Return Styles: Pseud0ch, Terminal, Valhalla, NES, Geocities, Blue Moon.

Pages: 1-

d/dx f(x) = f(x)

Name: Anonymous 2007-12-16 16:39

Are there any other functions from R to R whose derivative everywhere equals itself, beside x -> a*e^x for constant a in R?

Name: Anonymous 2007-12-16 16:44

f(x) = 0

Name: Anonymous 2007-12-16 16:49

for y=y' the only solutions are y=c*e^x and y=0(this latter one is refered to as the trivial solution)

Name: Anonymous 2007-12-16 17:21

>>3
Actually, y=0 is y=c*e^x for c=0.

Name: Anonymous 2007-12-16 17:28

>>2-3

x -> 0

is the same as

x -> 0*e^x

dum-dums.

Name: Anonymous 2007-12-16 19:11

Whilst I'm pretty certain it's correct, your posts hardly constitute a proof.


I'd suggest considering the equation F(x) - F(0)*e^x = g(x)

And then assuming there is a point a s.t g(a) |= 0 , and then working to a contradiction using the meant value theorem.

I tried it briefly, I'm too tired right now. (Shitty excuse I know)

Name: Anonymous 2007-12-16 22:03

Or by proving that f(x)=f'(x) implies that f(x)=a*e^x. This is very easy.

Name: Anonymous 2007-12-17 5:49

>>7
it really is
just look for basic differential equations on books
its bound to be there as an example

Name: Anonymous 2007-12-17 6:06

Isn't there some theorem somewhere that guarantees uniqueness of solutions or something?

Name: Anonymous 2007-12-17 18:05

>>7
>>8

are stupid. The proof will almost undoubtedly be of the form I described.

>>9 The theorem that guarantee's uniqueness of solutions will almost undoubtedly be harder to prove than the first result.

Name: Anonymous 2007-12-17 18:16

It is very simple. If f(x)=f'(x), then f'(x)/f(x)=1. Since f'(x)/f(x)-1=0, then d(ln(f(x))-x)/dx must also equal to 0 since it is equal to f'(x)/f(x)-1. Since it can easily be proven that only constant functions have 0 derivatives everywhere, then ln(f(x))-x must be a constant function C. The rest easily follows.

Name: Anonymous 2007-12-17 18:19

I also forgot to mention that if f(x) is 0 anywhere, then the proof obviously does not work. You have to take the case separately in order to prove that if f(x) is 0 anywhere with f'(x) also equaling 0, then f(x) must be equal to the constant function f(x)=0.

Name: Anonymous 2007-12-17 21:42

My calc 2 class basically started with d/dx f(x) = f(x) then did some shit with some properties and showed that the solution was e^(x).  So yeah its just e^(x).

If you only care about magnitudes and allow complex numbers, e^(sqrt(-1)*t) would qualify.  Its magnitude is 1 everywhere (just changed direction in the complex plane) and its derivative acts the same, just 90 degrees off.

Name: Anonymous 2007-12-17 23:17

>>13
>So yeah its just e^(x).

Right. Assuming you ignore 2*e^x, 3*e^x, pi*e^x, and so on.

Name: Anonymous 2007-12-17 23:32

>>13
e^(sqrt(-1)*t) does not work. Its derivative is i*e^(it) which is NOT equal to i*e^(it)

Name: Anonymous 2007-12-17 23:40

>>15
I made this post, and I was not paying attention to >>13 about magnitudes, so it was a mistake. However, an important thing is wording "functions from R to R" in the OP which clearly excluded complex functions or their absolute values. If, however, their absolute values were to be included, then it would be the same as the regular real function solutions.

Name: Anonymous 2007-12-18 9:23

someone please diff(e^(it) / (cost + isint)) step by step for me?

I tried to do it and it is NOT 1
or maybe I made a mistake somewhere, can someone do it?

Name: Anonymous 2007-12-18 10:27

>>11

Is easy to prove USING THE FUCKING MEAN VALUE THEOREM.


Plus your proof using implicit facts that have to be derived first, Working from first principles.

I stand by my statement.

Name: Anonymous 2007-12-18 10:30

>>17
>it is NOT 0
fixed

Name: Anonymous 2007-12-18 10:39

>>19
use quotient rule if you don't want to make the initial cancellation.
d/dt(e^it/(cost + isint)
= (cost +isint)ie^it - (-sint + icost)e^it/(cost + isint)^2
= (i(cost + isint) - (-sint + icost))e^it/(cost + isint)^2
= 0

Don't change these.
Name: Email:
Entire Thread Thread List