« first day (2541 days earlier)      last day (2774 days later) » 

00:12
So $f(0)f(x)=f(x)f(2x)$, for example? @Julius
Because they're both $g(|x|)$
Similarly, $f(0)f(x)=f(2x)f(3x)$
Oh, wait I see it
@Julius It's impossible. I will show that $f(x)=f(0)$ for all $x$, meaning that $f$ is a constant function.
Looks like it must be constant
right.. unfortunate
thanks to both of you
Proof: By the above observation, $f(0)f(x/2)=f(x/2)f(x)$. Dividing by $f(x/2)$ gives the result.
Unless…
Does something weird happen if $f(x/2)=0$?
@Julius Wait, there's a simpler proof
Can $x=y$?
That's fine for odd $x$, but in my case I'd prefer not so many zeros
Yes
'Cause if so, $f(x)f(x)=g(0)$ for all $x$
which means $f(x)$ is constant and always equals $\sqrt{g(0)}$
That's a simpler proof
I see. I was thinking about such a function as to find an example for four random variables $X_1,Y_1,X_2,Y_2$ such that $Cov(X_1,X_2|Y_1,Y_2)=f(|Y_1-Y_2|)$ where $Y_1$ and $Y_2$ take values 0,1,... and $X_1$, $X_2$ can be anything. Not so simple
(for some function f)
00:22
I don't know what $\rm Cov$ is
covariance. ah and $X_1 \neq X_2$
Greetings!
If $x\ne y$ above, by the way, we get a counterexample with $f(x)=\begin{cases}1,&x=0\\0,&x\ne0\end{cases}$
Please enlighten me: Find the smallest possible value of $x^2+4xy+5y^2-4x-6y+7$ I found out that it is $2$, at $x=4$ and $y=-1$ (don't ask how) but the proof was very frustrating, I tried every factorization I could think of..
I have no idea how I'd approach that
Hm. What happens if I plug in $x=y=10$
I get $907$
$907$ is prime
00:28
Now we're getting somewhere
Wait, are you expected to use calculus? @Mahmoud
Not at all
No*
This problem is in the Prologue chapter of a Calculus textbook ..
Hm. What's $(x+2y-2)^2$?
I'm aiming for the $x^2$, $~4xy$, and $-4x$ terms
It's $x^2+4xy+4y^2-4x-8y+4$
Your think minus that thing is…
Ha! You basically solved it
I'm impressed
$y^2+2y+3$
Right?
And that's $(y+1)^2+2$
So your think is $(x+2y-2)^2+(y+1)^2+2$
I don't know if I made a mistake somewhere
*thing, not think
@Mahmoud
00:34
And since the squared terms are $\ge 0$
So this would reach a minimum if I could make those squared terms $0$
Right
Typo in the first term, fixed :P
So if that's possible, clearly $y=-1$, right?
And then $x+2(-1)-2=0$ means $x-4=0$ means $x=4$
And we are done
Yay!
Yes, you did it sir, I spent the whole day trying to do it
Yeah so my thinking was
We can think of this as like a polynomial in $x$, whose coefficients just have $y$ in them
so it's $x^2+(4y-4)x+(5y^2-6y+7)$
and if we complete the square,
that ends up being $(x+\frac{4y-4}2)^2+\rm something$
where that "something" would have no $x$s in it
and that ended up working.
00:39
I couldn't stop thinking of $y$ as variable :P
Thank you @AkivaWeinberger $:)$
You're welcome!
Lesson of the day: If a certain approach you chose to solve a problem failed you, then you have the obligation to quit it and start all over
Or something similar
:P
"The definition of insanity is doing the same thing over and over again and expecting different results" @Mahmoud
2
"The difference between genius and insanity is mesured only by success" And that of course is completely off-topic, but I remembered it now and couldn't help myself.
2
 
3 hours later…
03:31
Hey guys, can anyone familiar with set notation and basic linear algebra (I'm sure all of you are lol) help me out? I'm revisiting linear algebra with a slightly more formal approach and I want to make sure my notes are not pure jiberish.
I just need to make a statement is well formed and it's saying what I want it to say.
Say you have $AX=B$ with solution set $T$ and the associated homogeneous system $AX=0$ with solution set $S$. Assuming $T \neq \empty$, then $(\forall w \in T)(\forall v \in T)(\exists u \in S)(w = v + u)$
The statement I want to check is the last one.
Basically I'm trying to say that once you have a solution to $AX = B$, you can tack on any vector from the associated homogeneous system and still get a solution. Actually, this might be a better formulation: $(\forall w \in T)(\forall u \in S)((v = w+u) \rightarrow v \in T)$
Any thoughts?
03:55
Hello, off topic. What is the purpose of the DeMoivre/Laplace Central Limit Theorem? That is, why would one ever want to use the normal distribution to approximate the binomial distribution in practice?
@TheSubstitute Interesting question.. I can't help you, but that seems like a good question for the main site; or maybe stats.stackexchange?
user84215
We use normal distribution for continuous variables since in description of nature we deal with these variables. Binomial distribution is related to discrete variables.
04:21
@DavidVarela I'd probably just have $(\forall w\in T)(\forall u\in S)(w+u\in T)$
or $w\pm u$ instead if we care
@AkivaWeinberger Thanks, that's much cleaner. It's reassuring to know I'm on the right track.
You could even say something like $T+S=T$ and $T-T=S$
where $A+B:=\{a+b\mid a\in A,b\in B\}$ means the set of things that can be written as the sum of something in $A$ and something in $B$ (and similarly for subtraction)
@DavidVarela
Because binomial coefficients are harder to compute than exponentials.
Hey everyone!
user84215
04:37
Hello.
@AkivaWeinberger Awesome! I think I saw the $+$ notation in Axler's book. Although now it seems kind of obvious (doesn't it always?), I hadn't even considered $T - T = S$! It led me to some new insights about solution spaces :D
I meant solution sets, not solution spaces.
user84215
05:15
NOTICE NOTICE
user84215
I have a new idea although it will likely be blamed and scorned by you as before. I have created a room called "LTD" (Learning Through Discussion) [LTD: Topology](https://chat.stackexchange.com/rooms/62439/ltd-topology). This is basically different from ordinary classes or study groups. In this room, there is a leader who presents study materials (in this case: Topology) and all in the room discuss with each other about them.
For example, suppose a theorem is stated then they can speak about the main ideas in the theorem and its proof (not mentioning details as it does in ordinary classes),
4
user84215
05:52
If you agree to participate in this room, please post your comments in order to determine a time to start.
06:32
@TheSubstitute throw a fair coin $1000$ times, what is the probability of getting $435$ or more heads? Using a binomial distribution gives you the correct result, 3 years from now when you're done with the calculations, the normal approximation gives a very good result and it's much much easier to compute
(The CLT is useful in many cases since it's not limited t o binomial distributions, a lot of stuff can be approximated as a normal distribution)
07:03
Why was I pinged ? @aminliverpool I can't find the message
The message was [DATA EXPUNGED]
Hi Dami
How's it going?
How's life going ?
Okay, I'm out
I didn't come here to get sniped
Nooooo
07:09
Anyway, I'm good (on vacation in Slovenia) what about you ?
I'm doing pretty well, thanks! Just finished dynamics homework
physics ?
The IMO is in Rio this year...
IMO ?
07:24
International Math Olympiad.
in The h Bar, 59 mins ago, by Sid
On a more "nerdy" note, the IMO 2017 has started in Rio!
Oh right
Cool
No, dynamical systems in math
Oh right that makes more sense
what are u learning in Dynamical systems @Daminark
like specific topics :)
07:44
@Astyx I visited ljubljana and the postojna caves around Easter and enjoyed my travel a lot, where are you in Slovenia?
In a small village on the frontier with Italy
@AlessandroCodenotti >< doh, thanks
@Astyx the message was basically pinging everyone
Hey @Baymax, we've been doing some symbolic dynamics, and now we're gonna start ergodic theory
(before that, more general stuff on topological dynamics, such as mixing, entropy, and transitivity)
08:00
hey everyone
lowly physics major here, are you guys going to kick me out?
Everyone! ATTACK!
:P
nah, there are a few physics people here
shh shhhhhhh @Alessandro
Nah but really though it's chill
08:03
whew, ok
whatcha guys talkin about
If ~5 messages can qualify as "talking about", dynamical systems
In general this chat tends to talk about algebra mostly. Algebraic topology is allowed as long as it is sufficiently categorical, right @Balarka?
How is it going with the alg top lectures btw? @Dami
I unfortunately didn't go to the last one, but it was apparently incomprehensible anyway. I've been slightly iffy about them since bootcamp is heavy and I'm tired :P
At some point I'll just work through concise to supplement
The extent of my knowledge about dynamical systems is knowing what Hamilton's equations are, the extent of my knowledge about topology is basic calculus and set theory, and the extent of my knowledge about algebra is group theory :) Physics perpective right here
08:18
Hey
I'm a little embarassed because this should be obvious but I can't for the life of me figure it out
I have an average of averages but for some reason depending on in what order I calculate it the result varies slightly
This happens even with fractions which should be 100% accurate
@Metaphysic99 you'll see soon that even in the context of physics, knowledge of all the stuff will grow deeper! :D
It's darts rounds. You need 501 points to finish a round. Each "throw" has 3 darts. I am to calculate the average points per throw across 10 rounds
I have the amount of darts needed to complete each round
So this is my dataset (easy, I know):
Round no. Points needed Darts needed
1 501 31
2 501 33
3 501 31
4 501 33
5 501 31
6 501 33
7 501 31
8 501 33
9 501 31
10 501 33
So I devised this equation to come up with the average points per throw for a single round: PointsNeeded / ( DartsToComplete / DartsPerThrow )
Or in other words, 501 * 3 / DartsToComplete
Daminark thanks for the encouragement I guess. The dream is to go into theoretical physics but sometimes I wonder whether I should have just majored in math because it's so important and used everywhere all the time
Physics is just like a bunch of special cases in math really
Should be fairly easy I thought. I thought I'd get averages for every round and then take the average of that
I wouldn't say either one subsumes the other, really
Physics requires the additional input of empirical data, so I'd describe it as more, constrained overlaying mathematical structures on stuff
08:30
You're going to do a lot of math anyway if you go for theoretical physics
Then I cross checked by trying ( PointsNeededPerGame * NumberOfGames * DartsPerThrow ) / ( SUM(DartsToComplete) from All games )But the result is ever so slightly off
Both equations formulated out are: ( ( 501 * 3 ) * 10 ) / ( ( 31 * 5 ) + ( 33 * 5 ) ) and ( ( ( ( 501 / 31 ) * 3 ) * 5 ) + ( ( ( 501 / 33 ) * 3 ) * 5 ) ) / 10 WolframAlpha tells me they aren't equal, where in simplifying have I made the mistake here?
This looks like first grade stuff but I have no clue why it's not equal even though one is the sum of all points needed divided by the total darts divided by darts per throw and the other is points needed divided by the darts needed per game divided by darts per throw and then taken as average
Shouldn't this be equivalent?
@Alessandro yeah true. Theoretical physics or any "real" physics is going to involve the creative (and constrained) use of mathematics. I'm still an undergrad so I suppose I'm seeing it from an undergrad perspective, they only teach us very specific math methods like how to solve this pde, this Green function, basically just a bag of tricks
user84215
09:32
Why do you not send your opinions about my idea?
@MetaPhysic99 welcome to the dark side
10:21
Hey guys how should I solve this equation:
8cos2x + (-sinx) = 0?
Use the fact that $\cos(2x) = \cos^2(x) - \sin^2(x) = 1-2\sin^2(x)$ and solve the quadratic in terms of $\sin x$
How is: \cos^2(x) - \sin^2(x) = 1-2\sin^2(x)?
10:38
$\cos^2(x)+\sin^2(x) = 1$
10:48
@Hippalectryon I've been developing a revolutionary theorem which will allow the calculation of very complex integrals.
Hi guys. Anyone can help with this problem please? Find the number of binary strings of length 40, for which, in any sub-string of length 6, there are maximum 3 "1". Much appreciated!
11:35
@Alex.vollenga still need help? I'm not 100% sure but I think I have an idea
what if you define a recursive sequence $S_n = \text{number of binary strings of length $n$, for which, in any substring of length $6$, there are maximum $3$ $``1"$'s }$
and then resolve the recursion?
I have calculated the different arrangements containing max 3 aces in each string of length 6 - this is 42 out of the total 64. Furthermore, there are 35 such substrings in the string of length 40. But next?
isnt that what youre trying to find?
No I mean the different arrangements within EACH substring of length 6. What we want to find is the TOTAL number of all strings of length 40, for which in ANY substring of length 6 there are at most 3 aces
user84215
12:01
My account was suspended about 30 minutes because of pinging several users. It was a joke.
well let's try my way @Alex.vollenga
let $S_n$ be the number of binary strings of length $n$ for which in any substring of length $6$, there are at most three $1$'s
what's $S_{n+1}$? hmm
it's going to be $S_n + \text{ something }$
do we have an extra $\binom {n+1} 6 - \binom n 6$ to consider? I'm bad at combinatorics
oh wait, we can do this all with combinatorics, can't we?
I'll wait for someone else to do it
@GFauxPas: Every time we move from one substring of length 6 to its next one, we remove one bit at the beginning and we add one bit at the end. The new one may or may not have max 3 aces.
so does $AA9999A$ count?
I'm trying to figure out how many more cases are added to the mix when moving from a string of length $n$ to a string of length $n+1$
@GFauxPas we are talking about binary strings, only 0 and 1. They must be of length exactly 6
oh, you said "ace", so I thought you meant cards
$110001$ has three ones, or nah?
12:15
no no, "1"
yes 110001 has exactly 3
But each such string can also have 2, 1 or 0
I'm asking like this, and I don't know the answer,
if I tell you, that, for example, $S_{20} = 100$
can you tell me $S_{21}$?
Let's start with $S_7$
I don't know - I guess it depends on the first - last bit?
you told me $S_6 = 42$, right?
I dont know the answer so I dont know if this is a dead end, this is a joint effort Alex :)
12:19
aaaahhh I see what you mean!
so what's $S_7$?
I will try with 7 - this is easy!
I can list the arrangements and then try to deduce the formula backwards!
If you can give me $S_{n+1} = f(n)$, then we can resolve this
I wish I knew hahaha
but I will try with 6 and 7
Wait!
right, but under certain hypotheses, you can then evaluate $S_k$ WITHOUT knowing $S_{k-1}$
if you have $1, 2, 3, \cdots, n+1$ places, then from $1, 2, \cdots, n$ you have $S_n$
and from $2, 3, 4, \cdots, n+1$ you also have $S_n$
so is it just $2 S_n$?
doesnt seem right
12:23
yes. But how can we find ONLY the arrangements for 7 bits that meet the criteria?
well how did you figure out $S_6$?
What I mean is the following: with 7 bits, we have 123456 and 234567, right? Only 2 subsets
for each subset, the arrangements that contain max 3 aces, are 42
How do we move from the first subset to the next one?
I want to say just double it each step, but that doesnt feel right
12:25
no
because the two sets are overlapping
by 4 bits
4 bits out of 6 are common
$2S_n - S_{n-4}$?
still doesnt seem right
The final result is 11428527951 - if this is of any help!
Which, if divided by 35,
it gives an integer outcome
(which means we are in the right path hahaha)
maybe easier to find the RATIO instead of finding the number
so $T_6 = \dfrac {42} {2^6}$?
0
Q: A question on convergence in Sobolev norm Vs convergence at a point of isolated discontinuity

Rajesh DachirajuLet $\Omega$ be a convex, open subset of $\mathbb{R}^d$, with a smooth boundary. Consider the set $$M = \{f/f:\Omega\to\mathbb{R} \bigwedge f\in C^0(\Omega)\bigwedge \lvert| f|\rvert_{W^{1,d+1}}\in\mathbb{R}\}$$ From Morrey's inequality, $f\in M \implies f\in C^{0,1/(d+1)}(\Omega)$. Lets take an...

I read somewhere a while ago that given a Hermitian matrix $A$, it can be made PSD by adding $\pm ||A|| I$, but I cannot remember which norm is being used. Does anyone know?
12:39
@GFauxPas I found that for string of length 7, we go from 42 to 74. Does this make any sense?
Is there an example for a linear algebraic group know whose $\mathbb{Q}$-rank is smaller than its $\mathbb{R}$-rank?
honestly Alex I'm not sure
74 out of 128
@user193319 I found this quora.com/…
so maybe it dosnt matter
oh wait, wrong title
ignore that completely
12:54
I found this: oeis.org/…
But there is no formula :(
Just numbers and only up to 33...
:( maybe there is no simple formula
I tried to find a pattern with no luck :)
well, what I was trying to do, which you might find helpful in other sequences
is to try to find $S_n = f(S_{n-1}, S_{n-2}, \cdots, S_0)$
Then, there techniques to write $f(S_{n-1}, S_{n-2}, \cdots, S_0) = \phi(n)$
yes, the recursive formula
but, alas, there doesnt seem to be one here, that we can think of at least
Alex, @ me if you find the answer, I'd be curious, thanks
13:04
indeed! I also tried to find a formula for the differences S(n)-S(n-1) etc, again with no luck!
I believe it is very difficult to calculate how you advance from 6 to 7 for example!
> Are all prime numbers the same in all bases? If 21 is a prime, are
10101 (in binary), and 15 (in hexadecimal) also primes?
What the flying carpet?
what does that mean
better to ask questions than not to, though, Leaky
Hi chat
hello Fawad, good morning
Morning
I am searching for question on mse. Integration of 1/x using x^0/0 formula. I remembered simplebeatifulart answered it
13:20
@Fawad Just a second... I have one somewhere.
@Fawad Interpret it as $\displaystyle \lim_{n \to -1} \int x^n \ \mathrm dx$
$\displaystyle \lim_{n \to -1} \int_1^x t^n \ \mathrm dt$
$= \displaystyle \lim_{n \to -1} \frac{x^{n+1}-1}{n+1}$
$= \displaystyle \lim_{n \to 0} \frac{x^n-1}{n}$
$= \displaystyle \lim_{n \to 0} \frac{\exp(n\ln x)-1}{n}$
$= \displaystyle \ln x$
0
A: Proof that $\int \frac{1}{x}$ is $\ln(x)$

Simply Beautiful ArtHere's a fun proof that I made in this answer: Consider the more basic power rule: $$\int_1^xt^{n-1}\ dt=\frac{t^n}n\bigg|_1^x=\frac{x^n-1}n$$ Of course, we can't simply have $n=0$, but if we take the limit as $n\to0$, you end up with $$\int_1^xt^{-1}\ dt=\lim_{n\to0}\frac{x^n-1}n$$ And sinc...

19
A: Demystify integration of $\int \frac{1}{x} \mathrm dx$

robjohnThere are several approaches. One is based on the limit $$ \lim_{n\to\infty}n\left(x^{1/n}-1\right)=\log(x)\tag{1} $$ By adjusting the constant of integration, we have $$ \int x^n\,\mathrm{d}x=\frac1{n+1}\left(x^{n+1}-1\right)+C\tag{2} $$ Taking the limit of $(2)$ as $n\to-1$ and using $(1)$ yiel...

Thanks thanks. Today we were officially started with integration
you won't find a better teacher
My math teacher don’t know that can be used using limits concept
13:35
your teacher, I'd guess, knows; but she is using a different approach
what definition of $\ln$ do you use in your class?
sometimes you use this is a definition: $\ln x = \displaystyle \int_1^x \frac 1 t \, \mathrm dt$
@GFauxPas Thanks for your help. Talk to you soon!
sure
Oversleeping is when you wake up and say, "I feel well-rested—this is wrong."
"for example, $g(X)$ is always discrete if $X$ is discrete"
@GFauxPas log x is value wich we get when raised to power of 10
13:38
@Fawad ln, not log
that doesn't seem right, let $g(x) = 5$?
@GFauxPas where did you get that from?
a critique of my answer on the site
@LeakyNun log (base 10) ln (base e)
Your first definition is wrong (on several counts simultaneously) and should read $$E(g(X))=\sum_jy_jP(g(X)=y_j)$$ The rest of the post is also rather confused at times (for example, $g(X)$ is always discrete if $X$ is discrete) but this initial mistake really messes everything up. — Did 33 mins ago
13:39
@Fawad we are asking ln
@GFauxPas I think they just take it as log base e in class
Oh. When raised to e
where e is the limit of the famous function
Fawad, if you had to define $e^4$ in words, without telling me the answer, what would you tell me?
@GFauxPas he's still in high school. no need for that formality.
don't think too hard
Calc II in high school?
nice
13:40
when did he say Calc II?
He's in India. Not everyone follows the same syllabus.
he's learning integration, I guess that's Calc I, mea culpa
@GFauxPas e power four
right, so, $e$ multiplied by itself 4 times, right?
what's $e^{1/2}$?
1 min ago, by Leaky Nun
@GFauxPas he's still in high school. no need for that formality.
Square root of e
13:41
yes, I am disagreeing with you
but you can see that $e^{\sqrt 2}$ is not so simple, right?
I'm not telling you what it is, I'm just saying it requires some thinking about
@GFauxPas I'm not seeing the purpose of this.
@GFauxPas e power square root of 2
I'm saying that the reason the professor might not be teaching the log based on the power rule is that it would be circular
13:42
You can listen to leakynun Gfaux
Okay, my bad then
@GFauxPas you married or not? Last time we talked it was you birthday
nope, not married, why do you ask?
9 mins ago, by Fawad
My math teacher don’t know that can be used using limits concept
$\lim\limits_{\substack{p\to\sqrt2\\p\in\Bbb Q}}e^p$
13:44
@AkivaWeinberger just use the limit of the function or the power series
Achievement get: Use "substack" correctly on the first try
@LeakyNun I was answering the question on how to define $e^{\sqrt2}$
@AkivaWeinberger so am I
@GFauxPas maybe s/he just doesn't know
When you say "just use the limit of the function" isn't that what I just did
18 mins ago, by Leaky Nun
$\displaystyle \lim_{n \to -1} \int_1^x t^n \ \mathrm dt$
$= \displaystyle \lim_{n \to -1} \frac{x^{n+1}-1}{n+1}$
$= \displaystyle \lim_{n \to 0} \frac{x^n-1}{n}$
$= \displaystyle \lim_{n \to 0} \frac{\exp(n\ln x)-1}{n}$
$= \displaystyle \ln x$
it could be he or she just doesn't know, could be
13:46
I don't see the need for the teacher to know ^ this @GFauxPas
ah, okay, gotcha
@AkivaWeinberger I mean, $\displaystyle \lim_{n \to \infty} \left(1 + \frac{\sqrt 2}{n}\right)^n$, the limit of the sequence not function, mea culpa
@LeakyNun definite integral is not stated. First indefinite integral
@Fawad fundamental theorem of calculus
he may not have gotten to that yet
13:54
@LeakyNun can you explain last step of evaluating limit
user84215
I want to repeat my post since I think some people have not seen it. I hope you are not angry with me.
We want to start it very soon.
user84215
NOTICE NOTICE
user84215
I have a new idea although it will likely be blamed and scorned by you as before. I have created a room called "LTD" (Learning Through Discussion) [LTD: Topology](https://chat.stackexchange.com/rooms/62439/ltd-topology). This is basically different from ordinary classes or study groups. In this room, there is a leader who presents study materials (in this case: Topology) and all in the room discuss with each other about them.
For example, suppose a theorem is stated then they can speak about the main ideas in the theorem and its proof (not mentioning details as it does in ordinary classes),
@Fawad $\displaystyle \lim_{h \to 0} \frac{e^h - 1} h$
Got. Thanks
Why you took integral from 1 to x? Why didn’t statrted from 0?
@LeakyNun
13:57
@LeakyNun Yeah, but it's still a bit of a pain proving that it satisfies the $e^n=e\cdot e\dotsb e$ for $n\in\Bbb N$
when you do it that way
@AkivaWeinberger yes it would be
In any case, I admit it doesn't matter
@Fawad because it would be infinite...
Doesn't matter what definition we use as long as we can prove they're equivalent
@AkivaWeinberger actually it wouldn't
$\displaystyle \lim_{n \to \infty} \left(1 + \frac{m}{n}\right)^n$
$= \displaystyle \lim_{n \to \infty} \left(\left(1 + \frac{1}{n/m}\right)^{n/m}\right)^m$
$= \displaystyle e^m$
14:20
Hi chat
What's going on ?
hi
chillin', you?
Same
how's your vacation going so far?
So far so good (it started yesterday noon actually)
I should even be taking exams right now, but I'm not interested in them and wanted my break badly
breaks are important
without them you get burnt out
14:33
Exactly
[A rant that looks so random when completely void of context] English culture seemed to finally pissed me off today
On to a 4 year break
but I must say, that's one hell of an accomplishment since in this whole reality, there are really only two classes of objects that can set me off and those morons managed to hit those
Actually, nope I was wrong, its not an english thing, my bad
create a third class to increase your odds
in fact, if you had infinitely many; nothing could set you off
@LeakyNun Oh, you're absolutely right, I forgot about that
Well, there's a little snag for the negative integers, but whatever
You just need to know $\lim_{n\to-\infty}(1+\frac1n)^n=e$ as well
which seems doable
14:55
hi chat
2 days ago, I played with a real number system with a continuum number of places and then conclude it is basically Laplace transform of a given functon evaluated at 1
$$F(1)=\int_0^{\infty}f(x)e^{-x}dx$$
Is there a simple proof that $1-x^2\ge(1-x)^x$ when $0\le x\le1$?
First thing that comes to my mind is differentiating
However that's purely barbaric
15:10
maybe use $1 - \dfrac 1 x \le \ln x \le x - 1$?
so $1 - \frac 1 {1-x} \le x \ln (1-x) \le -x$, not sure this is getting anywhere
$\dfrac 1 {x-1} \le \ln (1-x) \le -1$
nope, sorry
$1-x^2 = (1-x)(1+x)$
Well, it's equivalent to $\ln(1-x^2)=\ln(1-x)+\ln (1+x)\geq x\ln(1-x)$
so $\ln(1+x)\geq (x-1)\ln(1-x)$.
But the LHS is positive and the RHS is negative so long as $0<x<1$.
I guess one doesn't need the log for that, though. If you divide both sides by $1-x$ that becomes $1+x\geq \frac{1}{(1-x)^{1-x}}$
oh, wait. I'm being silly---$\ln(1-x)$ is negative if $0<x<1$.
But $1+x$ is the tangent line of $(1-x)^{x-1}$ at $0$, so this should just be a convexity argument.
15:32
I read somewhere a while ago that given a Hermitian matrix $A$, it can be made PSD by adding $\pm ||A|| I$, but I cannot remember which norm is being used. Does anyone know?

« first day (2541 days earlier)      last day (2774 days later) »