Do normies actually find this stuff hard?

Do normies actually find this stuff hard?

Other urls found in this thread:

en.cppreference.com/w/c/language/operator_precedence
ncbi.nlm.nih.gov/pubmed/27552985
twitter.com/NSFWRedditGif

-13

Only a normie finds this sort of bullshit interesting.

2

It's not about being hard, it's about using the convoluted retardation that is infix notation to post troll questions that make people disagree with each other.

>muh speshul notayshun
fuck off. operations in any ring are written like that by everyone for a reason

Literally everything of importance is done in RPN. You just don't realize it because there's a retard proof shunting algorithm that lets users continue to input retard infix.

>RPN
Implying anyone uses this for anything important. Just use proper parentheses ya dingus

You're confusing what you input with what actually gets used.
Go ahead and find me some examples of low level programmatic systems that calculate using infix. You can't because they don't, because it would be retarded and needlessly convoluted to have to write everything in terms of infix. Just because support has been set up to allow retards to input infix doesn't mean the programs themselves use infix, it just means there's a clever algorithm that parses out your bullshit into a system that actually works (RPN).

>You just don't realize it
you're a fucking retard. so since computers represent everything in bits should I speak in ASCII too?

fuck off

kek this

Should we use binary too because binary is better for computers? We should input numbers in binary so that they don't have to do an ASCII => Base10 => Base2 conversion, right?

Don't be retarded. Internal program semantics have nothing to do with which format is better, especially since the computer is not representing it in pictatorial RPN but rather abstract struct-based RPN which has nothing to do with handwritten RPN

Yeah but whats the fucking answer

why are compsci retards so disgusting?

-13 dum dum

I thought it was 17
3*6
18
-3
15
+2
17

>-3
it's 3-18, dumdum

Err, I'm a brainlet, but I thought it was -17?
The relevant operations from standard order of operations is multiplication, addition, and subtraction in that order

3x6 = 18
18+2 = 20
3-20 = -17

...

it's (-3)x(+6)

(-)x(+)=(-)

-3 x 6 = -18

Who gives an everloving fuck about the order of operations? Use brackets you autistic cunts.

I saw this on Veeky Forums once before and I was sure it was a troll.

And then today I literally see this horrible meme on my facebook, with a friend responding suggesting they've solved it.

I don't see how ASCII or binary are relevant, unlike with them you can actually use RPN and not have to deal with ambiguity in operation order. If you're able to use ASCII in your everyday life in a way that gets you any sort of practical benefit then sure, use ASCII. Same thing with binary. There are cases where using binary yourself probably would be beneficial so that one's less bad as an example, but still, it seems like you're trying to point to other random examples of things programs do to try to raise the complaint that not everything programs do is something you would want to do personally when that was never an argument in the first place. The case with RPN is that it gets used everywhere for calculation purposes because it's unambiguous and straightforward, which is something you obviously can benefit from using at a personal level.
>We should input... so that they don't have to do... conversion, right?
I'm not saying you should input RPN so computers don't have to convert to RPN, I'm saying you should use RPN for the same reason programs always do calculations using RPN which is that it's completely unambiguous in which operations will be done first without requiring a bunch of convoluted and arbitrary rules about operator precedence, and without requiring a bunch of nested parentheses to override that operator precedence scheme.
>Internal program semantics have nothing to do with which format is better
The fact it would be a nightmare having to define calculations in terms of infix has everything to do with which format is better. There is no reason to want to use infix except that you grew up with it and got used to it, it's objectively terrible in every way possible and the exact reason why these troll threads are capable of existing.
Try posting a troll thread with an RPN problem. Nobody will disagree with what the answer is because RPN has no ambiguity and the order you process the statement in is the order it's written in.

off yourself despicable freak.

>pic

What textbook is that from? How would I into black holes the non-meme way?

Why do you enjoy ambiguity so much? Take the infix dick out of your mouth, it's not helping you.

>tfw you cant even follow order of operations so you have to make some pollack method

Why do brainlets do this?

Nobody can follow it, again why do you think these retarded order of operation troll threads keep getting made?
It's not an intelligence issue, it's an ambiguity issue.
If anything makes you not intelligent it's voluntarily continuing to use a method of calculation that's both ambiguous and less efficient. You're doing extra work and getting more problems back for your efforts.

No ambiguity, no arbitrariness. Facebook viral images are not an argument, this is a non-issue among mathematicians.
One can write polynomials without parentheses, I shudder to think what they look like in RPN.
Use function notation for functions, infix notation for group operations. Looks good.

>no arbitrariness.
It's definitely arbitrary. There is no inherent reason why any particular operation should have precedence over another. If there were RPN wouldn't be able to exist and infix parentheses to override the arbitrary precedence scheme wouldn't be able to exist.
>No ambiguity
Pic related. Try −3^2 in Excel and in Google. There isn't agreement on whether to apply exponentiation first or whether to apply the unary interpretation of the negative sign first. Some infix calculations will interpret the negative sign as equivalent to subtraction and do it after the exponentiation (like Google) while others will interpret it as a unary operator and give it higher precedence over the binary operator of exponentiation (like Excel).

Seems pretty easy to me, OP. Now your turn. How do you solve this one?

(3-3) x (6+2)
0 x 8 = 0

(3) - 3 x 6 + (2)
(3) -18 + (2)
(3 + 2) = 5 -18 = -13

Not arbitrary. It's nice to be able to write readable polynomials without parentheses.
I don't give a shit about some dumbass computer program having trouble.

>Not arbitrary.
How are you still arguing that when it clearly isn't? What about multiplication makes it inherently deserving of higher operator precedence than addition? You can't answer that without either lying or being mistaken because statements where addition is done first clearly do exist, and if this scheme weren't arbitrary that couldn't be the case.
If you really need to use an arbitrary operator precedence scheme then fine, but don't pretend it isn't arbitrary.

It's conventional but it's not arbitrary (ie pulled from one's ass without rhyme or reason). Again, take a look at polynomials:
-3x^2 + 5x - 7 (conventional)
-3*(x^2) + (5x) - 7 (left-to-right)
x x * -3 * 5 x * + 7 - (rpn lol)

-3^2 = -1 * 3^2 = -9

Excel being wrong does not matter

Why are you changing the terms you're using for RPN instead of just doing:
-3 x 2 ^ * 5 x * + 7 -

Excel's not the only system that applies unary operators like a negative sign before doing exponentiation.
There's nothing inherently right or wrong about either approach. The only wrong thing is your decision to use ambiguous notation methods in the first place.

still looks like ass, and no, it's not about getting used to it

look here you bottom-feeder fa/g/atron compsci cripple, this is not the board for you, GOTO

It only "looks like ass" to you because you were arbitrarily indoctrinated into infix growing up.
There is no inherent sense in which using operators in a sensible stack based approach is aesthetically unappealing.

Plenty of engineers today still use RPN calculators without knowing anything about programming.
That said I'll go ahead and interpret that non-argument outburst as your subconscious beginning to accept infix is a bad choice.

2

Negative sign is not an unary operator, it's shorthand for 0 - x. And hopefully you wouldn't be stupid enough to argue that 0 - x ^ 2 = 0 + x ^ 2

It isnt ambiguous if you just follow the rules dumbass

>accept infix is a bad choice.
You sound like a retard who've just learned about polish notation and now needs an excuse to bring it up for no reason.

on the off chance that this isn't bait, addition does not have higher precedence than subtraction, they have equal precedence and should be applied left to right in the same step (the same is true of multiplication and division)

There is no ambiguity in infix notations, even if you consider it more arbitrary, as a programmer who has done this type of thing I absolutely agree that RPN and its variants have very nice properties for internal representation, however insisting that this makes it easiest or most practical for human use is very naive. For example (a/k + b) * (c/k + d) has clear delineation of the multiplicands, an abstraction which is very useful for a human to be able to quickly see, whereas it is far less clear in something like ak/b+ck/d+* what the argument separation is to a human (not so bad in this case but obviously as the expressions become more complicated they become difficult for a human to read because physical distance has little correlation to their distance in an AST)

>Negative sign is not an unary operator
It explicitly is unary in some environments like Excel for example.
Or, you know, that obscure language C that you might have heard of.
en.cppreference.com/w/c/language/operator_precedence
I've been arguing this for many years now, not that it matters when in the timeline of my life that I first learned how bad the infix system we're indoctrinated into is or that there's a much less convoluted solution that completely eliminates ambiguity without the need to ever use parentheses.

>There is no ambiguity in infix notations
See:
How is that not ambiguous when two extremely popular environments people use everyday (Google and Excel) assign different operator precedence schemes to the same ultra-simple three character calculation?
And yes, it does matter that this difference exists in a practical way because Excel is known to be used to produce calculations in a disturbingly large number of formal academic papers.
Here's a well documented massive collection (1/5 of the papers checked) of gene name errors caused by the popularity of Excel use among researchers of genetics:
ncbi.nlm.nih.gov/pubmed/27552985

>we should stop using infix because computers are stupid
Back to /g/.

Not working with a flawed system isn't stupidity.
That'd be like saying modern astronomers are stupid for not using epicycles.

I can also assume something to be true and build a shitty argument on top of that assumption: Only idiots use postfix and idiots belong on /g/, therefore, you belong on /g/.

3-3x16+2= 3-18+2=-15+2= -13

Easy

PEMDAS, left to right. Or you can just ignore the addition and subtraction functions and look at everything as either positive or negative integers and add them all up.

...

Finance fag here, nothing in the world gave me greater pleasure than my first year of grad school where the professors forced everyone to use these ancient hp-12c calculators that use RPN. Took everyone most of the year to figure it out, personally I found it more intuitive and flexible. Still fuckin use that calculator all the time.

>3x16
Where did you get that?

D-, Ned, D-

3-3 = 0
0x6 = 0
+2
= 2

substraction and addition have same priority, if both are in an equation you do left to right in that order.
likewise division and multiplication are in their own priority. You decide how to do it left to right.
Anyways, problems like these are stupid and are the reason people came with parenthesis instead of this shit

>if both are in an equation you do left to right in that order.
Thats not true
It doesn't matter what order you do them

dude u wot. 3-3*6+2 is equal to 3-(3*6)+2, so 3-18+2, so 13.

That's why calculators usually have 2 - buttons, a prefix - (usually denoted as (-) ) and an operand -, the former denoting a negative number (so (-3)^2), the other denoting an operand (so -(3^2) ).

yeah, it doesn't matter if you do it correctly, it matters if you do them 'out of order' incorrectly.

The dumbfuck that got it wrong solved it like this:

3-18+2 = 3-(18+2) , which is obviously false.

Of course, 3-18+2 = -18+3+2 = 2-18+3 = 3+2-18 and so on

>People are normies because they don't care about the arbitrary order of math operations they'll never need to use in their life.

Anyone can know this and anyone can not know this. Don't be a pretentious retard.

the left cant separate terms so it means it AUTOMATICALLY is oprresion

wtf it is like this
3-3*6+2
3-18+2
-15+2 (you do equal hierarchy operations from left to right)
-13

the hierarchys are:
parnthesis
root=exponent
multiplication=divition
sum=sustraction

They do, because instead of teaching them basic arithmetic logic they just tell them PEMDAS PEMDAS PEMDAS

3-3x6+2=??
-13=?^2
?=sqrt(-13)

fuckn normies the solution is √ƪ * ¾3(מ).
Proof is left to the reader

>i learned some compscience and muh pollack form

infix notation is much more natural for a human to read

itt: people pretending to be retarded

Why is this merimut term always coming up when people are talking about math?

>Long equation
what
>teaches you how 0 is everything and nothing
Do they teach BEDMAS to ameritards?