Function-3

Algebra Level 4

How many functions f : R R f:\mathbb{R}\rightarrow \mathbb{R} exists such that :

2 f ( x ) + 2 f ( y ) = f ( x + y ) + f ( x y ) + 2 y 2f(x)+2f(y)=f(x+y)+f(x-y)+2y

holds for all real x x and y y ?

0 1 2 3 4 More than 4, but less than 10 More than 10 but finitely many 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.

1 solution

Tom Engelsman
Feb 20, 2021

Upon observation, this functional equation resembles the famed Cauchy Equation (which admits linear functions as its solution). If f ( x ) = p x + q f(x)=px+q , then we have:

2 ( p x + q ) + 2 ( p y + q ) = p ( x + y ) + q + p ( x y ) + q + 2 y 2(px+q) + 2(py+q) = p(x+y)+q + p(x-y)+ q + 2y ;

or 2 p x + 2 p y + 4 q = 2 p x + 2 y + 2 q 2px + 2py + 4q = 2px + 2y + 2q ;

or 2 p y + 4 q = 2 y + 2 q p = 1 , q = 0 2py + 4q = 2y + 2q \Rightarrow p = 1, q = 0

which gives us the unique linear function f ( x ) = x . \boxed{f(x) = x}.

But were not done yet! If we have x = y = 0 x=y=0 , then we obtain 2 f ( 0 ) + 2 f ( 0 ) = f ( 0 ) + f ( 0 ) + 2 ( 0 ) f ( 0 ) = 0. 2f(0) + 2f(0) = f(0)+f(0) + 2(0) \Rightarrow f(0)=0. Now taking a leap-of-faith, let's assume f ( x ) f(x) is a differentiable function over all x , y R x,y \in \mathbb{R} . Differentiating the above functional equation wrt x x and y y yields:

2 f ( x ) = f ( x + y ) + f ( x y ) 2f'(x) = f'(x+y)+f'(x-y) (i)

2 f ( y ) = f ( x + y ) f ( x y ) + 2 2f'(y) = f'(x+y)-f'(x-y) + 2 (ii)

and adding (i) and (ii) next produces:

2 f ( x ) + 2 f ( y ) = 2 f ( x + y ) + 2 f ( x ) + f ( y ) = f ( x + y ) + 1 2f'(x) + 2f'(y) = 2f'(x+y) + 2 \Rightarrow f'(x)+f'(y)=f'(x+y)+1 (iii)

followed by differentiating (iii) wrt x x and y y :

f ( x ) = f ( x + y ) f''(x) = f''(x+y) (iv)

f ( y ) = f ( x + y ) f''(y) = f''(x+y) (v)

which leads to f ( x ) = f ( y ) = A f''(x) = f''(y) = A , and after integrating twice: f ( x ) = A 2 x 2 + B x + C f(x) = \frac{A}{2}x^2 + Bx + C (where A , B , C A,B,C are arbitrary real constants). We know f ( 0 ) = 0 C = 0 , f(0)=0 \Rightarrow C = 0, or f ( x ) = A 2 x 2 + B x f(x) = \frac{A}{2}x^2 + Bx (vi). If we now substitute (vi) into the original functional equation, we obtain:

2 ( A 2 x 2 + B x ) + 2 ( A 2 y 2 + B y ) = A 2 ( x + y ) 2 + B ( x + y ) + A 2 ( x y ) 2 + B ( x y ) + 2 y 2(\frac{A}{2}x^2 + Bx) + 2(\frac{A}{2}y^2 + By) = \frac{A}{2}(x+y)^2 + B(x+y) + \frac{A}{2}(x-y)^2 + B(x-y) + 2y ;

which expands & simplifies into:

A x 2 + A y 2 + 2 B x + 2 B y = A x 2 + A y 2 + 2 B x + 2 y A = k , B = 1 Ax^2 + Ay^2 + 2Bx + 2By = Ax^2 + Ay^2 + 2Bx + 2y \Rightarrow A = k, B = 1

where k R k \in \mathbb{R} . This gives the infinite family of functions f ( x ) = k x 2 + x \boxed{f(x) = kx^2 + x} , which includes our above linear function (when k = 0 k = 0 ), as the solution to the original functional equation.

Nice solution! I stopped when I found the infinite family of functions f ( x ) = k x 2 + x f(x)=kx^2+x (enough to answer the question). But are there any others? Or can you prove that there aren't?

Chris Lewis - 3 months, 3 weeks ago

Log in to reply

Thanks, Chris! This functional equation is rather limited on the type of functions that satisfy it (mainly due to the "2y" term on the RHS). Upon intuition, this rules out trigonometric, exponential, hyperbolic, & logarithmic as possibilities. This really leaves us with just the polynomial functions to check. The above differential equation technique (which is my most-preferred method for solving functional equations) really does the brute-force trick in finding the above infinite-parabola family.

I will confess, this method has had its share of Brilliant.org critics (especially when f(x) is not explicitly stated to be differentiable....hence, I wrote "leap-of-faith!") But nonetheless, it has worked for me on similar past problems.

tom engelsman - 3 months, 3 weeks ago

Log in to reply

100% agree that it's a solid way to solve these problems (and, as I said, it certainly solves this one; once there's an arbitrary coefficient, we're done). I wouldn't even call it a "leap of faith"; had you found, say, just one differentiable solution, then yeah, it'd be bold to then assume that that was the only one; but finding infinitely many means you don't have to search for any more, as far as this question goes.

However, I was wondering whether there might be more solutions; either belonging to a different functional family or else perhaps non-differentiable. Quite often functional equations like this have ghastly but valid non-continuous solutions.

Chris Lewis - 3 months, 3 weeks ago

In a case like this, where there are infinitely many twice continuously differentiable functions, then your technique works fine, because the extra assumption does not change the answer. However there are problems, such as the Cauchy problem itself, where there is only one continous solution, but infinitely many solutions of more general type.

In this case, you state the that 2 y 2y term limits the range of solutions. This is not true. As my other comment shows, that term can be removed by a trivial redefinition of the function, and we are left with what I called the function h ( x ) h(x) . If we assumed that f f was continuous, then h h would be continuous and hence identically zero. If we drop the requirement for continuity, then many more solutions appear!

It is always worth looking to see if this sort of problem can be solved without making additional assumptions. Harder, but more interesting.

Mark Hennings - 3 months, 2 weeks ago

If we write f ( x ) = g ( x ) + x f(x) = g(x) + x , then g g satisfies the functional equation 2 g ( x ) + 2 g ( y ) = g ( x + y ) + g ( x y ) 2g(x) + 2g(y) \; =\; g(x+y) + g(x-y) If we write h ( x ) = g ( x ) g ( 1 ) x 2 h(x) = g(x) - g(1)x^2 then we have h ( 0 ) = h ( 1 ) = 0 h(0)=h(1)=0 and 2 h ( x ) + 2 h ( y ) = h ( x + y ) + h ( x y ) 2h(x) + 2h(y) = h(x+y) + h(x-y) and it is easy to deduce that h ( q ) = 0 h(q)=0 for all rationals q q . Either we apply continuity to conlude that h 0 h \equiv 0 , or else we need to be more algebraic, and move over to a Hamel basis.

If A \mathcal{A} is a Hamel basis for R \mathbb{R} , so that every element x x of R \mathbb{R} can be written uniquely as a finite sum of the form x = a A q a a x \; = \; \sum_{a \in \mathcal{A}} q_a\,a where q a Q q_a \in \mathbb{Q} for all a A a \in \mathcal{A} , then we can define f ( x ) = x + a A F ( a ) q a 2 f(x) \; = \; x + \sum_{a \in \mathcal{A}} F(a)q_a^2 for any function F : A R F \,:\, \mathcal{A} \to \mathbb{R} . Thus there are a huge number of functions - we have a massive choice of Hamel basis, and a massive choice of functions F F .

Mark Hennings - 3 months, 2 weeks ago

Log in to reply

Nice solution, Mark....Hamel Bases, ay?!

tom engelsman - 3 months, 2 weeks ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...