« first day (3015 days earlier)      last day (2010 days later) » 

12:01 AM
alrightsy, I'll try hbar
 
12:19 AM
17
Q: Is there an integral for $\frac{1}{\zeta(3)} $?

mickThere are many integral representations for $\zeta(3)$ Some lesser known are for instance : $$\int_0^1\frac{x(1-x)}{\sin\pi x}\text{d}x= 7\frac{\zeta(3)}{\pi^3} $$ $$\int_0^1 \frac{\operatorname{li}(x)^3 \space (x-1)}{x^3} \text{d}x = \frac{\zeta(3)}{4} $$ $$\int_0^\pi x(\pi - x) \csc(x) \s...

Any ideas ?
 
1:15 AM
$\displaystyle\frac{sen^2xcos^2x}{96cos^2x+25}=\displaystyle\frac{sen^2(x+k)cos^‌​2(x+k)}{96cos^2(x+k)+25}$ Any idea how to solve this?
I am solving it in $x\in(\pi-2k, \dfrac{\pi}{2})$ with $k=\arccos(\dfrac{\sqrt{5}}{4})$. I noticed that if $\displaystyle\frac{sen^2xcos^2x}{96cos^2x+25} = g(x)$ then I have $g(x) = g(x+k)$
I showed $g(x)$ is monotone on $(\pi-2k, \dfrac{\pi}{2})$ but sadly $g(x+k)$ is not.
That would have let me conclude that, since they are monotone, they can not intersect. Damn.
 
 
2 hours later…
2:54 AM
@TedShifrin I hope so, haha
trying to finish the dissertation this may
so naturally, my life is filled with job applications, with an occasional side of math
 
3:46 AM
Is there a formal notion of the "degree of invertibility" of a differentiable map $f: M \rightarrow N$ between Riemannian manifolds?
 
4:23 AM
@Daminark Also Zorn is German for anger or wrath!
 
4:43 AM
$a-bx-e^{-x} = 0$
I m thinking how to $x$ in terms of a and b?
any help?
 
It seems it doesn't have a closed solution.
Closed form I mean.
 
ohh
hmm
 
If you didn't have the $a$ in there, there'd be a solution in terms of the Lambert W function
 
Taylor series expansion for $e^{-x}?$
 
hm actually $a,b>1$ can vary and a re parameters
 
4:48 AM
but that's not saying much, since the lambert-W function is defined so as to have $W(x)e^{W(x)}=x$
 
hm
 
5:00 AM
@Raptor hm that would work for small x
 
I know, but for larger $x$ it fails miserably
I was thinking along the lines of $a-bx=e^{-x} \implies e^x(a-bx)=1 \implies ae^x-bxe^x=1$. Not sure how much it helps but now the equation has a $xe^x$ term.
 
hmm
 
Sorry about that, I posted on the wrong chat
How can I delete my own messages in chat?
 
5:37 AM
click right side u see u can delete or not, it has some time to do that, if crossed only mods can delete
 
Nope, can't delete it. People need to create a meme server tbh
Like: cdn.discordapp.com/attachments/407559691791761416/… makes perfect mathematical sense
 
 
2 hours later…
7:29 AM
@Raptor you there?
Is $$\int dxdy=\int dx \int dy$$
Can anyone tell me whether this is true or not?
 
It is true for well behaved functions, see Fubini's theorem
 
@AlessandroCodenotti thanks
 
Nevermind, I misread it since I'm mobile, it is actually false
Just try it with $f(x,y)=xy$ on $[0,1]^2$, the integral on the left is perfectly fine, what do you get on the right?
@LoopBack
 
7:48 AM
Hi all! I'd just wanted to know whether any of you would like to support the "Math Challenges" proposal on Area 51, by following and upvoting some example questions - the latter we'd need for the proposed site to progress. If you have any questions regarding this, please ask them in this chatroom. Cheers!
 
8:04 AM
Hi
May i clarify bayes theorem answer here?
I believe answer should be 0.3
 
8:37 AM
never heard of it why don't you get off to a running start and state the theorem for everyone and then we can look at how many cyclists are probably going to be run over due to well I guess in my city the mean free path of a particle in box would be the best explanation
but instead of bicycles n stuff* wouldnt you much rather sink your teeth into the magical world of prime numbers (and totally popular with the community at large, the townsfolk in your locale are guaranteed to immediately elect you as mayor if you announce you are studying primes)
$$\Bigl\lfloor\frac{p_{\pi(n)+1}p_{\pi(n+1)}-1}{p_{\pi(n)+1}}\Bigr\rfloor=n \operatorname{iff} n \in {\{p_k-1:k \in \mathbb N}\}$$
 
 
3 hours later…
11:30 AM
Hi guys
 
 
2 hours later…
1:11 PM
Intense chat
 
1:44 PM
hellloo
hey
semiclassical
Semiclassical
can i ask you somthing
informally, how much set theory should a first year grad student in math know
would it be a bad sign, if they can't list the axioms of ZFC off the top of their head
 
Couldn’t tell you; I’ve never been a first year math grad student
Seems like an unlikely qualification tho
 
@user3865391 Most people can't even after finishing grad school I'd say
 
2:14 PM
@AlessandroCodenotti If you gave me an hour I might be able to come up with the names
 
I mean I want to specialize in set theory and I always look up replacement because it's annoying to write down properly without messing up...
 
@user3865391 if you don't specialize in set theory or a related area, then not much
 
QQ, how much am I supposed to know in advance if I want to do mathematics in undergraduate? Like RMS-AM-GM-HM inequality, multi-variable calculus, topology???, quaternions???, etc. I am not too sure what are the pre-requisites are for universities in first world countries?
 
how old is "QQ"
@MatheinBoulomenos ich bin nun in Freiburg gewesen
 
2:16 PM
back in my day it was Q_Q
 
@LeakyNun I like Freiburg! it's a nice city
how did you like it?
 
@MatheinBoulomenos du kannst mich besuchen :P
 
Are you in Freiburg or have you been there?
 
That will depend wildly on what country / university systme you're in, @Raptor. In the US one can essentially come in only knowing the background for calculus and then learn everything else at school
 
I am now in Freiburg
 
2:18 PM
That must be a really cramped schedule then.
 
Hmmm 300Km from here, a bit too far
 
I dunno.
I took multivariable calculus when I got to college and i did ok.
 
Heidelberg - Freiburg is like 1 hour and 45 minutes by train, I've done that before
I could but I'm quite busy
I actually visited Freiburg like a month ago
 
Worries me to be honest, as I really don't want to enter university having to learn a year's worth of mathematics in one month and I don't really have anyone that I can ask. I find multiple answers online and I'm not sure how accurate they are. The answers have a lot of variation too.
 
@Raptor I can only speak for Germany, here everyone has some probability, some calculus and some linear algebra and analytic geometry in high school, so that's highly recommended, but it's not strictly necessary, the lectures start from zero and are technically self-contained
 
2:23 PM
@Raptor I am not sure why you'd have to learn a year's worth in one month. I still don't know where you are based though.
You should expect variability largely dependent on that.
 
I am based in Bangladesh and it was just an assumption.
 
I still don't understand the assumption. But I know nothing about the Bangladeshi university system.
 
I'm not even sure if the high school math education is the best preparation for university math, you might profit more from starting with proofs, a lot of high school math is just rote memorization or rote calculation
 
Well, one reason why I want to avoid the Bangladeshi Universities is because public universities often teach in Bengali, while I have learnt math in English since the very beginning. And from what I have read, there is a lot of student politics too.
And I am already doing proofs for preparation though I should start doing more of them. I have just started induction.
 
where do you plan/hope to study?
 
2:39 PM
Though I am slightly put off by the quality of the book. I recently did a proof following the book's method and I must say that when I showed it to a friend, she described as really "stiff" and that mathematical proofs should look like "numbers scattered across a book."
 
"like numbers scattered across a book" sounds pretty stiff to me
 
Same here but perhaps everyone reads different books??? Stephen Hawkings is definitely not stiff.
math.stackexchange.com/questions/2985270/… has a lot more text than mine
 
Eh, that feels a bit overwritten
 
2:49 PM
@Raptor you can't compare popular science books to a textbook, textbooks will be more "stiff" since they have different goals and requirements
 
that's true
 
"Suppose $5\mid k^5-k$ for some positive integer $k$. Then
\begin{align}
(k+1)^5-(k+1)
&= [(k+1)^5-k^5-1]+k^5-k \\
&= (5k^2+10k^3+5k^4)+k^5-k\\
&=5k^2(k+1)^2+k^5-k.
\end{align}
Since $5\mid 5k^2(k+1)^2$ and $5\mid k^5-k$, we conclude that $5 \mid (k+1)^5-k^5$. Since $5$ divides $1^5-1=0$, we conclude by induction on $k$ that $5\mid n^5-n$ is true for any positive integer $n$."
 
Neat
 
of course, it'd help if I actually did the algebra right. $(k+1)^5-k^5-1=5k+10k^2+10k^3+5k^4=5k(1+2k+2k^2+k^3)$
The conclusion is unchanged, though
 
Why don't you post it as an answer.
 
2:58 PM
meh
 
Does someone know if centralizers behave well with ring extension, i.e. if $R \hookrightarrow S$ is a ring extension and $A$ an algebra over $R$, is $C_{A \otimes_R S}(a \otimes 1) = C_{A}(a) \otimes_R S$ for every $a \in A$?
 
I have a question that I'm not sure how to phrase
 
we'll help
 
So, actually the idea is quite simple, I have a set of M numbers that sums up to Q
 
Are there constraints on the numbers, e.g. are they positive integers?
 
3:05 PM
They are integers but not necessarily positive
I can choose to order them according to size $q_1 < q_2 < \dots < q_M$
Actually \leq rather than <
So theoretically I could choose all of them to be the same, but anyway
My question is this: I know the total sum is Q
What can now be said about the sum $\sum_{i=1}^M q_i^2$ ? Or the cubes? Or some other power of $q_i$ of my choosing?
I suppose I could say something about the bounds, but "the dream" would be to know the value of the sum itself
What is a good title for this kind of question? Technically it's not a homework question - I'm a physics student doing a thesis on the Bethe Ansatz and this seems the most feasible course of action towards getting the answer
This feels like a number theory thing somehow, but I can imagine that the answers I seek are not so trivial
And seeing as I didn't take any such course I'm out of my depth even in regards to putting the question into words :p
 
Well, note that if you allow any integer value---e.g $q_1=-2,q_2=3$---then there's an infinite number of ways to get a certain value of $Q$
 
I see the problem. So I should restrict the values to be positive integers?
 
If that makes sense for your problem, sure. (Or maybe nonnegative integers.)
If there's some motivation from the physics, presumably that will tell you what values of $q$ are allowed.
 
Hmm, now that I think of it, I think I can limit the values to positive integers. Non-negative wouldn't make much sense in my case
 
Okay. That certainly restricts things, though how much I don't rightly know
As an example, for $Q=5$ and $M=2$ you could have $(q_1,q_2)=(1,4)$ or $(2,3)$ subject to your ordering
 
3:15 PM
Yes, I'm looking at my notes, this would make sense. So I can adjust all values of q to be between 1 and, well theoretically Q-M+1, since I'd still have M-1 values that could be 1
 
in which case you could either have $q_1^2+q_2^2=1^2+4^2=17$ or $q_1^2+q_2^2=4+9=13$
right.
You're essentially interested in partitions of $Q$ into $M$ parts
and what the sum of squares for such a partition would be
 
That phrasing certainly helps - I remember doing an exercise on just that...
 
bounds are not hard to get, you can apply Jensen's inequality with $x \mapsto x^2$ (or $x \mapsto x^n$)
 
Presumably the 'optimal' lower bound will occur when $M$ divides $Q$, in which case the sum of squares is $M(Q/M)^2=Q^2/M$
 
I guess I shouldn't have mentioned the word bound.. although it's a great tool, the closer I would get to the actual value of the squares, cubes, etc, the better
 
3:18 PM
I expect there's not a nice answer for "how many values of $\sum_k q_k^2$ are possible where $\{q_k\}$ is a partition of $Q$ into $M$ parts"
There's presumably an upper bound on that, but I doubt you can rule out two different such partitions giving the same sum
might be wrong on that tho
 
@user55789 what kind of answer are you expecting?
 
@MatheinBoulomenos Actually I'm just trying to figure out a way to phrase the question and it's already getting somewhere
I'm just not so much into the maths lingo hence my question here
 
@Semiclassical a trivial bound is the number of partitions, of course, which has well-known asymptotics
 
right
 
@Semiclassical Maybe some context would help
 
3:22 PM
@user55789 i'm not asking about the terminology, I'm just asking what kind of information about the sum $\sum_i q_i^2$ you are expecting if you are not interested in bounds, since the exact value can't be computed just from knowing $\sum_i q_i$
 
As an example of how things can be annoying: 5,2,2 and 4,4,1 are both partitions of 9 into 3 parts, and the sum of squares of both is 34. So two such partitions can have the same sum-of-squares.
On the other hand, the partition 5,3,1 instead has sum-of-squares 25+9+1=35
 
Okay I suppose I will give some context then
 
this feels like a hard combinatorics problem, if you ask for the number of possible values
 
@MatheinBoulomenos agreed.
Might be on OEIS
The basic upshot is that two different partitions of Q into M parts can have the same sum of squares, but they don't need to have the same sum of squares.
 
The idea is that I have an energy function that is: $E(\{q\}) = \sum_{i=1}^M q_i^2 + 2 \sum_{j=1} \lambda^{(0)}_i d_i(\{q\})$
I'm looking into the idea that I'd like different sets of q's that correspond to $E=0$. The function $d_i$ has a huge amount of crap in it, but the part that depends on $q$ is: $(M/L) \sum_{\ell = 0}^\infty \frac{\phi^{(\ell)}(\lambda_j)}{\ell !} \sum_{m=1}^M (-q_m)^\ell$
The idea was to find different sets of $q$ that would obey $E=0$
 
3:33 PM
What determines the $\lambda_i^{(0)}$?
 
This can be chosen freely and basically depends on the parameters chosen for the initial system, in a sense they can be chosen freely
So my first idea was, perhaps I can relate this to some sort of geometrical figure. Without the $q$ dependence in the function $d$, I suppose one would have a hypersphere. But it's not so clear (to me at least...) how I would represent essentially a summation of q's with coefficients >2 to something geometrical, if that is even possible
 
Rephrasing my question, if $R \hookrightarrow S$ is a ring extension and $A$ a matrix in $M_n(R)$, is the centralizer of $A$ in $M_n(S)$ the tensor product of the centralizer of $A$ in $M_n(R)$ with $S$? I think this is true because the centralizer is defined by linear equations whose solutions can be viewed over $R$ and $S$. Is this argument ok?
 
The idea stemmed from the $E=0$ constraint, i.e. I could just solve one variable in terms of the other, then make a statement about the partitions and thats it
 
@abenthy seems correct to me
@user55789 the important question is: what kind of statement are you looking to make about the partitions?
 
I couldn't finish Proofs and Refutations
I guess you could say I'm Lakatos intolerant
5
 
3:48 PM
How do I find the Laurent series of $e^{\frac{1}{1-z}}$? Wolfram seems to have a very conclusive answer, but I don't know how it got there. I can write $e^z = \sum_{n=0}^{\infty} \frac{z^n}{n!}$ and for $|z| < 1$, $\frac{1}{1-z} = \sum_{n=0}^{\infty} z^n$, but that doesn't quite help me.
 
Chat
Hi
If you take a random set of numbers, copy it, reflect it and place it over top of the other set what is the entropy of the new distribution?
 
@MatheinBoulomenos I guess a first question would be: what partition can I make such that $\sum_{i=1}^M q_i^\ell$ has the same value to all orders $\ell$
 
If $f(z)=\exp((1-z)^{-1})$, then $$f(1/z)=\exp((1-1/z)^{-1})=\exp(z/(z-1))=\exp(1+(1-z)^{-1})=e f(z)$$
So that sets some constraints on what's allowed
 
@MatheinBoulomenos Thanks Mathein
 
@JoeShmo OEIS tabulates the coefficients of $e^{z/(1-z)}=e^{-1} e^{1/(1-z)}$ as A067764
 
4:03 PM
@MatheinBoulomenos do you know things about monoidal categories
 
actually, hmm, that's not quite right.
what that one is tabulating is the numerators.
right, A000262 gives the coefficients $c_n$ in the expansion $e^{z/(1-z)}=\sum_{n=0}c_n(z^n/n!)$
 
$e^{\frac{1}{1-z}}=\sum_{k=0}^\infty \frac{1}{k!}(1-z)^{-k}=\sum_{k=0}^\infty \frac{1}{k!}\sum_{n=0}^\infty \binom{-k}{n}(-1)^nz^n=\sum_{n=0}^\infty((-1)^n \sum_{k=0}^\infty \frac{1}{k!}\binom{-k}{n})z^n$
@MikeMiller only some basics
 
@Semiclassical in fact the question asks about $e^{\frac{z}{1-z}}$, but I thought I simplified the question
 
@MatheinBoulomenos I got a colleague to explain some notational question I was confused on
glad you were here though in case he fell through :)
 
ah. it's actually better to use that version, since it goes to $1$ as $z\to 0$ rather than $e^1$
What is the problem exactly asking, though? Judging from the OEIS page, there's no closed form for $c_n$
(There is a combinatorial interpretation of it, but no simple formula you can plug into)
 
4:11 PM
classifying the nature of the singularities
 
Ah.
You shouldn't need the explicit Taylor series for that
 
@JoeShmo here's a general fact: if $f$ has any non-removable singularity at $z_0$, then $e^{f(z)}$ has an essential singularity at $z_0$
 
hm, but our power term seems to have a removable singularity
 
A removable singularity where?
 
at $z=1$
 
4:14 PM
$z=1$ is not a removable singularity of $\frac{z}{z-1}$
 
you need to expand around the singularity you want to classify
 
$f(z) = \dfrac{z}{1-z}$ has a removable singularity at $z=1$
oh
 
it behaves like 1/0 at z=1
 
the expansions we were doing are around $z=0$
 
i see
 
4:15 PM
note that $\frac{z}{1-z}=\frac{1-(1-z)}{1-z} =\frac{1}{1-z}-1$
 
yes, agreed, thats how i "simplified" my expression earlier
oh yeah
yes yes
 
The Laurent series around $z=1$ is rather easy, just use $e^{\frac{1}{1-z}}=\sum_{k=0}^\infty\frac{(1-z)^{-k}}{k!}$
 
im an idiot. i don't even need to expend
although expending would be a good exercise
 
It's a good reminder that expansions about two different points can have very different coefficients.
With the Laurent series about $z=1$, the coefficients are explicitly calculable
With the Laurent series about $z=0$, there's no closed form (judging from the OEIS page)
 
$lim_{z \to 1} (z-1) -\dfrac{1}{z-1} = -1 \neq 0$
which is a necessary condition
 
4:19 PM
is that in the sense of $(z-1)\cdot -\dfrac{1}{z-1}$?
 
$lim_{z \to 1} (z-1) \cdot (-\dfrac{1}{z-1}) = -1 \neq 0$
 
gotcha
This of course goes to MB's point about $e^{f(z)}$ with $f(z)$ having non-removable singularity at $z=z_0$
 
yes, yes.
what about $\infty$?
doesn't seem to be an essential singularity there in this case since $\dfrac{z}{z-1} \to 1$ as $z \to \infty$
 
Note what I remarked earlier: $f(1/z)=ef(z)$
 
$\frac{3}{\sqrt{1}} + \frac{3}{\sqrt{2}} +..+\frac{3}{\sqrt{n}} \geq \sqrt{9n}$
any idea how to prove
this
for any $n \in N$
 
4:22 PM
Harmonic progression?
 
So $\lim_{z\to \infty}f(z)=\lim_{z\to \infty} f(1/z)/e=f(0)/e=1/e$
 
@Raptor yup but sqrt involved
 
@BAYMAX $\sqrt{n} > \sqrt{k}$ for $k<n$
 
This is interesting
I think you can factor the $3$ out
 
@Semiclassical agreed
 
4:24 PM
yeah, the 3 and the 9 are distractions
 
oh yeah!!!
 
Seems like the simplest move is the integral convergence test
 
and solve for $\zeta(\frac 12)$
 
hm
 
ok no,don't
it was a joke
 
4:26 PM
Come to think of it, I don't know what the large-$n$ asymptotics of $\sum_{k=1}^n k^p$ are when $0<p<1$
 
so sleepy but nxt Now $(1-\frac{1}{2})(1+\frac{1}{3}) .. (1 - \frac{(-1)^n}{n}) = \frac{1}{2}$
rapidsfires
 
um. that's false when n=3
 
this1 is hot sauce with hakka with vinegar
 
do you mean in the limit as n->infty or something?
 
we have to prove tht for each natural number $n$
 
4:28 PM
Odd thing I noticed: $(1+\frac 13)(1-\frac 14)\ldots (1-\frac {(-1)^n}n)=1$
n=1
?
 
Well, that product seemingly starts with $n=3$
but that doesn't help much. when you plug that in, you just get $1+1/3=1$ which is obv. false
 
n=2 starting i think
 
I believe that it should be true as $n$ approaches $\infty$. Did you mix up the questions?
 
oh my
 
Plugging into mathematica, it seems to be true that $(1+1)(1-1/2)(1+1/3)(1-1/4)...(1-(-1)^n/n)\to 1$ as $n\to \infty$
 
4:30 PM
i hae to prove that is true for each even natural number
even natural number n
 
hmm
that does seem to be true
In which case the most obvious approach is induction
 
yay
like spse true for n=2k
then we have to prove it for n = 2k+2
right?
 
Yeah. How does the product change when you increase $n$ by $2$?
 
$\prod_{n=1}^{\infty} \left( 1-\frac {(-1)^n}n\right)=1$ seems easier.
 
@Semiclassical extra two terms get multiplied
 
4:32 PM
yeah, that's the simplest way to write it
Okay. What are those two extra terms?
 
$(1-\frac{(-1)^{n+1}}{n+1})(1 - \frac{(-1)^{n+2}}{n+2})$
 
Right. And $n$ is even, so what can you conclude about $(-1)^n?$
 
1
 
Right. So how does that simplify your expression?
 
yeh the product is1
 
4:34 PM
So $(-1)^{n+1}=-1$ and $(-1)^{n+2}=1$
 
$\frac{n+2}{n+1} \frac{n+1}{n+2}$
 
Yep. So therefore if the product of the first n even terms was 1, then it'll still be one after adding two more
 
=1
 
@Raptor yup
@Semiclassical yep
 
Nicely done!
Next one
 
4:35 PM
nxt fire!!
 
The real point in there was distinguishing the case of n even vs. n odd
 
n odd was why we got stuck
 
one has $\prod_{k=1}^n (1-\frac{(-1)^k}{n})=1+\frac{1}{n}$ when $n$ is odd
 
$n$ is not prime and a perfecct square divides $(n-1)!$
sorry
 
so the product approaches 1 as odd $n\to \infty$
which is cute though not at all essential
 
4:38 PM
yes
thats nice
 
$(1+1)(1-\frac 12)(1+\frac 13)(1-\frac 14)(\ldots=2 * \frac 12 *\frac 43 *\frac 34*\ldots$
 
So you've got a non-prime $n$ such that $(n-1)!$ contains a perfect square.
 
@Raptor seems cute1!
 
yes
wait what?
cute???
 
it's a nice result, it's just not a very deep one
that's all
 
4:40 PM
yes
 
yaa
 
@BAYMAX one thing I notice immediately: Since $n<m\implies n!\mid m!$, any square factor of $n!$ is also a factor of $m!$
 
@Semiclassical I think n is not a prime and not a perfect square divides (n-1)!
 
What?
 
am i saying rubbish
 
4:41 PM
What's the exact statement?
 
You mean when n is composite
 
So far I'm only hearing premises (and inconsistent ones at that), not an actual question
 
Prove that a natural number $n$, that is not prime and a perfect square divides $(n-1)!$
this actual qn
 
So $n$ is composite?
 
yup
 
4:43 PM
And $n\neq k^2$
 
That's clearer, but still messy. State your premises clearly---what must $n$ fulfill?
 
how u got n to be prime?
 
It's a staid format, but phrasing this as "If P, then Q" will clarify matters
What are your premises P, and what conclusion Q are you trying to draw?
 
A number n is not prime and not a perfect square = P
 
Let me try at cleaning up your question:
 
4:47 PM
Q = n divides (n-1)!
 
So: Given a natural number $n$ which is neither prime nor a perfect square, show that $n\mid (n-1)!$.
 
If $n\neq p$ and $n\neq k^2$ then $(n-1)!=xn$
 
Contrapositive: If $n$ doesn't divide $(n-1)!$, then either $n$ is prime or $n$ is a perfect square.
 
yup
 
Can't say I've got intuition about this.
I don't see induction being a good approach, tho
 
4:49 PM
if $n$ doesnot diveide $(n-1)!$
 
Notice this: all factors of $n$ that satisfy both conditions are smaller than $n-1$
 
then $n$ has no factors
from $2$ to $n-1$
implying
 
or an odd number of factors
 
$n$ hs only two factors of itself that is 1 and n
and hence n is prime
 
You definitely need the condition that $n$ not be a perfect square. Otherwise you could take composite $n=4$ and yet have $4$ not divide $3!=6$.
So "$n$ doesn't divide $(n-1)!$" is not enough to conclude that $n$ is prime.
 
4:54 PM
hm
 

« first day (3015 days earlier)      last day (2010 days later) »