6

Let $f:\mathbb{N}^{+}\to \mathbb{N}^{+}$ be a strictly monotone increasing function such that $$f(f(m+1))-f(f(m))=f(f(m+1)+1)-f(f(m)+1),\forall m\in \mathbb{N}^{+}.$$

Show that: $$f(m+1)-f(m)=\text{ constant},\forall m\in \mathbb{N}^{+}$$

My attempt: Let $g(m)=f(f(m))-f(f(m)+1)$, then the condtion is $g(m+1)=g(m)$, so $g(m)$ is constant, or $f(f(m))-f(f(m)+1)$ be constant, then how prove it?

kimchi lover
  • 24,277
math110
  • 93,304
  • Why is the condition $g(m+1) = g(m)$? edit: I think the condition $$f(m+1)-f(m) = \text{ constant }$$ implies that $g(m+1) = g(m)$, but they are not equivalent - I don't think the implication runs the other way, I don't think that $g(m+1) = g(m)$ implies that $f(m+1)-f(m) = \text{ constant}$? Or does it in some way that I'm missing? – Zubin Mukerjee Oct 29 '17 at 14:45
  • 1
    Where did you get this question? –  Nov 06 '17 at 18:31

4 Answers4

5

Here is a counter-example:

Define $f:\mathbb{N}^+\rightarrow\mathbb{N}^+$ by the following rule

$$f(1) = 1,$$$$f(2) = 3,$$$$ f(k) = 2k-2, \forall k\geq 3.$$

Then, $\forall k \geq 3$,

$f(f(k+1))-f(f(k)) = f(2k) - f(2k-2) = 4k-2-(4k-6) =4\\ f(f(k+1)+1) - f(f(k)+1)=f(2k+1)-f(2k-1)=4k-(4k-4)=4.$

So the stated condition holds $\forall k \geq3$. It's easy to check that it also holds for $k=1,2$.

Furthermore, $f$ is strictly monotone. Hence satisfies the statement.

However,$$ f(3)-f(2) = 1 \neq 2 = f(2)-f(1).$$

Nyrune
  • 109
  • 4
1

So we are to prove that, given $f:\;N^ + \to N^ + $, we have $$ \bbox[lightyellow] { \eqalign{ & f\left( {f\left( {m + 1} \right)} \right) - f\left( {f\left( m \right)} \right) = f\left( {f\left( {m + 1} \right) + 1} \right) - f\left( {f\left( m \right) + 1} \right)\quad \left| {\;\forall m \in N^ + } \right.\quad \Rightarrow \cr & \Rightarrow \quad f\left( {m + 1} \right) - f\left( m \right) = \Delta \,f\left( m \right) = const\quad \left| {\;\forall m \in N^ + } \right. \cr} }$$ which means $$ \bbox[lightyellow] { \eqalign{ & f\left( {f\left( {m + 1} \right)} \right) - f\left( {f\left( m \right)} \right) = f\left( {f\left( {m + 1} \right) + 1} \right) - f\left( {f\left( m \right) + 1} \right) \cr & \quad \quad \Downarrow \cr & f\left( {f\left( m \right) + 1} \right) - f\left( {f\left( m \right)} \right) = f\left( {f\left( {m + 1} \right) + 1} \right) - f\left( {f\left( {m + 1} \right)} \right) \cr & \quad \quad \Downarrow \cr & \Delta \,f\left( {f(m)} \right) = \Delta \,f\left( {f(m + 1)} \right) = \Delta \,f\left( {f(m) + \Delta \,f(m)} \right) \cr} } \tag{1}$$

In the post it is further specified that $f$ is
a strictly monotone increasing function.
and according the definition of monotonic function as per Wikipedia
a) either the monotone specification is superfluous, and the function is just strictly increasing,
b) or it is actually meant that it is strictly increasing and with monotone increase.

So let's examine both cases.

a) strictly increasing

The image of $f(m)$ will not coincide, in general, with its domain. So from (1) we cannot deduce the claim in general and therefore we shall reject the hypothesis.

For example, a function like this one $$ \begin{array}{c|ccccc} m & 1 & 2 & 3 & 4 & \cdots \\ \hline {f(m)} & 1 & 3 & 6 & 8 & \cdots \\ \end{array} $$ gives $$ \begin{array}{l} f(f(2)) - f(f(1)) = f(3) - f(1) = 5 = \\ = f(f(2) + 1) - f(f(1) + 1) = f(4) - f(2) = 5 \\ \end{array} $$ but $$ f(2) - f(1) = 2\quad \ne \quad f(3) - f(2) = 3 $$

b) strictly increasing and with monotone increase

In this interpretation it is meant that $$ \bbox[lightyellow] { \eqalign{ & m_{\,1} < m_{\,2} \quad \Rightarrow \cr & \Rightarrow \quad \left\{ \matrix{ f(m_{\,1} ) < f(m_{\,2} ) \hfill \cr \left( {\Delta \,f(m_{\,1} ) \le \Delta \,f(m_{\,2} )} \right)\; \hfill \cr} \right.\quad \vee \quad \left\{ \matrix{ f(m_{\,1} ) < f(m_{\,2} ) \hfill \cr \left( {\Delta \,f(m_{\,1} ) \ge \Delta \,f(m_{\,2} )} \right) \hfill \cr} \right.\quad \Rightarrow \cr & \Rightarrow \quad \left( {0 < \Delta \,f(m_{\,1} ) \le \Delta \,f(m_{\,2} )} \right)\;\quad \vee \quad \Delta \,f(m_{\,1} ) \ge \Delta \,f(m_{\,2} ) > 0 \cr} } \tag{2}$$

This tells us that $\Delta f$ is either non-decreasing or non-increasing, and identity (1) that in various points it is stable, thus in this acception the claim follows immediately.

G Cab
  • 35,272
  • To the anonymous downvoter : since we are here to give our best contribute, and to learn from each other nice ideas and faults as well, it would be nice to leave a minimal justification of where the fault is. – G Cab Nov 05 '17 at 23:14
-4

BIG FAT EDIT:

I will build a counterexample to the problem. The problem is to prove that a function satisfying $f(f(m)+1)-f(f(m))=constant=K (A)$ implies $f(m+1)-f(m)=constant=C (B)$ because the original statemtn is equivalent to (A).

Lets start building the function and see what intuition we can gain in the process. Once we determine a value for 1, this is $f(1)$ shall we call it $a$ then if a>2 we have all the values till a that are determined by nothing, but once we put a alue to a we get a value for a+1. So we need to know which values of the naturals we are giving to the function and register them in order to remember when we will need to satisfy (A).

Lets say $a=3$ and also $K=3$ $f(2)=5$ and $f(3)=7$, I assigned those because they were free, but $f(4)=10$ yet we dont mind if $f(5)=12$ because this doesnt exclude $f(6)-f(5)=3$ (I care about 5 because it belongs to the span of the function) so we get the thing here. We only need to give consistent values to the span of f so I conserve the necessary properties.

So, we register the span (3,5,7)once in $x=3$ we jump $y=3$ to make $f(4)=10$ (and add 10 to the span (3,5,7,10))but because 4 didnt beling to the range previously registered we dont need $f(4)+3=f(5)$ so to prove how great we are we actually say $f(5)=f(4)+2=12$ (and add 12, (3,5,7,10,12)) and then we can build a function so that in these new added "spans" we "jump" 3 in the y axis to preserve the property, (also happy that this preserves monotonicity) AND we can actually violate this for other naturals in the domain. Whenever we are not on the "spans" we only add 2 to the previous value creating a slope 2 line and we add a new span, when we get to one span we jump 3 to preserve the property.

This produces a broken line so the conclusion doesnt follow.

  • Try to present a proof first and then you might have solved it. Seriously, the text above is not remotely rigorous enough to pose as a proof. If you believe you have a good idea try to formulate it in clear and concise manner. PS I have not downvoted you but as it stands this is not an answer. – MathematicianByMistake Nov 01 '17 at 22:26
  • I did give an example. Its as specific as it can be. – Big Hands Nov 01 '17 at 22:28
  • im really excited i really think i solved it :) check the example and tell me what you think. if you can find an error i will recognize it. if not, i need that bounty :V – Big Hands Nov 01 '17 at 22:34
  • Excitement is not a proof of a proof so first things first-Explain your first line:"Lets say $f(1)=a$, so as long as $a−1>1$ then only $a$ determines the subsequent value" – MathematicianByMistake Nov 01 '17 at 22:39
  • well, there ya go. Lets focus on the example, the first paragraph is the motivation. Let $a$ be 3. we have got no value for 2 neither for 3 but we know once in 3 we get a value for 4. so lets say the value for 2 is 5 and for 3 is 7, a line of slope 2, comment to know you are listening :) and then continue – Big Hands Nov 01 '17 at 22:41
  • You are just assigning random values and then articulating phrases that make no sense. Perhaps its a language barrier but in any case this is not a proof. – MathematicianByMistake Nov 01 '17 at 22:47
  • man, i mean, im building the function that is the counterexample. Im starting: f(1)=3, f(2)=5, f(3)=7, shall we continue? – Big Hands Nov 01 '17 at 22:49
  • "man" does your function satisfy the assumptions of the problem?$f(f(m+1))-f(f(m))=f(f(m+1)+1)-f(f(m)+1),\forall m\in \mathbb{N}^{+}$ – MathematicianByMistake Nov 01 '17 at 22:53
  • are you trying to mock with me? it does :V. if you let me prove it you would realize it. Of course i didnt list all the function, did you think so?. but just for m=1 we have $f(m)=3$ $f(m+1)=5$ $f(f(m))=7$ $f(f(m+1))=12$ $f(f(m+1)+1)=15$ $f(f(m)+1)=10$ $$ – Big Hands Nov 01 '17 at 22:58
  • "if you let me prove it" I thought that thing above was the proof. What does it even mean to "list all the function"? You have a closed type that satisfies the properties but contradicts the required result? If so post it. You are just posting random values.. – MathematicianByMistake Nov 01 '17 at 23:01
  • im not! but its not the type of thinf that springs to the naked eye. im explaining how it is built but you are not even trying. it is better for me to know you are understanding what im saying. ill rewrite my answer, but damn, guys in this site should know maths is not a spectator sport. – Big Hands Nov 01 '17 at 23:03
  • I’m too tired and sleepy now to perfectly check the details of your not (so simple and clear) construction, but at least it looks (very) promising and plausible, so I upvoted it. – Alex Ravsky Nov 02 '17 at 02:28
  • Be respectful, focus on the maths and not on insulting others. – MathematicianByMistake Nov 02 '17 at 22:56
  • @BigHands: your intuition is right (see also the example in my answer), and +1 for that, but it is somewhat difficult to follow in your exposition: try and make it simpler and more "canonic". – G Cab Nov 06 '17 at 15:56
  • Massive revive but this is a well done proof. its a shame it was recognized properly. –  Jan 15 '18 at 19:51
-4

It seems (there are 4 a.m. here and I’m tired and sleepy now, so I may have hallucinations :-) ) that there indeed is a counterexample. Put $f(m)=m^2$ if $m$ is even and $f(m)=(m-1)^2+1$, if $m$ is odd. Then $f(m+1)-f(m)$ equals $1$ for even $m$ but equals $4m-1$ for odd $m$, so the function $f$ is strictly monotone increasing, but $f(m+1)-f(m)$ is not constant. It remains to remark that $f(m)$ is even for each $m$, so $$f(f(m+1))-f(f(m))-f(f(m+1)+1)+f(f(m)+1)=$$ $$f(m+1)^2-f(m)^2-f(m+1)^2-1+f(m)^2+1=0.$$

PS. An even simpler counterexample we obtain if we put $f(m)=2m$ if $m$ is even and $f(m)=2m+1$, if $m$ is odd. Then $f(m+1)-f(m)$ equals $3$ for even $m$ but equals $1$ for odd $m$.

Alex Ravsky
  • 90,434