Involving Involutions

Calculus Level 1

An involution is a function, f ( x ) , f(x), that "un-does" itself such that f ( f ( x ) ) = x . f\big(f(x)\big) = x.

Here are three examples:

f ( x ) = 1 x f ( f ( x ) ) = 1 1 x = x f ( x ) = x f ( f ( x ) ) = ( x ) = x f ( x ) = x x 1 f ( f ( x ) ) = x x 1 x x 1 1 = x . \begin{array}{lcl} f(x) = \frac{1}{x} & \implies & f\big(f(x)\big) = \frac{1}{\hspace{2mm} \frac{1}{x}\hspace{2mm} } = x \\ \\ f(x) = -x & \implies & f\big(f(x)\big) = -(-x) = x \\ \\ f(x) = \frac{x}{x-1} & \implies & f\big(f(x)\big) = \frac{\frac{x}{x-1}}{\frac{x}{x-1}-1} = x. \end{array}

How many of these special types of functions are there?

Just these 3 More than 3 but less than 10 There are infinitely many

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

20 solutions

Henry U
Nov 16, 2018

The function f ( x ) = a x f(x) = \frac ax is an example for all real a 0 a \neq 0 since

f ( f ( x ) ) = a a x = a x a x x = a x a = x f(f(x)) = \frac a {\frac a x} = \frac {a {\color{#D61F06} \cdot x}} {\frac a x {\color{#D61F06} \cdot x}} = \frac {ax} a = x .

This is incorrect, as this is only 1 type of function. The problem is asking for how many types there are.

Blan Morrison - 2 years, 6 months ago

Log in to reply

I see that my solution doesn't answer this question completely (partly because the wording has been changed) but it's also dificult to define a type of funtion.

sin x \sin x is obviously a type , but are cos x , tan x , \cos x, \tan x, \ldots the same type "trigonometric functions" or are sin ( 2 x ) , sin ( x 2 ) , sin ( 1 x ) , \sin (2x), \sin \left( x^2 \right), \sin \left( \frac 1x \right), \ldots the type "derived from sine"? And what about sin ( sin x ) , sin ( sin ( sin x ) ) , \sin (\sin x), \sin( \sin( \sin x)), \ldots . Are they the type "composition of sine" or every one a new type "n-times composition"? Are polynomials one type or distinguished by their degree?

Henry U - 2 years, 6 months ago

Log in to reply

See @Brian Moehring 's solution to see the different types there are.

Blan Morrison - 2 years, 6 months ago

Log in to reply

@Blan Morrison His solution is by far more general, but I still think that type is a very vague phrase. All his functions can be seen as the type even function , but of course they're of many different types like squareroots, trigonometric functions, etc.

Henry U - 2 years, 6 months ago

Log in to reply

@Henry U This is true. Your solution provides one of infinite subsets, with each subset containing an infinite amount of functions.

Blan Morrison - 2 years, 6 months ago

If x/(x-1) is a different type than 1/x, then surely is (x-1)/x, (x+1)/x, x^i/x, log 78 /x...........

The thing is that replace <the numerator:a> with any expression, as layer on it will get cancelled by itself!

Anirudh Sharma - 2 years, 3 months ago

a is constant number, hence we can change it and can get infinite function

UGESHWAR YADAV - 2 years, 1 month ago

Then, define a "type" of function. I'm fine with he's example.

Antoine Quere - 2 years, 4 months ago
Levi Walker
Nov 15, 2018

Consider a new function g ( x ) = h ( f ( h 1 ( x ) ) ) g(x) =h(f(h^{-1}(x))) , with f ( x ) f(x) being an involution and h ( x ) h(x) being any function with an inverse. g ( g ( x ) ) = h ( f ( h 1 ( h ( f ( h 1 ( x ) ) ) ) ) ) = h ( f ( f ( h 1 ( x ) ) ) ) = h ( h 1 ( x ) ) = x g(g(x)) = h(f(h^{-1}(h(f(h^{-1}(x)))))) = h(f(f(h^{-1}(x)))) = h(h^{-1}(x)) = x Thus, g ( x ) g(x) is also an involution. Since there are infinitely many invertible functions, there must be infinitely many involutions.

Note that this shows the set of involutions is closed under conjugation. This is definitely an interesting fact.

However, your conclusion requires the map h h f h 1 h \mapsto h\circ f\circ h^{-1} to be finite-to-one [for at least one f ]. f\text{].} This is probably true, but it isn't implied by your solution.

Brian Moehring - 2 years, 6 months ago

Log in to reply

Yeah, that's a hole that I should probably fix. f ( x ) = a x f(x) = a-x is one such involution with no singular points, but of course there are others.

Levi Walker - 2 years, 6 months ago

Now, this a solution! How did you think of this???

Department 8 - 2 years, 6 months ago

That's a really nice solution, but how do you know that this algorithm creates infinitely many distinct(!) functions?

For example: f(x) = -x and h(x) = 2*x do not create a new function.

Konrad Knatz - 2 years, 6 months ago

Log in to reply

That's a good question - f ( x ) = x f(x) = -x wouldn't generally work, but f ( x ) = a x f(x) = a - x will, with the additional constraint that h ( x ) h(x) isn't linear - for example a x 2 , l n ( a e x ) \sqrt{a - x^2}, ln(a - e^x) , and 1 a 1 / x \frac{1}{a - 1/x} are all involutions.

Levi Walker - 2 years, 6 months ago
Brian Moehring
Nov 15, 2018

A one-dimensional involution can be interpreted graphically as any function which is symmetric over the line y = x y=x . For instance, you can take any continuously differentiable even function g ( x ) g(x) such that g ( x ) < 1 |g'(x)| < 1 for every x x . Then rotate its graph 4 5 45^\circ clockwise and you'll end up with the graph of an involution.

Using this construction, the functions given in the problem are g ( x ) = a 2 f ( x ) = a x g ( x ) = 2 + x 2 f ( x ) = 1 / x g ( x ) = 2 + x 2 + 2 f ( x ) = x x 1 g(x) = \frac{a}{\sqrt{2}} \implies f(x) = a-x \\ g(x) = \sqrt{2 + x^2} \implies f(x) = 1/x \\ g(x) = \sqrt{2+x^2} + \sqrt{2} \implies f(x) = \frac{x}{x-1}

It should be apparent that there are infinitely more g ( x ) . g(x). For instance, we could use

  • g ( x ) = a + x 2 + b g(x) = \sqrt{a+x^2} + b where a > 0 a>0
  • g ( x ) = a cos ( b x ) + c g(x) = a\cos(bx) + c where a b < 1 |ab| < 1
  • etc.

Very interesting! Although, how do we know that the first paragraph is true?

Blan Morrison - 2 years, 6 months ago

Log in to reply

I made a few comments in the first paragraph, but most of them are common or at least direct to show. For instance,

  • The graph of an even function has x = 0 x=0 as a line of symmetry, so the relation formed by rotating the graph 4 5 45^\circ clockwise has y = x y=x as a line of symmetry.
  • f f is an involution, if and only if ( x , y ) f ( y , x ) f (x,y) \in f \implies (y,x) \in f , if and only if the graph of f f is symmetric over y = x y=x

That leaves only one remaining claim: If g ( x ) < 1 for all x , then rotating the graph of g by 4 5 clockwise yields the graph of a function. \text{If } |g'(x)| < 1 \text{ for all } x \text{, then rotating the graph of } g \text{ by } 45^\circ \text{ clockwise yields the graph of a function.} To show this, suppose the resulting graph is not the graph of a function, so that there are two points that lie on the same vertical line. Rotating this 4 5 45^\circ counterclockwise shows two points on the graph of g g lie on a line with slope 1 , -1, which by the mean value theorem implies g ( x 0 ) = 1 g'(x_0) = -1 for some x 0 . x_0. This provides your contradiction.

Brian Moehring - 2 years, 6 months ago

This is somewhat more complicated than the typical basic level questions.

Malcolm Rich - 2 years, 6 months ago
Jeremy Galvagni
Nov 16, 2018

The simplest function not on your list is the identity function: f ( x ) = x f(x)=x

Any function that has y = x y=x as a line of symmetry also works.

I believe you mean "any function that has a line of symmetry of y = x y=x ," as f ( x ) = x 2 f(x)=x^2 is not an involution.

Blan Morrison - 2 years, 6 months ago

Log in to reply

Any function that has this as a line of symmetry also works. "this" being the line defined by f ( x ) = x f(x)=x . I will clarify.

Jeremy Galvagni - 2 years, 6 months ago

Any homographic function f ( x ) = a x + b c x + d f(x)=\dfrac{{\color{#3D99F6}{a}}x+b}{cx+{\color{#D61F06}{d}}} , ( c 0 , a d b c ) \quad (c\ne0, \; ad\ne bc) , is an involution, iff a + d = 0 {\color{#3D99F6}{a}}+{\color{#D61F06}{d}}=0 .

Blan Morrison
Nov 15, 2018

Consider the function f ( x ) = x + a x 1 f(x)=\frac{x+a}{x-1} , where a a is any real. Now, find f ( f ( x ) ) f(f(x)) : f ( f ( x ) ) = x + a x 1 + a x + a x 1 1 f(f(x))=\dfrac{\frac{x+a}{x-1}+a}{\frac{x+a}{x-1}-1} Multiplying a a (in the numerator) and -1 (in the denominator) by x 1 x 1 \frac{x-1}{x-1} : x + a x 1 + a x a x 1 x + a x 1 + x + 1 x 1 = x ( a + 1 ) x 1 a + 1 x 1 = x ( a + 1 ) a + 1 = x \dfrac{\frac{x+a}{x-1}+\frac{ax-a}{x-1}}{\frac{x+a}{x-1}+\frac{-x+1}{x-1}}=\dfrac{\frac{x(a+1)}{x-1}}{\frac{a+1}{x-1}}=\frac{x(a+1)}{a+1}=x Therefore, since a a can be any real, there are uncountably infinitely more solutions. β \beta_{\lceil \mid \rceil}

If you are considering that infinite a a 's = infinite functions, wouldn't that mean that the question setter's first function itself concludes infinite functions?

Parth Sankhe - 2 years, 6 months ago

Log in to reply

The question asks for how many more functions, not how many.

Blan Morrison - 2 years, 6 months ago

Log in to reply

If that is the thing, then why did you give f ( f ( x ) ) = x + a x 1 f(f(x))=\frac{x+a}{x-1} , you could have given the function f ( x ) = 1 x f(x)=1-x or f ( x ) = 2 x f(x)=2-x and so on.. they are infinite in number... but in the question all functions of the type f ( x ) = a x f(x)=a-x are considered as 1.

And the example you are giving is also a function which just represents one particular type... it doesn't proves that there are infinite functions...

Vilakshan Gupta - 2 years, 6 months ago

Log in to reply

@Vilakshan Gupta

  1. f ( x ) = a x f(x)=a-x is already given, so that won't prove anything else.

  • It depends on how you define function.

  • @Henry U gave a similar example as mine, and no one is disputing him.

  • Blan Morrison - 2 years, 6 months ago

    Log in to reply

    @Blan Morrison Well, my point is... even if you give 100 or even 1000 more functions which satisfy f ( f ( x ) ) = x f(f(x))=x , that doesn't mean there are infinite functions until and unless you prove it. The solution with a proof can be considered as of Brian Moehring and Levi Adam Walker.

    Vilakshan Gupta - 2 years, 6 months ago
    Kuuaebo Cowou
    Nov 27, 2018

    for any a f(x)=a-x is an example

    f(x)=a-x

    f(f(x))= a-(a-x)) = a-a+x = x

    Plum Bum
    Dec 1, 2018

    Suppose we have such a function, f ( x ) f(x) . As is commonly done to functions, let's look at this one as a relationship on the x-y plane. f ( x ) f(x) being an involution means that for every x x we plug in, we can take the corresponding y y , plug it back in, and out comes that same x x value we started with. If you pick enough x x 's and make up enough y y 's... You might notice some kind of mirror symmetry in the picture. But all we need in order to argue that there's infinitely many involutions is this: Imagine the function being built two points at a time, starting at say, x = 0 x=0 . We can make up any corresponding y y value for x = 0 x=0 . And as soon as we get the point ( 0 , s o m e t h i n g ) (0, something) , we also get a second point ( s o m e t h i n g , 0 ) (something, 0) . We move a little bit to the left of 0, getting two new points. Then we move a little bit to the right of 0, getting two more points. We do this for every number on the x x axis, and we have our involution. But since we made up each y y value, we had infinitely many choices for a function, no matter how strange-looking.

    Tricia Melia
    May 12, 2019

    The family of functions f(x)=a^(1/(log base a of x)) are all involutions, since a^1/log base a of (a^1/log base a of x)=a^log base a of x=x.

    Aijaz Bhat
    Dec 1, 2018

    In mathematics there exists infinite no. Of invertible functions. Here are just pool of three enlisted.

    Deepak Maurya
    Nov 30, 2018

    In fact h(f(x))=x holds if the function h is inverse of f. Inverse functions have property that they are symmetric about y=x.Here we have f in place of h so any function which is symmetric about y=x will hold this property.

    Emmanuel Moysis
    Nov 27, 2018

    I don’t know this type of math. I guessed the right answer :P hahahahahahahaaha.

    Peter Macgregor
    Nov 26, 2018

    If a function is an involution, then (from the definition given in the problem) it is its own inverse.

    Remember that the graphs of a function and its inverse are mirror images in the line y = x y=x

    So any function which is symmetrical in the line y = x y=x is its own inverse, and therefore is an involution.

    There are infinitely many such graphs. For example, any of the lines perpendicular to y = x y=x will do, so

    y = x + a y=-x+a represents an infinite family of invloutions.

    Let's illustrate this with a specific example. Let's choose y ( x ) = x + 3 y(x)=-x+3 and x = 7 x=7 . Then

    y ( y ( 7 ) ) = y ( 7 ) + 3 = ( 7 + 3 ) + 3 = ( 4 ) + 3 = 4 + 3 = 7 y(y(7))=-y(7)+3=-(-7+3)+3=-(-4)+3=4+3=7

    Since we have found an infinite family of involutions (and there other such families) we can certainly say

    There are an infinite number of involutions \boxed{\text{There are an infinite number of involutions}}

    N Kansara
    Nov 26, 2018

    All functions f ( x ) = 1 x k f(x) = \frac {1}{x^k} are example of it

    Are you sure? What about f ( x ) = 1 x 2 f(x) = \frac{1}{x^2} ? Notice that f ( f ( 1 ) ) > 0 f( f( -1) ) > 0 .

    Calvin Lin Staff - 2 years, 6 months ago
    Nicolas Guignes
    Nov 26, 2018

    For any A R + A \subset \mathbb{R}_{+} , the function f f defined by f ( x ) = x f(x) = -x if x A x \in A or x A -x \in A and by f ( x ) = x f(x) = x otherwise, is an involution.

    Nathan Oupresque
    Nov 25, 2018

    Just take

    • f(x) = x for all x exept for :

    • f(a) = b

    • f(b) = a

    Ben Richardson
    Nov 25, 2018

    The first function 1/x will also work with any value (e.g. 2/x, 3/x, 4/x ....)

    The question asks for how many types , not just functions. You have only listed 1 type.

    Blan Morrison - 2 years, 6 months ago
    Louis Ullman
    Nov 22, 2018

    A function, f ( x ) f(x) , is another function, g ( x ) g(x) 's inverse iff you get the graph of g ( x ) g(x) when you reflect the graph of f ( x ) f(x) over the line x = y x=y . This means that for a function to be an involution, it has to remain the same when reflected over the line x = y x=y . It is obvious that there are infinitely many graphs of this type that aren't of the form a x a-x . Therefore, there are infinitely many more involutions than the examples provided.

    Abd Fml<3
    Dec 7, 2018

    If f ( x ) = n f(x)=n then f ( f ( x ) ) = f ( n ) f(f(x))=f(n) wich is n n .

    Patience Patience
    Nov 27, 2018

    Root of x ^2

    0 pending reports

    ×

    Problem Loading...

    Note Loading...

    Set Loading...