PDA

View Full Version : (Spivak) - a function with strange behaviour.


kioria
Apr12-04, 07:30 AM
1) Find a function, f(x) which is discontinuous at 1, \frac{1}{2}, \frac{1}{3}, \frac{1}{4} ..., but continuous at any other points.

Solution (I have come across, probably wrong and a half):
f(x) = { 1 for all real x; 0 for 1/x where x is natural numbers.

Can any one tell me the answer to this?

ReyChiquito
Apr12-04, 07:48 AM
The function you have founded have indeed that property, there is an infinite set of functions that can do the job...

kioria
Apr12-04, 07:50 AM
The function you have founded have indeed that property, there is an infinite set of functions that can do the job...

Excellent! Thank you. :biggrin:

Nexus[Free-DC]
Apr13-04, 05:59 PM
Does the function need to be defined everywhere? If not, you can construct an elegant solution as follows:

f(x)=\frac{(x-1)(x-1/2)(x-1/3)...}{(x-1)(x-1/2)(x-1/3)...}

This function is equal to 1 except at the points \frac{1}{n}, where it is undefined.

matt grime
Apr13-04, 06:34 PM
Part of the definition of a function is its domain, and it needs to be defined on its domain.

If you mean, say, is 1/x a function from R to R? No: you've not defined what it is at zero, until you do it is at best a function from R\{0} to R.

This is a big problem that is not taught properly when it first arises and causes many unnecessary problems.

The one you gave has the nice property that one may define f at the points in question so that it is 1, and is continuous at all those points.

Gokul43201
Apr16-04, 06:15 PM
see attachment for a family of solutions

kioria
Apr17-04, 08:56 AM
Thanks for all the help :)

kuengb
Apr17-04, 09:13 AM
Kioria, I just want to add that the function you gave is also discontinuous at x=0, not only at x=1/n. But you can change it like this to kill that bug:

f(x) = { 0 for all real x; x for x=1/n , n any natural number

matt grime
Apr17-04, 10:15 AM
If you're going to get picky then the original definition doesn't define a function.

kioria
Jun19-04, 11:05 AM
Part of the definition of a function is its domain, and it needs to be defined on its domain.

If you mean, say, is 1/x a function from R to R? No: you've not defined what it is at zero, until you do it is at best a function from R\{0} to R.

This is a big problem that is not taught properly when it first arises and causes many unnecessary problems.

The one you gave has the nice property that one may define f at the points in question so that it is 1, and is continuous at all those points.

Interesting. Good stuff.

fourier jr
Jun19-04, 05:57 PM
The Gauss Transformation is a function that is like that. It looks like this:

G(x) = 1/x + [ 1/x ] for x in the interval (0,1]
G(x) = 0 @ x=0

[] means floor, aka least integer function.

stefanfuglsang
Jun25-04, 03:52 AM
\Gamma( \frac {-1}{x} ), for x > 0