How I can construct a continuous onto function from [0,1) to (−1,1) ?
I know that such a function exists and also I have a function f(x)=x2sin11−x which is continuous and onto from [0,1) to (−1,1). But I don't know how I can construct this type of function.
There are many function which is continuous onto from [0,1) to (−1,1). I can construct a continuous onto function from (0,1) to (−1,1). But when the domain interval is left-closed then how I can construct ?
Please give the idea to construct the function.
Answer
I suggest to start by drawing: draw the bounding box [0,1)×(−1,1), place your pencil at (0,0), and trace different functions. The following are observations and hints, not logical proofs.
A first idea is that, since you want a continuous function, a monotonous one might be troublesome in mapping a semi-open interval like [0,1) to an open one like (−1,1). You can find one location in [0,1) where f(x) approaches 1 and another location in [0,1) where f(x) approaches −1. If these locations are in some [ϵ,1−ϵ] (ϵ>0), continuity might impose that the values −1 or 1 will be reached strictly inside [0,1), which you do not want.
One remaining choice is that values y=−1 and y=1 are both approached on an open end like [. 1). So you might need a function that oscillates infinitely often close to x=1. In other word, it can be wise to open [0,1) to something like [0,+∞).
To summarize, three main ingredients ingredients could be useful, in the shape of a fellowship of continuous functions that you could compose. Many choices are possible, here is one (borrowing from Tolkien's Ring poem):
- three functions for the unit-interval [0,1) up to the sky [0,+∞) (hereafter f0, f1, fϕ),
- one for the oscillation lords in their infinite [0,+∞) hall of sound (hereafter f2),
- one function to bring them all and in ]−1,1[ bind them (hereafter f3),
in the Land of functions where continuity lies (indeed, continuous functions tend to cast continuous shadows to connected intervals).
The first ingredient f1(x) is easy with functions you know, defined on some [a,b[ with a singularity at b. It is easy to remap [a,b[ to [0,1[, so let us stick to that interval [0,1[. Examples: 11−x, −log(1−x), tan(πx/2), and many more.
If you want more flexibility, you can start with any function f0 that maps [0,1[ to [0,1[: xp with p>0, sin(πx/2), log(1+x(e−1)).
For the second ingredient f2 on [0,+∞), the sine is a nice pick, and a lot of combinations of sines and cosines, like a chirping sound. But you have a lot of fancy alternatives. And you can easily plug in a function fϕ that maps [0,+∞) to [0,+∞) (for instance exp(x)−1, xp).
The choice of f2 is possibly the most sensitive, since you will need to strictly bound it afterward inside (−1,1), therefore you will need a third ingredient: a function f3 that compensates (as a product for instance) the envelope of f2 so that the result does not exceed 1 in absolute value. So for the sine, xp or expx−1e−1 will do the job. A function whose magnitude is strictly less than 1, and tends to 1 as x→1 is likely to work.
Finally, a function can be obtained by composing f(x)=f3(x)×f2(fϕ(f1(f0(x)))).
In your case, you have for instance f0(x)=x, f1(x)=11−x, fϕ(x)=x, f2(x)=sin(x), f3(x)=x2.
While not fully generic, you can cook of lot of recipes with those ingredients. For instance (see image below):f(x)=sin(πx7/2)sin(tan(π√x/2)).
No comments:
Post a Comment