« first day (2708 days earlier)      last day (2609 days later) » 
01:00 - 23:0023:00 - 00:00

01:03
What is everyone's favorite problem here ? Looking for some fun problems to do
 
1 hour later…
02:17
@SandeepSilwal Favourite problem, hmmmmmm....
Hi @Narcissus, @Eric, @SemiC
@BalarkaSen Hey there. I don't have the background to understand your answer in SGA still btw.
I haven't done much AT yet.
Ah I can try to elaborate if you want
But maybe not quite right now
I should probably read some more Hatcher first anyhow.
Everyone should read Hatcher!
02:22
But I want to be hip :).
Reading homological algebra from H&S and commutative algebra from Matsumura, as well as looking at Raymond now (by Adeek's suggestion).
Sounds pretty cool
But I can fit some Hatcher in surely
Does one actually do the exercises in Hatcher?
The exercises in Hatcher are a big part of the book
03:20
I haven’t done all of the exercises in Gatcher but I’ve done maybe 2/3
worth it for sure
I think I have done nontrivial amounts of Chapter 2 and Chapter 3
never did anything from 4
03:37
@Balarka!!! You're clearly unsleeping too seriously!
Ahah. Mistaken!
Oh yeah?
I fell asleep at 1 AM and woke up at 7
Oh, well, that's something.
Very quiet here.
How's your day
Yeah
Not much activity
03:39
I'll get onto Gatcher then. Should be increasingly easy as I learn more homological algebra prerequisites.
It's evening now. Tennis on from Australia :)
Why did H turn into G re Hatcher?
Narcissus is joking about Mike's typo I guess
I dunno, Mike possibly made a typo, and I just carried it on :P.
03:40
Heya DogAteMy
aha @Narcissus
This is how inside jokes are born :').
Or you're turning everything into Russian (no H in Russian).
Mike's phone has made many unfortunate typos before
Interesting, I didn't know that
заговорщик (That translates backwards interestingly)
Yes, H in English turns into G in Russian.
03:42
I guess Gromov calls it $g$-principles then
I'm sure Russians can learn the english h :P
Я этово не знаю
Hatcher -> заговорщик -> Conspirator
Hatching schemes I suppose
I guess DogAteMy is only here ethereally
He's always conspiring to make you falsely believe that you're studying algebraic topology
03:44
:')
Not I.
Annoying to answer a neophyte's question, get no response, no upvote, no question, no nothing.
Answering in MSE is mostly a thankless task of late
Maybe I need some new year's resolutions.
Mostly I get a response.
Oh hell, it's Leaky.
03:47
lol
"continuous bijection is either strictly increasing or strictly decreasing"
what needs to be assumed about the topological space for this to hold, if the topology is the order topology?
I mean, it certainly doesn't hold in Q even though it is order topology
Interesting. It doesn't?
does it?
I haven't constructed any counter-example in Q
Then how the hell are you saying "it certainly doesn't hold"?
because my gut tells me so, sorry :P
That doesn't make a "certainly"
03:50
alright, I've constructed it
order-biject Q with Q\{0}, and then do the 1/x trick
can I make it a "certainly" now?
I don't understand what you just said ... But I guess you're saying $f(x) = 1/(x-\sqrt2)$ gives an example.
The usual proof does depend on the intermediate value theorem, so your gut had a rationale
@TedShifrin no it doesn't
it doesn't map to Q
No, it doesn't
So explain your cryptic remark.
the order-bijection between Q and Q\{0} does use sqrt(2)
I don't know what you're talking about
03:53
it maps numbers around sqrt(2) to numbers around 0
I'm describing it
let a_n be a strictly increasing sequence of rational numbers that go to sqrt(2)
and then map [a_1, a_2) to [-1/1, -1/2)
map [a_2, a_3) to [-1/2, -1/3)
(note: [a,b) and [c,d) order-bijects by mapping x to c+x(b-a)/(d-c))
and then do the same thing to b_n a strictly decreasing sequence of rational numbers that go to sqrt(2)
@TedShifrin ok?
And fill in below $a_1$ and above $b_1$ ...
certainly
So we do $1/x$ conjugated by this homeomorphism. OK.
yes, indeed
So, as I said, the usual proof uses the intermediate value theorem. so I guess you probably need a linear continuum.
03:56
usual proof of what?
oh
The usual proof of the "theorem"
is there a topology equivalent of intermediate value theorem?
of course there is
yes ...
it's called "connected"
yes ... but of course you need order topology in domain and range, too.
03:57
right
so linear continuum.
so our order topology needs to be connected
I'm outta here ...
Cya @TedShifrin
04:34
Do you think it's possible to derive a law of "gravitation" for the dynamical graph described in math.stackexchange.com/questions/2583334/…?
I'm having a hard time trying to describe what I want formally, but intuitively I feel like there should be a way to probabilistically predict whether the distance between two nodes increases or decreases over some time scale given some other information e.g., in-degree/out-degree of the two nodes, average degree of the graph, etc.
 
1 hour later…
05:56
....With the recent spread of the topology epidemic, more and more users of this chat were converted to topology. It would not be long before everyone be converted to algebraic topology and differential geometry, and becoming slaves of Balarka's rein...
That would make a good lore
The math chat was once ruled by the founders of the various eras:
Balarka Sen for topology and differential geometry
Waiting for integrals and series
Simpleart for googology
Tobias Kindeltoft for abstract algebra
Akiva for number theory
Alessandro for set theory
and an unknown group for analysis
Tobias Kildetoft for Lie theory and modular representation theory (?)
Probably, I have no data for anything before my joining of the chat in 2013
But at least in the past 2 years, Tobias mainly mediated the abstract algebra theme and questions related to that
(Lore) The chat has witnessed rise and falls of the eras as the founders fade in and out of existence and exert their dominance to this chat
While no eras have stayed long, the era of topology proved resilient and always behind a corner
as the chat being occupied by it every now and then
06:29
@LeakyNun hi , maybe you can help ? math.stackexchange.com/questions/2588440/…
07:08
@Liad why did you delete it?
@Secret What is this fantastic anime? I must watch it
if i plot two histograms that overlap, what are good colors to use to make it readable? (alpha=.5 for each histogram)
currently red and blue, doesn't look great
hmm, green and blue much easier to read
07:32
Good luck finding someone to turn that into an anime lol
07:43
@LeakyNun because i solved it
@Liad good
@LeakyNun maybe you'll be able to help me with a problem in logic theory
$E$ is a ralation of 2 variables, i wrote formulas that states that $E$ is an equivalent ralation (such as $E(x,y) \to E(y,x)$). now i need to define $\phi_n$ a formula that states that there exists and unique an equivalent class of order $n$ @LeakyNun
what have you tried?
i can write the equivalent class of order $n$ formula
$E(x,y) and x!=y$ for example (n=2)
and i added not exists 3 different elements
so it would be exactly 2
functions similar to second degree histogram? imgur.com/a/WrQEf
07:57
@Liad you forgot the unique part
08:16
@Adeek I have started on Raymond's text by your recommendation.
NEW KGLW ALBUM!!!
aaaAAAA
:goes unhinged:
@BalarkaSen If I have $\Bbb R$ as an inner product space over $\Bbb Q$, can we have $x,y \in \Bbb R$ such that $\langle x,y \rangle = e+\pi$?
@LeakyNun RATTLESNAKE RATTLESNAKE RATTLESNAKE RATTLESME
ISOLATION TREPIDATION
08:26
aka "is $e+\pi$ rational?"
let's publish a paper on April 1 proving that $e+\pi$ is irrational
DONT FEAR NOTHING SNAKE IS BLUFFING
@LeakyNun That's easy. We will use recent advances in number theory to prove a generalization, namely e+π(n) is irrational for every positive integer n, not just e+π(1).
Now where's my hat for the occasion...
rip...formatting away my emoji
@user21820 nice
@user21820 right, where's your hat
@LeakyNun There; just found it.
09:01
$1 \to Z(G) \to G \to \operatorname{Aut}(G) \to \operatorname{Out}(G) \to 1$
hmm, this is the first exact sequence I saw that contains 6 elements
whats Out
@BalarkaSen KGLW?
King Gizzard and Lizard Wizard
Oh, cool, I'll check it out
They put out 5 albums last year
the last they dropped in 31st
"Gumboot Soup"
09:05
wait, $e + \pi$ is irrational and we had a paper on that already?
(cannot tell if user21820 has included the April fool reference of Leaky's response)
That should be (very) open
guess so
corrugations are 10/10
I am suspecting we don't know what is the continued fraction representation of the above I think
despite we knew the continued fraction of the individuals
actually, might be related. Must the sum of two normal numbers be normal?
Intuitively it would seemed that having an almost uniform distribution of any finite strings of 1 and 0s among the binary expansion of the two numbers should suggest the carryovers from the sum of them should be unlikely to produce periodic patterns
but again I am not sure. How normal is $e$ relative to $\pi$
@Secret You certainly need to assume some sort of independence of the two numbers to get their sum to be normal
otherwise you could just take a number and its negative.
09:16
oh hey @TobiasKildetoft!
Problem is, we cannot check independence by assuming the transcendentals as linear combination of rationals (as that is the Hamel basis of $\Bbb{R}$ which we don't have an explicit way to construct due to its existence depends on the axiom of choice)
@Secret if $x$ is normal $-x$ is too
Sniped by Tobias
@AlessandroCodenotti That's the (rather) trivial case. what I am wondering is take two normal numbers a and b that are not equal. That seemed to be a harder question. I wonder if something like the standard deviation of the positions of ones of a relative to b will help give some estimate of their independence
because take $\pi$ vs $\pi$. Then clearly the standard deviation of the digits of $\pi$ wrt itself is zero because they are the same pattern
(missing background on how to prove the integer multiple of a normal number is normal)
What if you take a normal number and construct a new one that has as $n$-th decimal $9-$the $n$-th decimal of the original number
hmm... I am not sure. Then clearly by construction, only the nth digit will be different, thus the standard deviation will be small. However the sum and product will be affected rather significantly as suddenly you have the nth digit to go from a low probability to give a carryover (assuming the original digit is x < 5) when summed to having a higher probability to give a carryover (because now 9-x > 4, thus it becomes a lot easier to produce a carryover when the numbers are summed)
09:26
@AlessandroCodenotti i.e. 1-x lol
If x is in (0,1) yes
I have not worked out the algebra of carryovers yet, thus so far the only thing I can say is changing the probability of the nth digit to produce a carryover when summed or multiplied to another number will change the probability of a finite possibly long string of digits to produce carryovers
so it is a highly nonlocal effect in general
Nevertheless, I think your argument does support that the notion of standard deviations mentioned above may not help much on checking the independence of two normal numbers
hmm...
09:58
The variance of a (real) random variable should be $Var(X) = \int |X-EX|² dX > P(|X-EX| > e) • e²$ for $e > 0$. Is there a name for this statement?
 
1 hour later…
11:13
Hi guys. Is it true, that all limit laws hold also in $\mathbb C$?
I was wondering if the scalar multiplication law holds: $\lim_{n \to \infty} ( \alpha \cdot a_n ) = \alpha \lim_{n \to \infty} a_n$ for an $\alpha \in \mathbb C$ ?
The sum and the product law hold if I'm not mistaken?
@philmcole The scalar multiplication follows from the product
@TobiasKildetoft Ok and all laws are exactly the same in $\mathbb C$ ?
@philmcole all of the ones just involving addition and multiplication, yes, since those just boil down to those operations being continuous
@TobiasKildetoft Understood. Thanks a lot!
Hello and a Happy New Year!!

Let $f:[-1,1]\rightarrow \mathbb{R}$ be defined by $f(x)=\begin{cases}\frac{1}{k} & \text{ for } x\in \left (\frac{1}{k+1}, \frac{1}{k}\right ], \ k \in \mathbb{N}\\ 0 & \text{ for } x=x_0=0 \\ \frac{1}{k} & \text{ for } x\in \left [\frac{1}{k}, \frac{1}{k-1}\right ), \ k \in \mathbb{Z}\setminus \mathbb{N}_0, \ \text{ i.e. } (-k)\in \mathbb{N}\end{cases}$

I want to examine the existence of $f'(x_0), \ \ f_+'(x_0), \ \ f_-'(x_0), \ \ \lim_{\substack{x\rightarrow x_0+0, \\ x\in D}}f'(x), \ \ \lim_{\substack{x\rightarrow x_0-0\\ x\in D}}f'(x)$.
11:39
@Astyx Bonne année
An integer $z$ is called even iff it is a multiple of $2$.
What will happen if we plug nonzero number into this?
@famesyasd Not sure what you mean by "plug in" since this is just a definition
e.g. is 3/2 not an even number or is something definition undefined or something I dunno
so I mean this definition makes sense for integers; does it make sense for nonintegers, such as 3/2 for example? like do we say that 3/2 is not an even number or what?
@TobiasKildetoft Can you give me another advide? I want to compute the limit $\lim_{n \to \infty} (\frac{n^5-10}{n^2+1})$ by only using the limit laws. I know that this sequence diverges, but I want show it once rigorously.

I tried dividing out $n^5$ so the numerator and denumerator would be two convergent sequences so I could put the limit inside the fraction. Like this

$\lim_{n \to \infty} (\frac{1-\frac{10}{n^5}}{\frac{1}{n^3}+\frac{1}{n^5}})$

But then the denumerator would converge to zero which is not allowed.
@philmcole it is fine to have a denominator that goes to $0$ as long as the numerator does not
then you just need to keep track of which side it approaches $0$ from
11:55
@TobiasKildetoft I just thought, because one limit law is "for $a_n \neq 0$ and $\lim_{n \to \infty} a_n \neq 0$ we have $\lim_{n \to \infty} \frac{1}{a_n} = \frac{1}{\lim_{n \to \infty} a_n}$", that this is not allowed...
@philmcole But you could also just divide out $n^2$ to get an increasing numerator and a denominator that converges to a constant.
sure, you can't use the law in this way
Happy new year everyone. please may someone help me to understand the triviality of this answer:math.stackexchange.com/q/2588100? thanks
 
3 hours later…
15:23
Hi @Balarka
What's the Lefschetz principle in algebraic geometry exactly?
What's the context? Lefschetz has done a lot of stuff in algebraic geometry and I don't remember which one is the theorem, which one is the principle and which one is the formula.
It should say something along the lines of "A (first-order) sentence (in the language of fields) holds over an algebraically closed field of characteristic zero iff it holds over $\Bbb C$"
Ah yes that thing
I don't know what the precise statement is. To Lefschetz it was more of a philosophy than an actual theorem; good man: he knew that general nonsense is motivated by concrete, pictorial things, like $\Bbb C$
@Alessandro Look up GAGA though. It's related.
15:32
Googling that only brings up Lady Gaga :P
In mathematics, algebraic geometry and analytic geometry are two closely related subjects. While algebraic geometry studies algebraic varieties, analytic geometry deals with complex manifolds and the more general analytic spaces defined locally by the vanishing of analytic functions of several complex variables. The deep relation between these subjects has numerous applications in which algebraic techniques are applied to analytic spaces and analytic techniques to algebraic varieties. == Main statement == Let X be a projective complex algebraic variety. Because X is a complex variety, its set of...
"GAGA mathematics" brings up some unexpected results
@BalarkaSen thanks
Cedric Villani is probably very unhinged
That's the image of a deranged man
15:36
He's not even the weirdest among mathematicians to be fair
Don't do PDE kids
4
15:48
There is some discussion of the Lefschetz princple on MO
@BalarkaSen oh hai Boltzmann (lower left corner)
@BalarkaSen how rude
16:18
Guys what's exactly happening (on the formal level) when I do some illegal stuff like writing $1/0$ in real numbers or composing two functions like $f:R \rightarrow R, \; f(x) = sinx$ and $g:R$ \ ${0} \; \rightarrow R, \; g(x) = 1/x $ getting $gf$ = $1/sinx, \; gf: R \rightarrow R$
@Narcissusjewel that is awesome it is clearing so many gaps for me
hi @BalarkaSen
does $1/0$ mean that the pair $(1,0)$ belong to the function $"/"$ ? so if I would write $1/0;$ then it would be a false statement?
On the formal level that can't happen @famesyasd
16:35
@AlessandroCodenotti what can't happen? the possibilty of writing $1/0$?
yes, $1/\sin x$ is not a function $\Bbb R\to\Bbb R$
16:49
I have found a set of polynomials that have as their roots the Möbius function or any other arithmetic function. Bounds on the sum polynomial roots are governed by the polynomials, was it the second coefficient. But unfortunately the second polynomial coefficient is the partial sums of the Möbius function called the Mertens function. So it does not help.
This is what the 12 times 12 companion matrix for the Möbius-function-as-roots-polynomial looks like:
$\left(
\begin{array}{cccccccccccc}
1-x & 1 & 1 & 1 & 1 & 1 & 1 & 1 & 1 & 1 & 1 & 1 \\
1-x & -x & 1 & 0 & 1 & 0 & 1 & 0 & 1 & 0 & 1 & 0 \\
1-x & 1 & -x & 1 & 1 & 0 & 1 & 1 & 0 & 1 & 1 & 0 \\
1-x & -x & 1 & -x & 1 & 0 & 1 & 0 & 1 & 0 & 1 & 0 \\
1-x & 1 & 1 & 1 & -x & 1 & 1 & 1 & 1 & 0 & 1 & 1 \\
1-x & -x & -x & 0 & 1 & -x & 1 & 0 & 0 & 0 & 1 & 0 \\
1-x & 1 & 1 & 1 & 1 & 1 & -x & 1 & 1 & 1 & 1 & 1 \\
So what I am really saying is that I have found the companion matrix to the polynomial, not the polynomial itself.
17:05
Hey! I need help remembering a funny result I heard: it was about groups, and it help for all n != 6
held*
Outer automorphism group of S_n is trivial for n≠6?
How do I expand $s^{-1}(1+s^{-2})^{-1/2}$ in a binomial series?
Valid for $s>1$
Or how can I rewrite $\binom{-1/2}{k}$?
New question: $\mathcal{L}\{s^{2k}\dfrac{\Gamma(1/2)}{\Gamma(k+1)\Gamma(1/2-k)}\}=?$
17:28
Great @Daminark! Thanks
Could someone give me a sanity check? I've been self-studying differential equations and multivariate calculus in preparation for next semester. Both of the textbooks my school use are on the less-rigorous side. Anyways, so far I've found things pretty darn easy. I'm reading thoroughly, doing the exercises, and theres nothing really telling me I'm going to fast, but darnit it feels wierd
I guess my question is, are either Calc III or DE courses students don't often struggle with on their first go-round?
18:00
user image
2
Arnold's Thonk Map
@BalarkaSen thonquake
19:00
@Lozansky Try a few example $k$, see if there's a pattern
@AkivaWeinberger I got $\sum_{k=0}^{\infty} (-1)^k \dfrac{(2k)!}{2^{2k}(k!)^2}s^{2k}$
Sounds right
@AkivaWeinberger Unfortunately, I need a series valid for $s>1$...
I think it has to be convergent if I am to take the inverse Laplace transform term by term
Oh, I miswrote. It should be $\mathcal{L}^{-1}\{s^{2k} \dfrac{\Gamma(1/2)}{\Gamma(k+1)\Gamma(1/2-k)}\}$
Q. Given
*x_1 + x_2 + x_3 + ... + x_k = N*
and *1 <= x_i <= S* for all *i in [1, k]*,
find the number of different solutions.

A. As far as I know, we can apply multinomial coefficient formula *(N-1)C(k-1)* to get the number of positive solutions to the above equation. But it will also include solutions where *x_i > S*. How do I exclude those?

Can I do the following:

Let *y_i = x_i - S*, then we can write the above equation as:
*y_1 + y_2 + y_3 + ... y_k = N-kS*

The number of positive solutions of the above equation will be *(N-kS-1)C(k-1)* and I'll subtract this from the above formula.
20:04
Hi,
$f$ convex function on $\mathbb{R}^n$, $K$ a compact subset of $\mathbb{R}^n$: Is it true that: $\max \limits_ {x \ in K} f(x) = \max \limits_ {x \in \partial K} f(x)$?
 
1 hour later…
21:26
@LeakyNun @Secret turns out I've had quite a few flaws but ironically correcting them made my proof of that one postulate much easier than it otherwise wouldve been.
It does require a bit of possibly measure theory so if you could check it out later tonight that would be awesome
i wil post a revision before i go to bed tonight
@TedShifrin minor issues here and there but you may find the general concept of this interesting. drive.google.com/file/d/1RU1kpRmfIE82QPJ-2FMqlkvELd6de2fe/…
It is a pdf on an alternative method of how to handle the floor function in differential equations.
Hi guys. Can somebody give me a hint how to show that the sequence $$a_1 = 1, \quad a_{n+1} = \frac{2}{3} \big( a_n + \frac{1}{a_n} \big)$$ is strictly increasing?
I've tried it using induction over $n$ but I'm stuck at the induction step.
@philmcole Hint : Study $x\mapsto {2\over3}(x-{1\over x})$ on an interval (I'll let you guess which)
Hi Semi
21:41
Hi
Still doing cool physics ?
Should that be x+1/x or am I forgetting something?
@Astyx you mean really $-1 \over x$ or $+1 \over x$
Yeah...
hi everyone
21:43
bonjour @Astyx
My bad
Thanks
Not much Physics at the moment
Salut
My way of getting intuition for these kinds of questions is to do a cobweb plot
21:44
Same
There’s an animation there which illustrates it
@Astyx Ok I chose the interval $[1,2]$ and it is an increasing function. I don't get how to show it then for the sequence..
Reminds me of that fractal
Not the interval I’d suggest...
@philmcole use induction to prove that $a_n \ge 1$ for all $n$
and then induction again to prove that $a_{n+1} > a_n$ for all $n$
21:47
@LeakyNun Yes. I'm stuck at the induction step for $a_{n+1} \gt a_n$.
You know $f$ is increasing
for $x \ge 1$ that is
Yes
do you know (at the level of a proof or otherwise) what $a_n\to$ as $n\to \infty$
@Semiclassical yes, $\sqrt{2}$
21:48
Yeah
$[1,2]$ is still a feasible interval
So I’d take the interval to be [1,sqrt(2)]
anyway @philmcole do you understand now?
@Semiclassical Yeah. I just need to show that this limit actually exists. Therefore I want to show that the sequence is strictly increasing and bounded. I did the bounding part already. I'm stuck at the increasing part.
@Semi You can take $[1; +\infty[$ and it would still work
21:50
@LeakyNun I still don't get the induction step for the increasing part
@philmcole do you know that $f$ is increasing?
for $x \ge 1$?
@LeakyNun That's what I need to show, no?
$f$ is increasing rewrites as $a\lt b \implies f(a)\le f(b)$
no
$f$ is $x \mapsto \frac23 \left( x + \frac1x \right)$
you want to show that $a_n$ is increasing
21:51
assume $a_k < a_{k+1}$
Uh. $2\mapsto 5/3<2$
then $f(a_k) < f(a_{k+1})$ since $1 < a_k$ and $1 < a_{k+1}$
Ok. Basically I then need to show that $f$ is increasing instead.
So I’m not seeing how going past sqrt(2) is a good idea, since f is decreasing past there
@Semiclassical but what is wrong with my proof?
no
$f$ is increasing at $[1,\infty)$
it's just that $f < \operatorname{id}$ after $\sqrt2$
21:53
on*
@Astyx what do you mean?
Which proves that $f$ converges
$f$ is increasing on $[1;\infty)$
are you picking on prepositions?
Yes :p
21:55
1. $f$ is increasing in $[1,\infty)$
2. $a_n \ge 1$ for all $n$
3. $a_1 < a_2$
4. $a_k < a_{k+1}$ implies $f(a_k) < f(a_{k+1})$ which is equivalent to $a_{k+1} < a_{k+2}$
5. $a_n$ is strictly increasing (3 4 induction)
Ok, I'll try it. Thanks for your help @Semiclassical, @LeakyNun, @Astyx !
Strictly increasing is needed here (I don't remember wether english-speaking countries make the distinction ?)
indeed, strictly increasing is needed
My intuition was always very visual when it came to these things...but that doesn’t quite mesh with actually proving things
22:02
@LeakyNun Can you explain why at step 4. $f(a_k) \lt f(a_{k+1})$ is equivalent to $a_{k+1} \lt a_{k+2}$?
@philmcole because how is $a$ defined?
$a$? What do you mean? $a_n$ is the sequence
$a_n$
how is $a_{n+1}$ defined?
Ah okay! Got it.
Yeah makes sense :)
@LeakyNun Is it generally easier to prove that a function is increasing instead of a sequence?
@philmcole yes
22:15
@LeakyNun Ok. However I'm still kind of stuck at the same step just with a function instead of a sequence. Here is what I mean: Let $x \lt y$, then $f(x) = \frac{2}{3}(x+\frac{1}{x}) \lt \frac{2}{3}(y+\frac{1}{x}) \lt ? \lt \frac{2}{3}(y+\frac{1}{y}) = f(y)$
you know that won't work because $\frac23 \left( y + \frac1x \right) > \frac23 \left( y + \frac1y \right)$
Ok.
But how else
differentiation :P
Ok. No other way?
Eh, there’s certaibly some way
22:19
We are not allowed to use stuff which we haven't introduced yet..
Not saying one is necessarily obvious. But that function is simple enough that calc shouldn’t be needed
Do you have any suggestions?
I'm thinking
there must be a way to convert a proof with calculus to a proof without
I think so too
but it's one of the more difficult functions with the $\frac{1}{x}$...
$y-x+\frac1y-\frac1x = y-x+\frac{x-y}{xy} = \frac{y-x}{xy}(xy-1)$
eureka
22:22
Well, you want to show 2/3(x+1/x)<2/3(y+1/y) so long as 1<x<y
$x \ge 1$ and $y>1$ so $xy>1$, i.e. $xy-1 > 0$
Top!
Thanks!!
22:43
Is $\Bbb Z_{(p)}/(p)\Bbb Z_{(p)} \cong \Bbb Z/(p)$? Seems like it to me, is there anything obvious to point out to the contrary?
01:00 - 23:0023:00 - 00:00

« first day (2708 days earlier)      last day (2609 days later) »