I need to prove that if $F$ is a closed set such that $A\subset F$ then $\overline A \subset F$, that is, $\overline A$ is the smallest closed set w.r.t. to inclusion that contains $A$ (I suppose?)
Let $F_{\alpha}$ be closed and $A\subset F_{\alpha}$ for each $\alpha$ in an indexing set $I$. We must show $$\overline A = \bigcap_{\alpha\in I} F_{\alpha}$$
The natural definition of that general type is that $\operatorname{cl}A$ is the set of points $x$ such that every nbhd of $x$ meets $A$. The other natural definition is that the theorem that you want to prove.
Let $x\in \overline A$. Then $x$ is either a limit point or an isolated point of $A$. Assume the first, then $x$ is a limit point of each $F_{\alpha }$ so $x\in F_{\alpha}$ for each index since each $F$ is closed. If $x$ is an isolated point of $A$ then $A\subset F_{\alpha}$ so finally $x\in F_{\alpha}$ for each index so $\overline A \subset \bigcup_{\alpha \in I}F_{\alpha}$-
@PeterTamaroff Yes, that’s another; I just don’t normally see any great need to distinguish the points that are already in $A$ from those that are not. Eventually one wants to prove that all of these are equivalent anyway, but I prefer to pick a nice one as my definition.
@PeterTamaroff You said that you’d already shown that if $F$ is a closed set containing $A$, then $\operatorname{cl}A\subseteq F$; from this it’s immediate that $\operatorname{cl}A\subseteq\bigcap\{F\supseteq A:F\text{ is closed}\}$. (Note the typo in my earlier comment: that should have been $F\supseteq A$.)
To get the opposite inclusion, merely note that $\operatorname{cl}A$ is one of the closed sets containing $A$, so $\bigcap\{F\supseteq A:F\text{ is closed}\}\supseteq\operatorname{cl}A$.
@ChuckFernández Perhaps the best introductory graduate level text in general topology is Willard’s General Topology, which I believe is available in an inexpensive Dover edition.
@ChuckFernández I’ve mostly forgotten that one, since I never liked it, but I rather think that Willard is a little more comprehensive and a little more advanced; it’s certainly more modern in notation, terminology, and overall coverage.
@BrianM.Scott I forgot to tell you I don't have the assumption that $\operatorname{cl}A$ is closed. This exercise is aimed to prove it is closed by showing it is the intersection of closed sets.
@ChuckFernández I’ve a few objections to it, but overall it’s very good. This isn’t just my opinion, either: the book is generally considered to be the standard against which other undergraduate topology texts should be compared.
@PeterTamaroff Ah, okay. That makes things a little harder. The contrapositive is easiet. Suppose that $x\notin\operatorname{cl}A$. Then you should be able to show without much trouble that there is an open set $U$ such that $x\in U$ and $U\cap A=\varnothing$. Now let $F=X\setminus U$: $F$ is a closed set containing $A$, and $x\notin F$.
@BrianM.Scott Let me read that. THis is something we were discussing earlier: If $x=\lim \, x_n$ with each $x_n \in A$, then $x$ is either a limit point or an isolated point of $A$.
@PeterTamaroff Don’t think about sequences: you’re trying to prove something that’s true in all spaces, even those in which sequences don’t determine the topology.
@BrianM.Scott But that just arose in the discussion of $x\in \overline A \iff x=\lim\;x_n $ with $_n \in A$. That is, $x$ is in the closure of $A$ if and only if $x$ is the limit of a sequence of points in $A$.
@BrianM.Scott Let $(X,d)$ be a metric space. $x \in X$ is a limit point of a set $A\subset X$ if every neighborhood of $x$ contains a point of $A$ other than $x$.
@PeterTamaroff You can’t construct it explicitly, but it’s easy to show that it exists. That’s practically the definition of $x$ not being in the closure of $A$.
@BrianM.Scott If $x$ is not in the closure of $A$ then $x$ is not a limit point of $A$ neither it is an isolated point of $A$. Doesn't that mean $x\notin A$?
$x$ is isolated if there is a nbhd $N$ of $x$ such that $N\cap A=\{x\}$
@PeterTamaroff The fact that $x$ is not a limit point of $A$ means that $x$ has a nbhd $N$ such that $N\cap A\subseteq\{x\}$. The fact that $x$ is not an isolated point of $A$ then implies that $N\cap A$ must be empty, since it can’t be $\{x\}$. So $N$ is the $U$ that I wanted.
@PeterTamaroff $A\subseteq B$ simply means that every element of $A$ (if any) is an element of $B$. $A\subsetneqq B$ means that every element of $A$ is an element of $B$, and there is at least one element of $B$ that isn’t in $A$.
@BrianM.Scott Well, if the student doesn't reach back then that's another issue. But well, I think you're the experienced teacher. I shouldn't be telling you what is or isn't true about teaching. =X
@BrianM.Scott How does it follow that $X-U$ is closed? Maybe some argument about $C(X-U)$ being open?
@PeterTamaroff I always found it distressing when a student just didn’t seem to grasp something no matter what I tried, and I think that many $-$ perhaps even most $-$ teachers feel the same way, but at some point most of us realize that it’s inevitable, and all we can hope to do is minimize it.
@PeterTamaroff The complement of an open set is closed.
@BrianM.Scott OK. Brian, I proved that the graph of a continuous function $f:X\to Y$ where $X$ and $Y$ and metric spaces is closed by a sequence argument.
I ask because of " **Don’t** think about sequences: you’re trying to prove something that’s true in all spaces, even those in which sequences don’t determine the topology."
@PeterTamaroff But you know that in a metric space distinct points have disjoint nbhds; it’s possible to use no more than that fact about metric spaces in your proof $-$ no sequences at all $-$ and then the proof will carry over pretty much verbatim to the general setting.
@PeterTamaroff Probably not, though there’s one that you might find interesting after you’ve had a bit more general topology. It’s A ‘More Topological’ Proof of the Tietze Extension Theorem, The American Mathematical Monthly, 85 (1978), 192-3.
@ChuckFernández I’ve made it a point not to look at that site. I know what my students thought of me over the years, both good and bad. I expected quite a bit; some students liked this, or at least liked the fact that they learned a lot, and many (predictably!) disliked it.
Area is a quantity that expresses the extent of a two-dimensional surface or shape, or planar lamina, in the plane. Area can be understood as the amount of material with a given thickness that would be necessary to fashion a model of the shape, or the amount of paint necessary to cover the surface with a single coat. It is the two-dimensional analog of the length of a curve (a one-dimensional concept) or the volume of a solid (a three-dimensional concept).
The area of a shape can be measured by comparing the shape to squares of a fixed size. A shape with an area of three square metres...
@PeterTamaroff Most instructors establish set percentage cutoffs for the various letter grades (A, B, C, D, F) at the beginning of the course. If the cutoff for an A is $92$%, a student who averages at least $92$% on graded work will receive an A. This requires the instructor to design exams to match his grading scale. I never cared to do this. Like my father, I preferred to design exams to test what I wanted to test and then interpret the results.
I’d guess that over my entire teaching career something around $80$% was a typical cutoff for an A, and a $50$% could be a C. Students didn’t like this, partly because they weren’t accustomed to it, and partly because they couldn’t tell exactly how well they were doing.
Many teachers will tell you that preset cutoffs are more objective, but they’re fooling themselves: they just transfer the subjectivity from the grading to the composing of exams.
@PeterTamaroff Academic performance. The most common system is the four-point system: each A is worth $4$, each B is worth $3$, each C is worth $2$, each D is worth $1$, and an F (failure) is worth nothing. Add up the values of your grades in the courses that you’ve taken and divide by the number of courses, and that’s your GPA.
Different courses may have different weights: one course might be a three-credit course, while another was a five-credit course. In practice, then, one takes a weighted average of the grades.
@BrianM.Scott I don't have a lot of data, but it seems like kids these days are taking just the one that will make them look better. When I was going through this process (Six years ago, I guess? Yikes.) I got the impression that you took both no matter what.
@PeterTamaroff It’s easier to show that if $x\notin\Gamma_f$, then $x$ has an open nbhd disjoint from $\Gamma_f$; from that it follows immediately that $\Gamma_f$ is closed.
@ChuckFernández I get tired of writing out neighborhood all the time; both nbhd and nhd are common abbreviations.
@BrianM.Scott Then I should be saying "Thus, no point not in $\Gamma_f$ is a limit point of $\Gamma_f$ so that $\Gamma_f$ must contain all it's limit points, from where $\Gamma_f$ is closed.
@PeterTamaroff What of it? It would fail to be closed iff it had a limit point that wasn’t in it. If it has no limit points at all, then it certainly has none that aren’t in it!
@PeterTamaroff For some people. It’s exceedingly compactly organized, so that sometimes you don’t see the point of a result until many pages later, but anyone who knows everything in baby Rudin has a superb grounding.
@PeterTamaroff Sorry: I had to step away for a moment and didn’t finish. It is, however, a good example of Rudin’s concision: everything that you need is there, but you have to think about it in order to use it effectively.
(By the way, that’s 2.20 in my edition of the book!)
@PeterTamaroff That works for metric spaces. In a more general setting one starts with open sets, and an open nbhd of $x$ is simply an open set containing $x$. Then a set $N$ is a nbhd of $x$ (not necessarily open) if there is an open set $U$ such that $x\in\U\subseteq N$.
@PeterTamaroff Then you’ll definitely want to learn that material at some point. It’s useful in some parts of general topology and absolutely essential for algebraic topology. But you’ll want to get a good grounding in basic general topology first.
Let $(X_i,d_i)$ be metric spaces for $i=1,\dots,n$ and convert $X=\prod_{i=1}^n X_i$ into a metric space $ (X,d)$ by setting $d(x,y)=\max\limits_{1\leq i\leq n}\{d_i(x_i,y_i)\}$. Let $O_i\subset X_i$ be open for each $i$. Prove that $O=\prod_{i=1}^n O_i$ is an open subset of $X$ and that each open subset of $X$ is a union of sets of this form.
@PeterTamaroff In fact that’s the standard base for the product topology, and what you’re proving is that the metric that you’ve defined generates the product topology.
OK. This is my train of thought so far: $(a)$ Every open ball in $X$ is the product of open balls of the $X_i$. $(b)$ Every open subset $O_i$ is the union of open balls in $X_i$. $(c)$ $O$ is the product of the union of open balls in each $X_i$, so $O$ is the union of the product of open balls in each $X_i$, which are open balls in $X$. $(d)$ Thus, $O$ is the union of open balls of $X$, so it is an open subset of $X$. $(e)$ Freaking put all of the above in a proof.
@BrianM.Scott Maybe that'll clear out when I study Topological Spaces.
You might find it useful to ask yourself exactly what $B_d(x,\epsilon)$ looks like for $x\in X$: $y\in B_d(x,\epsilon)$ iff $d(x,y)<\epsilon$ iff $\max_i d_i(x_i,y_i)<\epsilon$ iff $d_i(x_i,y_i)<\epsilon$ for all $i$ iff $y\in\prod_i B_{d_i}(x_i,\epsilon)$. Thus, $B_d(x,\epsilon)=\prod_i B_{d_i}(x_i,\epsilon)$.
Once you have $(a)$, everything else follows easily. For instance, if $x\in O$, then $x_i\in O_i$ for each $i$, so for each $i$ there is an $\epsilon_i>0$ such that $B_{d_i}(x_i,\epsilon_i)\subseteq O_i$. Now let $\epsilon=\min_i \epsilon_i$. Then $B_d(x,\epsilon)\subseteq O$. Since $x$ was an arbitrary point of $O$, $O$ is open.
This is from John Coates' Example Sheet: If $n$ is not a perfect square and satisfies, \begin{align} n-n^{2/3} < \varphi(n) < n-1 \tag{1} \end{align} show that $n$ is a product of two primes.
@robjohn I don't know where to start. For one, we know $n$ is not a prime. But, that's just an observation, a priori irrelevant.
Oh: Since $n$ is not a prime, it has atleast two prime factors. Since $n$ is not a perfect square, enough to conclude that $n$ has atmost two factors. Is this fine @rob?
Yes. Agreed. The first expression comes from the multiplicativeness of $\varphi$. (of course, it requires us to know what $\varphi (p^k)$ is for a prime $p$.)