

>>12310539
OMG this nigger is so triggered LMAO

>>12308826
Probably Andy sixx

>>12301437
It's quantum, but the paradigm after that might be fixed-size matrices. ML would be drastically faster if you could multiply two matrices in O(1) time. An n-by-n matrix can simulate any smaller matrix by appending 0s in both directions, and simulate binary by restricting everything to the top row. So an n-by-n matrix processor would have a superset of all currently-used operations. That sort of chip would waste a lot of power on a home computer, but it would be a good server-side option.

>>12307769
Right now the main applications for a quantum processor would be:
>faster indexing/search for databases
>breaking a lot of widely-used cryptography
>"true randomness" for seeding cryptographic algorithms
It's useful for big business and three-letter agencies, but the average consumer has no reason to want one.

>>12314099
>It's quantum, but the paradigm after that might be fixed-size matrices.
lolno. Neural compute chips are already a thing and tensor cores already help in matrix multiplication.
Quantum is something completely different, you can't compare it.

>Right now the main applications for a quantum processor would be:
True, but additionally
>linear algebra, solving leq
>getting info about the solution of leq like maximal entry or largest eigenvalue
>a few machine learning applications like quantum PCA and quantum reinforcement learning
>providing natural nonlinearity
>binary optimization
>finding minima in general

>>12307690
Fuck 0, me and all my homies from classical antiquity use " ".

>>12301437
>Binary is outdated or gay
ftfy

>>12301450
I mean, you could easily consider a quantum state as containing significantly more 'states' than just binary, it being a unit vector in C^n and all. The real problem is that measuring it destroys that information and only returns a binary state probabilistically rather than deterministically, so it is very difficult to work with that information directly

>>12301437
>>12301482
>>12301486
full spectrum optical

>>12316440
Light scatters too much.

>>12316620
>>12310111
what this is for

>>12316416
kek’ed

>>12301948
>There is only 1 fundamental unit in the physical universe
?
There's no fundamental unit. 'unit' in general is fictitious.

>>12301482
>>12301486
Op-amps.
"""Analog computing""" is at least 50 year old technology.

>>12301482
I know its a joke but you would still would have the trouble of splitting the range into readable states, albeit tending to infinity.