f ( x y ) = f ( x ) + f ( y )
The function f : R + → R with domain the positive real numbers and codomain the real numbers satisfies the above equation for all positive reals x , y . In addition, f is continuous at x = 1 , and f ( 3 ) = 7 .
Determine whether f is necessarily differentiable at 1 . If it is, compute f ′ ( 1 ) . If it isn't, simply input 0 as your answer.
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.
This is marvelous! I was skeptical at first when you intend to prove that it's differentiable everywhere because I thought it's an impossible feat. And at first, I was confused why you prove the case for all natural numbers n , and all rational numbers r because they seem so unrelated to the question at hand and then you pull off the impossible! Thanks for the fantastic question and answer!
Log in to reply
Thank you, Pi Han Goh. At the beginning, I was doubting whether or not to post this question, because I didn't not know how to formulate the question in a way that a potential solver had to think a little deeper and not just to guess that the function had to be a logarithm and to use it to find a value. This risk exists anyway but your comment tells me that the right thing was to post it. By the way, with the experience that you have in brilliant with near 8000 problem solved, I appreciate your comment even more!
Note that all we need is continuity at a point, and from that we can prove that this function is continuous everywhere by scaling.
Log in to reply
Thank you, Calvin. I edited the question and my answer to include your suggestion.
The Cauchy functional equation g ( x + y ) = g ( x ) + g ( y ) x , y ∈ R is a well-known problem in analysis. It is a standard piece of bookwork (proved very similarly to the way you worked above!) to show that any solution of the Cauchy functional equation which is one of: (1) continuous at some point, (2) bounded on some interval, (3) monotonic, must be of the form α x for a constant α .
Given our function f , the function g ( x ) = f ( e x ) must be a solution of the CFE which is continuous at 0 , and hence g ( x ) = α x , so that f ( x ) = α ln x , making f ′ ( 1 ) = α = 7 / ln 3 .
It is, on the whole, slightly easier to prove continuity with the fully additive function g than it is for f . Similarly, it is an easy induction that g ( q ) = q g ( 1 ) for any rational q ; after that continuity finishes things off!
Log in to reply
You are right! This is just a basic example that I thought had to present in brilliant!
Your mind is like a genuine creator which does not presume thing without proper resources. Your elaborations here give me a feeling of rare and patient.
Log in to reply
Thank you @Lu Chee ket! I really appreciate your comment.
Log in to reply
Dear Arturo Presa,
Tan 2 x and Log x are analysed in the way as shown above. Seemingly they are the only functions to respective properties being specified; did you find any others? Can you introduce special or new but quite typical functions which suppose to be important from your work efforts which you think that many people may not have ever known them before?
Log in to reply
@Lu Chee Ket – Thank again for your interest! The two questions that I have posted that you mentioned before are typical problems based on functional equations. The natural question here would be to find the function, or all possible functions satisfying the equation and the additional conditions, but due to the format of brilliant.org, in general one should ask the question in a way that the answer is multiple choice or just a number. My only merit, if there is any, is to have modified two old functional equations in a way that makes the solver to solve them to be able to find the answer. In contests and olympiads this kind of question is recurrent. The following is an example:
IMO 1983/1. Find all functions f defined on the set of positive real numbers which take positive real values and satisfy the conditions: (i) f ( x f ( y ) ) = y f ( x ) for all positive x , y ; (ii) f ( x ) → 0 as x → ∞ .
By the way, you can find an elementary (but interesting) introduction to Functional Equations here in brilliant.
Log in to reply
@Arturo Presa – Another website that you can read is imomath.com (Functional Equations)
@Arturo Presa – The topic concerning these is not pleasantly easy to my feeling unless I do have plenty of interest to think about them. All possible means there could valid with something else, which is my main interest. My curiosity is whether you do obtain anything quite valuable because I believe this is the motive why we want to get headache by getting into many seemingly trivial concerns or worries for whether there can be anything else other than the most immediate or most obvious that we can guess about. I am asking for a fruit from plenty of efforts.
My solution is in no way half as good as the other solutions, but I just want to share it.
Putting x = y = 3 into the equation would give us f ( 3 ) = 2 f ( 3 ) , which gives us f ( 3 ) = 7 / 2 .
Then, induction can be used to prove the following statement: f ( 3 1 / 2 n ) = 2 n 7
Since 3 1 / 2 n approaches 1 and 2 n 7 approaches 0 and it is continuous, this can be used to compute f ′ ( 1 ) .
The limit sign will be omitted.
f ′ ( 1 )
= 3 1 / 2 n − 3 1 / 2 n + 1 7 / 2 n − 7 / 2 n + 1
= 3 1 / 2 n − 3 1 / 2 n + 1 7 / 2 n + 1
= 2 7 3 1 / 2 n − 3 1 / 2 n + 1 1 / 2 n
= 2 7 3 1 / 2 n ( 1 − 3 − 1 / 2 n + 1 ) 1 / 2 n
= 2 7 1 − 3 − 1 / 2 n + 1 1 / 2 n
= 2 7 − 3 − 1 / 2 n + 1 ln 3 ( 1 / ( 2 n + 1 ) 2 ) ( ln 2 ) ( 2 n + 1 ) ln ( 1 / 2 ) × ( 1 / 2 n )
= 2 7 3 − 1 / 2 n + 1 ln 3 ( 1 / ( 2 n + 1 ) 2 ) ( 2 n + 1 ) 1 / 2 n
= 2 7 3 − 1 / 2 n + 1 ln 3 ( 1 / 2 n + 1 ) 1 / 2 n
= 7 3 − 1 / 2 n + 1 ln 3 1
= ln 3 7
The limit from the negative side can be checked by a similar method, based on f ( 1 / 3 ) = f ( 1 ) − f ( 3 ) = − 7 .
I found your way very interesting!
Unfortunately, you have not proven that the function is differentiable. All that you have shown, is that for a very specific sequence of points p i , we get that
lim p i − 1 f ( p i ) − f ( 1 ) = ln 3 7 .
A function is differentiable (at 1) if and only if the limit is constant for all possible sequences that tend to 1. That is harder to establish, and you will need to use the condition of continuity, in order to rigorously push through.
As an explicit counter example, the function f ( x ) = x if x is rational and f ( x ) = − x if x is irrational, is continuous at 0 but f ′ ( 0 ) clearly doesn't exist. However, we can pick a sequence of (rational) points, such that the limit of the difference is 1.
Let f ( x ) = g ( lo g x ) , so that f and g are defined over positive real numbers. If f is continuous, then g is also continuous.
We get f ( x y ) = g ( lo g x y ) = g ( lo g x + lo g y ) [1]
and f ( x ) + f ( y ) = g ( lo g x ) + g ( lo g y ) [2]
[1] = [2]; g ( lo g x + lo g y ) = g ( lo g x ) + g ( lo g y )
Substituting x as e x and y as e y we get
g ( x + y ) = g ( x ) + g ( y ) .
Given that g is continuous, g satisfies Cauchy's Functional Equation.
We get g ( x ) = c x for some constant c .
Which gives f ( x ) = g ( lo g x ) = c lo g x .
Solve for c we get c = lo g 3 7 .
Therefore, f ′ ( x ) = x c
At x = 1 , f ′ ( 1 ) = c = lo g 3 7 . ~~~
Wonderful!
If one differentiates the above functional equation with respect to x and y, one will obtain:
y
f'(xy) = f'(x) (i),
x
f'(xy) = f'(y) (ii).
and equating (i) with (ii) gives:
f'(x)/y = f'(y)/x, or x f'(x) = y f'(y) = A (real constant) (iii)
and integrating (iii) with respect to x yields:
f(x) = A*ln(x), f(3) =7.
which finally results in f(x) = (3/ln(7))*ln(x).
Certainly f(x) is differentiable at x = 1, and f'(1) = (3/ln(7)) * (1/1) = (3/ln(7)) = 6.372.
PS: Great proof, Arturo Presa!
At the start of this solution, you are assuming that the functional equation is differentiable, and it is not immediately obvious why this is true. Also, you can't then conclude "Certainly f(x) is differentiable at x = 1 " since that is part of your assumption.
In fact, if there wasn't the condition of continuity, then there are non-differentiable functions that satisfy the functional equation.
Very nice ideas! Thank you!
At the start of this solution, you are assuming that the functional equation is differentiable, and it is not immediately obvious why this is true. Also, you can't then conclude "Certainly f(x) is differentiable at x = 1 " since that is part of your assumption.
In fact, if there wasn't the condition of continuity, then there are non-differentiable functions that satisfy the functional equation.
I came up with a very simple solution, although I don't think it is a rigorous proof:
When I saw that the function had the property of f ( x y ) = f ( x ) + f ( y ) , I thought that a logarithm function would have that property.
So I defined f ( x ) = a ln x
f ( 3 ) = 7 , so a = ln 3 7
f ( x ) = ln 3 7 ln x , and so f ′ ( x ) = x ln 3 7
f ′ ( 1 ) ≈ 6 . 3 7 1 6 7
Problem Loading...
Note Loading...
Set Loading...
By making x = y = 1 into the given functional equation we get that f ( 1 ) = 0 . Then you can see that if x is any positive real number and y = x 1 , we obtain that f ( x ) + f ( x 1 ) = f ( 1 ) = 0 and therefore f ( x 1 ) = − f ( x ) . ( 1 ) Using (1) and the functional equation we get that for any real numbers x and a the following equality holds: f ( x ) − f ( a ) = f ( x ) + f ( a 1 ) = f ( a x ) = f ( a x ) − f ( 1 ) . Therefore the continuity of f at 1 implies the continuity at any positive number a .
Now let us prove by Mathematical Induction that for any positive number a and any natural number n the following is true f ( a n ) = n f ( a ) . ( 2 ) Indeed, making n = 1 we obtain that f ( a ) = f ( a ) . Assuming that f ( a k ) = k f ( a ) , and using the functional equation given in the problem, we get that f ( a k + 1 ) = f ( a k ∗ a ) = f ( a k ) + f ( a ) = k f ( a ) + f ( a ) = ( k + 1 ) f ( a ) . By replacing a by a n 1 in (2) we obtain that f ( ( a n 1 ) n ) = n f ( a n 1 ) , or equivalently, f ( a ) = n f ( a n 1 ) for any positive number a and any natural number n . Dividing both sides by n we obtain f ( a n 1 ) = n 1 f ( a ) ( 3 )
Now, we can prove that for any real number r , f ( a r ) = r f ( a ) . ( 4 ) First we are going to assume that r is a positive rational number, so it can be written like m n , where m and n are natural numbers and then using (2) and (3) f ( a m n ) = f ( ( a m 1 ) n ) = n f ( a m 1 ) = m n f ( a ) . The equality (4) is also going to be true in the case the r is a negative rational number. Actually, if r is a negative rational number then using (1) and assuming that (4) is true for positive rational numbers, we obtain that f ( a r ) = − f ( a − r ) = − ( − r ) f ( a ) = r f ( a ) . The case when r = 0 it is easily derived from the fact that f ( 1 ) = 0 . To prove that (4) is true for any real number r it is enough to consider a sequence of rational numbers ( r n ) n , such that lim n → ∞ r n = r . Then taking limits in both sides of f ( a r n ) = r n f ( a ) and using the continuity of f ( x ) we obtain that f ( a r ) = r f ( a ) . This completes the proof of (4).
The rest is very simple. If x is any real positive number, from (4) we obtain that f ( x ) = f ( 3 lo g 3 x ) = ( lo g 3 x ) f ( 3 ) = 7 lo g 3 x . Therefore, f ( x ) is differentiable at any positive number and f ′ ( 1 ) = ln 3 7 ≈ 6 . 3 7 .