0

In the text of 'Introduction to Mathematical structures and Proofs', by Larry Garstein; there are below sub-problems as part of Q. 14 in Ex.#3.3, as shown here. Have taken a few questions below:

(d) Show that if $f$ and $g$ are surjective, then $f\cup g$ is surjective.
(e) Show that if $f\cup g$ is injective, then $f$ and $g$ are injective.
(h) Give an example of functions where $f$ and $g$ are injective, but $f\cup g$ is not injective.

I present here my attempt using three approaches:
(i) logic based approach that uses set-representation for domain & range,
(ii) calculus based approach by taking example of functions representing $f,g, f\cup g$.

(d) Show that if $f$ and $g$ are surjective, then $f\cup g$ is surjective.

(i) Both $f,g$ have their co-domain = range; that means that if consider the union of set of members in range (whether finite, or infinite) will have at least one member of the combined domain (set formed by union of domain sets of both $f,g$) that is pre-image of that.
Also, being union of sets, the common elements of the two sets (combined set of domain, or combined set of range) are repeated only once.

(ii) Let there be a polynomial $p_f$ that represents function $f$, & similarly $p_g$. Each polynomial denotes a mapping. Being surjective for any polynomial function means that two criteria are satisfied, as below :
(a) The function is unbounded from above & below, i.e.
$$ \lim_{x\to +\infty} f(x) = +\infty, \text{ and } \lim_{x\to -\infty} f(x) = -\infty $$ (b) Then, it is an added requirement that the polynomial is an odd order one. For an even order polynomial, for any $y\in \mathbb{R}$, there is $M > 0$ such that $$ x > M \Rightarrow f(x) > y, \text{ and } x < -M \Rightarrow f(x) < y $$ Hence, by intermediate value theorem, there is $x_0 \in [-M,M]$ such that $f(x_0) = y$.

So, the union of two surjective functions too is.


(e) Show that if $f\cup g$ is injective, then $f$ and $g$ are injective.

(i) If $f\cup g$ is injective, then there is an underlying assumption: having common elements eliminated from the set of domain, & range of union of the functions $f,g$. Also, each element in the domain of set $f\cup g$ maps to a unique element in the range.

On separating out the respective members of domains of $f,g$ with duplication of common elements, get the two functions still injective.

(ii) For being an injective function, the polynomial representing the function should be monotonic. This can be checked by verifying that the first derivative is not infinity anywhere.
Stuck on this part.


(h) Give an example of functions where $f$ and $g$ are injective, but $f\cup g$ is not injective.


Need help as am unable to think such example. Particularly, am not clear how it is possible theoretically.

jiten
  • 4,524
  • 1
    What is $f\cup g$? – mfl Sep 26 '18 at 16:30
  • @mfl I have given link of the book's page having the given questions. It is the union of two functions $f,g$. – jiten Sep 26 '18 at 16:32
  • I have seen the link. What is the union of two functions? – mfl Sep 26 '18 at 16:35
  • @mfl union of the mappings given their respective domains & ranges. – jiten Sep 26 '18 at 16:36
  • 1
    We have $f\cup g:A\cup B\to C\cup D.$ How is it defined $(f\cup g)(a)$ for any $a\in A?$ – mfl Sep 26 '18 at 16:39
  • @mfl Given domain of $f = A, g = B$ & range of $f = C, g =D$; need to find which domain contains $a$ and then apply the mapping. If it is common to both domains, then it is possible that the union is not a function anymore, as two different mappings are possible. Your question is simpler, as $a$ belongs to only one domain $A$, hence maps to a point in $C$. – jiten Sep 26 '18 at 16:42

1 Answers1

3

For part $d$ and $e$, the task is to prove a general statement and we shouldn't assume structures such as we can we are dealing with real numbers, continuous functions, or polynomials.

$f:A \to B$, $g: C \to D$.

From the assumption, $f \cup g: A \cup C \to B \cup D$.

  • For part $d$:

If $f$ and $g$ are surjections. Take an element from $y \in B \cup D$. If $y \in B$, we can find a preimage in $A$, hence we can find a preimage in $A \cup C$. If $y \in D$, then we can find a preimage in $C$ and we can find a preimage in $A \cup C$.

Hence $f \cup g$ is a surjection.

  • For part $e$:

If $f \cup g$ is an injection, let's verify that $f$ is an injection. Suppose it is not, then we can find $x,y \in A$, $x \ne y$ such that $f(x)=f(y)$. That is we have found $x,y \in A \cup C$, $x \ne y$ such that $f(x)=f(y)$, violating the injectivity of $f \cup g$.

For part $h$:

Just let $B=D=\{1\}$, $A=\{1\}$, $C=\{2\}$.

Siong Thye Goh
  • 149,520
  • 20
  • 88
  • 149
  • I request the calculus based approach for the part e, as shown in (ii) section under that. I am unable to continue it. – jiten Sep 26 '18 at 16:50
  • 1
    In that case, you are no longer answering the original question. what are your domains of $f$ and $g$? – Siong Thye Goh Sep 26 '18 at 16:54
  • It is the default one, reals. Same for codomains. – jiten Sep 26 '18 at 16:55
  • 1
    we are having the assumption that $f \cup g$ is a function. Since $A=C$ and $B=D$, we have $f=g$. $f \cup g = f$, so injectivity follows from injectivity of $f$. – Siong Thye Goh Sep 26 '18 at 17:00
  • Thanks, please solve my earlier comment's query. Also, our assumptions of common domains & codomains should not naturally lead to $f=g$. The functions are not the same as set of values possible. A function $f$ can be $x^3$, while $g = x+1$; with same domain & codomain. – jiten Sep 26 '18 at 17:03
  • 1
    To have $f \cup g$ remains a function, we can't afford to have $A=C$, $B=D$ and yet $f \ne g$. For the example that you claim $(0,0) \in f$, $(0,1) \in g$, $f \cup g$ is not a function. If you further impose the $f$ is continuous and injective, then we have $f$ is monotonic, hence $f=f \cup g$ is monotonic, hence it is monotonic and injective.... (monotonic is so restrictive, why are we even going there is puzzling to me) – Siong Thye Goh Sep 26 '18 at 17:20
  • Being monotonic is a further restriction over being injective, as needs continuity. So, how can I build the calculus based approach to part (e), if such approach is possible at all. – jiten Sep 26 '18 at 17:26
  • beyond my ability to help you. – Siong Thye Goh Sep 26 '18 at 17:28
  • I hope that you mean that a calculus based approach is not possible for part (e), as is possible for part (d). But my answer in part (d) (ii) too is just a statement of necessary conditions. – jiten Sep 26 '18 at 17:34
  • Please help with my post at : https://math.stackexchange.com/q/2945020/424260. It is regarding finding ways to get $\ge 4$ contiguous $1$s in a block of $8$ bits. Particularly, am stuck on how over-counting occurs as stated in my second comment for answer by @N.F.Taussig. – jiten Oct 07 '18 at 03:29
  • Kindly join the chatroom at: https://chat.stackexchange.com/rooms/89042/notation-for-statement-with-nested-iff – jiten Mar 23 '19 at 03:13
  • Kindly join the chatroom for help on finding equivalent problem to 'rod-cutting' problem at: https://cs.stackexchange.com/a/29190/78009 – jiten Mar 31 '19 at 04:36
  • I had a very humble request but that was also not granted. Your presence caused a boost that's all was needed to explore fast. Although in this topic, help is rare from you; but still your presence bolsters me. Anyway, will post when make a progress. – jiten Mar 31 '19 at 05:19
  • Please take a look at my question posted at : https://cs.stackexchange.com/q/106286/78009 – jiten Mar 31 '19 at 12:22
  • Kindly join the chatroom: https://chat.stackexchange.com/rooms/89042/notation-for-statement-with-nested-iff, as my earlier problem's algo is given in multiple ways elsewhere. And to understand that need understand sequences, as infinite sum of 1/1-x in double summation & further higher forms (based on product of such terms). I want to understand sequences first, & best source touched is that of Bromwich. So, am reading it page by page. Have a simple problem in chatroom for pg 4 of the book. – jiten Apr 09 '19 at 07:50
  • Have posted the question at : https://math.stackexchange.com/q/3180830/424260. kindly detail your answer for the limit to be 0. I have got no code nor any details on squeeze theorem's application; hence leading me into a maze to search & that too without code. Please help. – jiten Apr 09 '19 at 10:28
  • refer to Peter Foreman's answer. he has proven it. – Siong Thye Goh Apr 09 '19 at 10:30
  • he has lead me into a maze as cannot see how an oscillating sequence can be taken at is mean value; so he should have given details of squeeze theorem. For me, it is a struggle to grasp. No code, in fact I feel no code applies, but machine (wolfram) will give answer, then code must be there. – jiten Apr 09 '19 at 10:36
  • Please respond. Feel very difficult in coding as limits are given. Should there be code to solve limits, or apply squeeze theorem (squeezing lemma); but if yes then again, how to do it. Please guide. – jiten Apr 09 '19 at 13:28