« first day (2104 days earlier)      last day (2754 days later) » 

11:02 PM
@flawr I had a math question
can one derive the fourier transform by taking the inverse fourier transform which can be derived basically from common sense, and then solving for F()? I couldn't get anywhere with that approach
 
File.read("/usr/share/dict/words").split.map{|i|i.gsub("'s","")}.shuffle[1..5].‌​join(" ").capitalize+"."
 
> I think Randall has a time machine and checks Reddit a couple years ahead for the phrase "I don't have a relevant xkcd!" and then comes back and makes it to prevent that timeline from occurring.
 
@Maltysen Do you know linear algebra?
 
number overflow -> 43 mil jackpot
 
@feersum yeah
basically
 
11:10 PM
So if you do the type with discrete coefficients (that can form only periodic functions), it's relatively straightforward
 
@PhiNotPi But it was!
 
@feersum you mean the fourier series?
 
You can show that that sin/cos n pi x forms a basis for the space of periodic functions
@Maltysen Right
 
that's what I was shown
no linear algebra, just a proof of the series, then take limit period → ∞ gives transform
but since the inverse transform seems so easy, I was wondering if it was possible to use that to prove the transform
 
Well for the series, you can show that it's a basis, meaning that to invert it you use the inner product to find the weight of each function
I don't know exactly how to show that for the continuous version.
 
11:14 PM
@feersum oic what you mean
yeah, that's why I was saying that the inverse is super easy to derive
but I was wondering if i can use the inverse to derive the forward
 
1
Q: Palindromize this string!

OliverYour task is to palindromize a string as follows: Take the string. abcde Reverse it. edcba Remove the first letter. dcba Glue it onto the original string. abcdedcba But here's the catch: since this challenge is about palindromes, your code itself also has to be a palindrome. Remembe...

 
but the change of basis idea is pretty cool
 
Oh when I said invert I actually meant the "forward" way
 
@feersum ohhhhh
 
Oh it should be n pi x / L where L is the period of the function
 
11:16 PM
so that's why the forward is just the same as the inverse
(*-1)
oh you meant inverting the change of basis
 
11:35 PM
@noɥʇʎPʎzɐɹC I keep trying to figure out how they got that number, but it's not just simple integer overflow
 
@El'endiaStarman Ah, whoops. I made a math error in the JS benchmark. It's actually plenty fast, but GPGPU still gives a massive improvement.
TIL calculations that result in NaN really stall JS.
 
@Maltysen I don't quite get what you're asking
 
1
Q: Depalindromize this string!

OliverGiven a palindrome generated according to this challenge, depalindromize it. Test cases abcdedcba -> abcde johncenanecnhoj -> johncena ppapapp -> ppap codegolflogedoc -> codegolf Remember, this is code-golf, so the code with the fewest bytes wins.

 
@mınxomaτ s/that result in NaN// :P
 
Well, I fixed that
Hence this whole project
 
11:48 PM
@flawr thanks but @feersum cleared a lot of things up for me, which made me realize that my question was pretty stupid
 
Ok=)
All in all it is worth noting that the fourier transform and its inverse are basically the same thing. (in the discrete aswell in as in the continuous case)
(up to some "constants")
 
@flawr is there an intuitive reason why that is so?
 
There was a really nice explanation, let me see if I can find it.
The idea of wrapping a signal around a circle (with a given frequency/angular velocity) and averaging (=sum or integral) the points to get that share of the signal is a really helpful image.
If you got that, it is easy to see that the inverse basically consists of doing the same again.
 
does anyone here use cmder? (windows console emulator)
 

« first day (2104 days earlier)      last day (2754 days later) »