/sci/ - Science & Math » Thread #11970182

18KiB, 1200x1033, CCNOT.png

View Same Google iqdb SauceNAO

Reversible Computing

Anonymous Mon 03 Aug 21:20:26 2020 No.11970182 View Reply Original Report

Quoted By: >>11970210 >>11971483 >>11974307 >>11975445 >>11975595 >>11979049

Can someone explain reversible computation?
Doesn't make much sense to me, to be honest.

Anonymous

Anonymous Mon 03 Aug 2020 21:28:18 No.11970210 Report

Quoted By: >>11970233

>>11970182
>In this modeling framework, an inverter (logic gate) (NOT) gate is reversible because it can be undone. The exclusive or (XOR) gate is irreversible because its two inputs cannot be unambiguously reconstructed from its single output. However, a reversible version of the XOR gate—the controlled NOT gate (CNOT)—can be defined by preserving one of the inputs.
It saves enough data so you know how to go backwards

Anonymous

Anonymous Mon 03 Aug 2020 21:40:44 No.11970233 Report

Quoted By: >>11972511

>>11970210
I'm not interested specifically in the CNOT, I'm interested in reversible computation on the whole.
I only used an image of it because I needed an image.

As long as we're talking about the CNOT:
How is that even different from a normal XOR gate?
A XOR gate does not necessarily consume it's inputs, therefore you can undo a XOR as long as didn't get rid of both inputs.
If that wasn't true then XOR swap wouldn't work.

Anonymous

Anonymous Mon 03 Aug 2020 21:52:58 No.11970278 Report

Quoted By:

Injective logic gates are reversible in general if you can restore atleast one input

Anonymous

View Same Google iqdb SauceNAO Aberration-correction-using-phas (...).png, 18KiB, 679x573

Anonymous Tue 04 Aug 2020 00:03:27 No.11970917 Report

Quoted By:

I thinking there is a fourier machine running much higher in frequency than our perception of time permits. I reflect upon phase conjugate mirrors, these 'particles' seem to skip a beat

Anonymous

Anonymous Tue 04 Aug 2020 03:39:16 No.11971483 Report

Quoted By:

>>11970182
how much thermodynamics do you know? Peter Shor has a good explanation on cstheory stackexchange.
https://cstheory.stackexchange.com/questions/4204/formal-notion-for-energy-complexity-of-computational-problems

Anonymous

Anonymous Tue 04 Aug 2020 05:04:52 No.11971639 Report

Quoted By: >>11972543

It's supposed to let you "reuse electrons" for further operations after reversing the logic, so they don't get converted to heat when they are dicipated or discarded or become a garbage state such as when an and gate (not reversable) takes two bits and ends up with one bit as output. Where did the other bit go, it didn't really go anywhere but part of the electrical signal allowed through the gate dissapated as heat in the transistor via the resistance that exists in it even as the gate is open. When the gate is closed there is more resisting so more heat.

I might be wrong on parts of this.

Anonymous

Anonymous Tue 04 Aug 2020 13:47:53 No.11972511 Report

Quoted By: >>11974278

>>11970233
>A XOR gate does not necessarily consume it's inputs, therefore you can undo a XOR as long as didn't get rid of both inputs.

But you always do. There's something called memory. And almost all algorithms depend on doing successive calculations punctuated by saving and recalling from memory. There's only one operation being performed the the CPU at once so even something as easy as a+b+c has lost all the inputs to a+b before it gets to adding that to c.

Anonymous

Anonymous Tue 04 Aug 2020 14:03:23 No.11972543 Report

Quoted By: >>11972547 >>11974746

>>11971639
You are slightly wrong, because the input to a CMOS gate is very high impedance. Look at emitter collector current for a working transistor. The leakage current could just drive the output slightly higher than the input level current.

Anonymous

Anonymous Tue 04 Aug 2020 14:05:16 No.11972547 Report

Quoted By:

>>11972543
>emitter collector
emitter base*
The base is the input signal.

Anonymous

Anonymous Tue 04 Aug 2020 14:06:55 No.11972552 Report

Quoted By:

I think it would have something to do with information theory and that some information is recoverable.

Anonymous

Anonymous Tue 04 Aug 2020 14:09:37 No.11972560 Report

Quoted By: >>11972572

https://en.wikipedia.org/wiki/Billiard-ball_computer

Anonymous

Anonymous Tue 04 Aug 2020 14:17:30 No.11972572 Report

Quoted By:

>>11972560
https://en.wikipedia.org/wiki/Fluidics

Anonymous

Anonymous Tue 04 Aug 2020 23:05:52 No.11974278 Report

Quoted By: >>11975436

>>11972511
I'm not following. Unless you're using an old-ass accumulator based architecture, then no, you aren't forced to consume inputs.
If I do 'XOR R1, R2, R3', then Registers R2 and R3 are unmodified, while the calculation is stored into R1.

Anonymous

Anonymous Tue 04 Aug 2020 23:14:14 No.11974307 Report

Quoted By:

>>11970182
reversible computation just means that your set of computational operations forms a group.

Anonymous

Anonymous Wed 05 Aug 2020 00:21:51 No.11974746 Report

Quoted By: >>11975423

>>11972543

Do you mean that most of the heat generated is from the high resistance region near the collector-base junction and that most heat is generated when the gate is closed, but also some when open?

That's how I figure it but not sure.. I absolutely made it seem like most heat comes from an open gate.

Anonymous

Anonymous Wed 05 Aug 2020 03:30:38 No.11975423 Report

Quoted By:

>>11974746
Whoops, wasn't thinking there. The inputs are high impedance. Output is from the rails.

Anonymous

Anonymous Wed 05 Aug 2020 03:37:20 No.11975436 Report

Quoted By: >>11975795

>>11974278

Ok. You just demonstrated a+b. Where is that results +c? I'm aware this is still possible, but you'll run out of registers pretty quick.

Anonymous

Anonymous Wed 05 Aug 2020 03:42:27 No.11975445 Report

Quoted By:

>>11970182
CSWAP is the best logic gate.
>reversible
>conservative
Change my mind.

Anonymous

Anonymous Wed 05 Aug 2020 04:37:19 No.11975595 Report

Quoted By: >>11975806

>>11970182
it's just applied bijective functions

Anonymous

Anonymous Wed 05 Aug 2020 05:47:22 No.11975795 Report

Quoted By: >>11976911

>>11975436
What difference does it make if it's stored in a register or memory from a reversible computing standpoint?

And yes, memory isn't infinite, duh.
That's something that every computer must contend with except for a hypothetical Turing machine.

Anonymous

Anonymous Wed 05 Aug 2020 05:50:00 No.11975806 Report

Quoted By:

>>11975595
And for some reason that makes it more energy efficient?

Anonymous

Anonymous Wed 05 Aug 2020 14:20:37 No.11976911 Report

Quoted By: >>11977789

>>11975795
>What difference does it make if it's stored in a register or memory from a reversible computing standpoint?

So you want to store a+b+c=d somewhere in memory? In general today the computer just stores d and eventually overwrites a,b,c. If you want to define a computer where no bit of data that goes into the result is ever overwritten go ahead, but that's not the case with modern computing not would it seem to be useful. I thought the idea of a reversible computer would perform something like factoring large numbers. A true reverse computer would solve the P=NP problem, because it could undo all the successive arithmetic that led to the answer

Anonymous

Anonymous Wed 05 Aug 2020 18:53:08 No.11977789 Report

Quoted By: >>11979044

>>11976911
>So you want to store a+b+c=d somewhere in memory?
No.
But if you want a+b+c=d to be reversible, then you must store at least 3 of those four numbers.

>but that's not the case with modern computing not would it seem to be useful.
Good thing we're not talking about normal computers then, we're talking about reversible computing.

Anonymous

Anonymous Thu 06 Aug 2020 01:35:19 No.11979044 Report

Quoted By: >>11979076

>>11977789

But what would be the point? It would be interesting to put a large product of two primes in r1 then do a reverse multiply and get it's factors, but that's not what you are talking about. What would be the value of your computer?

Anonymous

Anonymous Thu 06 Aug 2020 01:36:28 No.11979049 Report

Quoted By: >>11979076

>>11970182
Neither science nor math

Anonymous

Anonymous Thu 06 Aug 2020 01:48:01 No.11979076 Report

Quoted By: >>11979116

>>11979044
Read up on why reversible computing is important before replying to him again. He obviously knows a lot more than you on the subject. RC is important because it will let us run some algorithms with less energy per operation than is possible with a nonreversible computer.

>>11979049
It's both, retard. Reversible computation is performing physical arithmetic operations with no increase in entropy. By Landauer's principle the operation must also be logically reversible. Lp is based on thermodynamics.

Anonymous

Anonymous Thu 06 Aug 2020 02:03:02 No.11979116 Report

Quoted By: >>11979379

>>11979076
>RC is important because it will let us run some algorithms with less energy per operation than is possible with a nonreversible computer.

Such as?

Anonymous

Anonymous Thu 06 Aug 2020 03:47:53 No.11979379 Report

Quoted By:

>>11979116
There are many. Any algorithm which doesn't require the deletion of information can generally be made reversible. A logical NOT is reversible. XOR is reversible if you keep one of the inputs as well as the output. Fundamentally irreversible algorithms are things like lossy compression, where information deletion is the goal.

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches