« first day (1379 days earlier)      last day (3936 days later) » 

15:00
@Alex Yes and no. I would take it in the ordinary meaning of continuity of functions between topological spaces. But a function $\Omega\to\mathbb{R}^n$ is continuous if and only if all component functions are continuous, so ...
@DanielFischer Uhm wouldn't the topological and component continuity agree?
@MartinSleziak Well, I guess in the end it's best to have confidence in the validity of your own prove instead of always believing Wolfram :)
@Alex That's the point. The two concepts are identical (for functions into topological products).
@DanielFischer Oh I see you, you are saying you choose to view it from the more general topological idea.
15:15
can the information of any formal proof ( no matter how rules of inference and assumptiosn,etc it uses ) of any formula be condensed in simply one formula on the form of an implication ( that already considers all applications of rules of inferences and also embedds all new assumptions, say , in the middle of a proof ) ?
This matrix multiplication is freaking me.
r9m
r9m
@robjohn should I make it available somewhere ? :D .. I was just saying I couldn't help but name the file like that .. :P
@Chris'ssis it should be $$\frac13\gamma-\frac13\log\left(\frac1{3\pi} \cosh\left(\frac{\pi\sqrt3}{2}\right)\right)$$
@robjohn It's awesome! What ways did you use?:-)
user116900
@Sawarnik What is so hard about matrix multiplication?
15:29
@r9m I don't know. If they are all on chat, they should be available to anyone, so there should be no problem making a collection
@JasperLoy Its kind of very strange.
user116900
@Sawarnik Just remember that it is done row by column.
Ok :)
c c
c c
I'm reading a nice intro to tensors p19, $tr(A)=\frac{1}{\vec{c_1}*\vec{c_2}*\vec{c_3}}[\vec{c_1}.A.(\vec{c_2}*\vec{c_3}) + cycl.]$
I don't understand why could be cycl first (cyclic terms?)
@Chris'ssis I noticed it was $$\frac13\gamma+\frac13\sum_{k=1}^\infty\frac{\zeta(3k)-1}{k}$$ and just did the usual tricks with zeta
15:31
@r9m :P
@Chris'ssis Then used the product I computed yesterday :-)
@robjohn :-)))))))))))))))))))))))))) AWESOME
user116900
@Chris'ssis Too many )
@robjohn How difficult does it seem to you? Is it good for a uni contest?
c c
c c
I think it's a cycle, and since $\vec{c_2}*\vec{c_3}=\vec{c_1}$ it's: $tr(A)=\frac{1}{\vec{c_1}*\vec{c_2}.\vec{c_3}}[\vec{c_1}.A.\vec{c_1} + \vec{c_2}.A.\vec{c_2} + \vec{c_3}.A.\vec{c_3}]$ anyone familiar with tensors could confirm, please?
15:35
@JasperLoy I apologize ... :-)
user116900
@Chris'ssis I think I prefer Spider-man to Superman. I just watched The Amazing Spider-man 2 today.
@JasperLoy I didn't see it yet. Is it cool?
user116900
@Chris'ssis Yes, but the ending is quite sad, his girlfriend died.
@JasperLoy :(
I prefer Batman (The Dark Knight)!
15:41
@JasperLoy Spoiler
Also read the comics, snaps
hi all, what is the name of a function that does not have an inverse function? such as y=x+exp(x)? I thought it was transcedental, but according to wikipedia that is not it
non-injective
$$
\begin{align}
\frac13\lim_{n\to\infty}\left(\sum_{k=1}^n\frac{\zeta(3k)}{k}-\log(n)\right)
&=\frac13\gamma+\frac13\sum_{k=1}^\infty\frac{\zeta(3k)-1}{k}\\
&=\frac13\gamma+\frac13\sum_{k=1}^\infty\sum_{n=2}^\infty\frac1{kn^{3k}}\\
&=\frac13\gamma-\frac13\sum_{n=2}^\infty\log\left(1-\frac1{n^3}\right)\\
&=\frac13\gamma-\frac13\log\left(\prod_{n=2}^\infty\frac{n^3-1}{n^3}\right)\\
&=\frac13\gamma-\frac13\log\left(\frac1{3\pi}\cosh\left(\frac{\pi\sqrt3}{2}\right)\right)
\end{align}
$$
user116900
Geezis, who approved the wrong edit? I had to reedit...
@MikeMiller Thanks, had something with bijective in mind as well
15:45
@JasperLoy reddit
user116900
@N3buchadnezzar Never visited.
user116900
@N3buchadnezzar That is what they told me in the other room.
@r9m Hey.
@Bernhard Well, if the function is injective but not surjective, its inverse will just be defined on an open interval (and not all of $\Bbb R$). So injectivity should be the only requirement.
15:50
@robjohn This is exactly my way too.
@Chris'ssis: I just proved this this morning: $$\log(\Gamma(1-x))=\gamma x+\sum_{k=2}^\infty\frac{\zeta(k)}{k}x^k$$
I'd never seen that before today
@robjohn I know this identity for some time.
@GabrielR. I have something for you. Are you there?
@robjohn zeta function can be expressed in terms of polygamma.
@Chris'ssis That leads to $$\log(\Gamma(2-x))=(\gamma-1)x+\sum_{k=2}^\infty\frac{\zeta(k)-1}{k}x^k$$
which converges a lot faster
@robjohn I see.
@Chris'ssis In any case, it allows computation of non-integer Gamma values, much easier than others I have used
15:56
@robjohn Indeed.
@robjohn I didn't know you weren't aware of this one (I could have told you about it).
@Chris'ssis I may have seen it on the Wikipedia page since it is there, but I never really noticed it
These identities are really amazing.
@robjohn Have you seen Gabriel's question to me yesterday? The one involving harmonic numbers?
user116900
Geezis, another high rep user did a stupid edit.
@Chris'ssis I don't think so... I will look.
@robjohn I did it elementarily (happily).
user116900
15:59
Both the low rep user and high rep user changed the meaning of the question I answered, LOL.
@robjohn take them pls.
user116900
Geezis, I cannot stop laughing.
@JasperLoy laughing is good.
user116900
I think I should just delete my answer, lol.
brb (I need to buy some food for my pets)
16:08
@chris'ssis I'm there
God @chris'ssis I didn't get the second link
user116900
@GabrielR. That's the problem with deleting, lol.
@GabrielR. are you done?
@chris'ssis yes thanks
16:12
I'm really out now for 20 min or so.
brb
We have several post on meta about turning posts into CW by sufficient number of edits, for example here or here. (The later is a which is tagged as , to make things even more confusing.)
Not so long ago it was announced that posts will be no longer automatically turn into CW.
Should we now somehow correct the old answers? (By posting a new answer about change in the policy, or making a comment asking the answerer to update their answer.) Or should we simply let it be, it will come up on meta eventually?
user116900
@MartinSleziak Wow that is new to me.
user116900
@MartinSleziak I think you should post a new answer.
The thing I am concerned about is that there are probably several threads on meta that are related to this. I am not sure whether bumping all of them is a good idea.
(But of course, I might have overestimated the number of posts where this information is relevant.)
user116900
@MartinSleziak I think it's fine. If people scold you for bumping, just ignore them...
16:17
@JasperLoy What is Community Wiki?
user116900
@MartinSleziak You can quote me in your defence, lol.
user116900
@GabrielR. Means anyone with a certain rep can edit your post, and you no longer get points on your post even with new votes.
Here is another post on the same topic.
user116900
One thing you can do is to find them all, and then edit at one shot so that people know you are doing serial editing.
user116900
In this case, bumping is also good to draw people's attention to it. For example, I did not know about the change in policy!
user116900
16:23
@N3buchadnezzar Also never visited, lol.
I have posted one answer and made comments on some answers, that they should be updated. Maybe they are not so many questions about this, as I originally thought.
user116900
OK, thanks for your work.
@MartinSleziak Not a good idea: that would cause them to all to come to the top of the list. That was one purpose of the auto-CW conversion: prevent constant edits from pushing the question to the top of the heap. Now, the mods get flagged and then we have to decide whether it is someone gaming the system instead of it being automatically done.
@MartinSleziak So, a lot of the posts might still be CW. When a post was marked CW before, the mods had to be flagged by the OP to fix it.
user116900
@robjohn Anyway, this is meta, so nobody gets more rep from the bumping.
@JasperLoy Isn't the auto-CW conversion also removed from main posts?
user116900
16:35
@robjohn Yes, but that is not my point. =)
@robjohn I think you might have misunderstood me. I was not talking about all posts that are CW. I was just saying that it might be useful to update information about CW-fication posts by SE software.
@JasperLoy was I replying to your comments?
user116900
@robjohn Oh, OK then =)
Or maybe I have misunderstood you.
So you're saying that the new policy is bad idea? Or that my suggestion to update meta is bad idea?
@MartinSleziak like which posts?
16:38
Like the two posts I've bumped a while ago on meta.
@robjohn Hey
@MartinSleziak If removing CW moves the posts on the front page, I don't think it is worth it. However, I am not sure that it does.
@robjohn I have never mentioned removing CW from posts.
For example here on our meta an answer says, that a post is turned CW after 10 edits. This is no longer true. So I think it is reasonable to update that answer.
@MartinSleziak Ah, I see. You want to change the FAQ-like posts that talk about CW... yes, that should be done.
@JasperLoy It was not Martin's point either, :-)
user116900
@robjohn You need more coffee!
16:41
The reason I brought it up here was that the same thing might be mentioned in many posts on meta. (I am not sure how many.) So if this meant that we have to bump say 20 posts and each of them just add this new information, that would be not ideal.
user116900
@MartinSleziak But if you left a comment for others to edit, won't that also bump them?
@MartinSleziak I would think that this list would contain most of the posts that might need editing, and from a quick sampling, not many do.
@JasperLoy Yes, if someone edits their post, it causes a bump. But I did not feel like changing someone's answer substantially. That's why I chose to comment and not to edit.
Hi @Jas
user116900
@MartinSleziak Ah, and that is why I said you should just go ahead and post 20 new answers!
user116900
16:44
@ಠ_ಠ How is everything?
@robjohn I had brief look at questions tagged .
Hi @Jasper and mr eyeglasses and @robjohn
It seems that they are not too many questions about this.
user116900
@TedShifrin Hi!
@N3buchadnezzar Hay!
@TedShifrin howdy
user116900
16:45
@MartinSleziak Yes, which is why all the more you should bump them all right now!
@JasperLoy That was just an estimate. (Probably wildly exagerated.)
@robjohn Meh, integration :p
user116900
Meh and Bleh are very interesting words...
Heya @N3
$$
\int_0^\infty \frac{x^{3/2}}{(1+x)^4}\,\mathrm{d}x
$$
I am trying to wrap my head around why $\int_\varepsilon^R = \int_R^\varepsilon$
Keyhole contour,
16:47
Learning complex variables, @N3?
Exam in two days =)
@N3buchadnezzar I have an exam on integrals in two days too :)
Off by a minus sign there, @N3, unless you change branch of your integrand.
user116900
@N3buchadnezzar No meat, no pudding
@TedShifrin How should I choose that ?
16:49
@N3buchadnezzar That looks like a Beta integral
/I am not sure what branch to pick, except the principal branch
@robjohn YEah I know, but I have to use complex :p
user116900
Pick from the lowest branch.
user116900
Those are lhf.
Not equal. You change direction and what happens to $z^{3/2}$ when you go around the keyhole?
@N3buchadnezzar Ah... is this in a question or just here? (I don't want to write up something here if there is a place on main for it already)
16:50
You go 2pi i around it
@robjohn Not on the main
@N3buchadnezzar okay... I will write something up
@TedShifrin Not sure what you mean
Should I quit talking, then? @robjohn
16:52
Hey everyone.
How do you define $z^{3/2}$, @N3?
I only have problems showing that the two paths traveling in opposite directions are in fact equall. I have been able to compute the residues and everything else.
Hi @Anthony
@TedShifrin Nah... go ahead. Yours might be better
@TedShifrin Something like $e^{\frac{3}{2} (\log |z| + i \theta)}$ ?
16:53
OK, @N3, so what happens when $\theta$ changes by (almost) $2\pi$?
Are lognormal returns in finance always base e?
I asked this yesterday, but if you have two identical balls and two distinguishable bins, is there any chance that the probability of ending up with a ball in each bin is 1/3, and not 1/2? This is the case with electrons, but I guess they're special?
Is the base e lognormal distribution uniquely defined by its mean and variance?
Something like
$$
e^{\frac{3}{2}(\log |z| + 2\pi i)} = e^{\log(|z|^{3/2})} e^{3i \pi } = -z^{3/2}
$$ ?
@TedShifrin Maybe you'll know, you said you're gonna teach probability :P
16:56
So that answers your question, @N3: you change direction and the function changes by a factor of $-1$, so altogether ....
@Jas I hate walking to school in the heat
user116900
@ಠ_ಠ Have you seen a counsellor?
I don't know your question, @Anthony, and I don't get the electrons. Presumably there's something conditional going on?
@TedShifrin Scratches head... Gimme a few minutes to think about it. Sigh, I should see it
@Jas I'm not at school yet and I have finals
16:58
@N3: what happens if you do a line integral backwards on a curve?
user116900
@ಠ_ಠ OK, right after that then
@TedShifrin But if you have two identical balls, and two distinguishable balls, the probability that both will end up in different bins-that's 1/2?
@TedShifrin You get a minus sign, eg chooses the opposite parametrization
Right @N3, and the function is essentially $-1$ times the other one ...
Start at the beginning, @Anthony.
Okay so it is because the value of log z increases each time you do a full rotation
17:02
@TedShifrin You have two identical balls, and two labeled bins $1,2$. You throw both balls into the air, and they must land in either $1$ or $2$... It makes sense that the probability that they both land in different bins is 1/2, I just always thought that removing the labels gave strange results.
Right @N3
Thats where my problem was. Thanks =)
As a soon to be educator, it is often hard to pinpoint where a student has missintrepetations or false believes of a subject..
Experience, @N3, plus caring enough to think about it.
user116900
@N3buchadnezzar What level will you be teaching?
@Anthony: you list cases and 2 out of 4 times they're in different bins. Now what if they're labeled? I don't see a difference.
17:05
@JasperLoy 16 to 18 years
@TedShifrin Thank you for caring about my silly questions =)
user116900
@N3buchadnezzar Very good. I quit teaching here because the system is rubbish.
You're fine, @N3...
Whats with all the ...`s today ?
@TedShifrin Alright... I guess I was just misunderstanding what unlabeled meant? I guess it only changes number of outcomes, not probabilities.
I always do that :( ... :)
17:08
...
@N3buchadnezzar You can question them where you think they have doubts?
Oh, I see your issue, @Anthony. You were thinking 3 cases in the unlabeled case, but they are not equally likely!! Always have to be careful about that!
user116900
@N3buchadnezzar I just hope to go to another place for further studies and maybe do some teaching there after that.
Well I mean very obviously if I take two "identical" balls and throw them, they're still distinguishable in a sense, each has a 1/2 probability of landing in either bin. With electrons, they are actually identical, and their probabilities interfere, you can't say one will one one way with half probability etc. I don't know, it's just bothersome. @TedShifrin
@N3buchadnezzar I get $\frac{21\pi}{8}$ if I didn't make any mistakes.
17:12
So I don't know what electrons you're working with here. Are we taking repulsive forces into account?
Something like
$$
\int_0^\infty \frac{z^{3/2}}{(1+z)^4}= \pi i \cdot \lim_{z\to -1} \frac{d^3}{dz^3} z^{3/2}
$$ ?
Ugh, I prefer using Laurent series to those derivative formulas @N3
@N3buchadnezzar checking I see a mistake... hang on a sec
@TedShifrin Do you have me on ignore?
17:16
I see a mistake as well ugh
r9m
r9m
@TedShifrin ah .. I made that mistake ..
@r9m And it really surprised me....
@r9m, which?!
r9m
r9m
@TedShifrin thought the three cases occur with same probability ..
Ah @r9m
Be careful with your branch of $z^{3/2}$, too, @N3.
17:20
@N3buchadnezzar error fixed... the answer should be $\frac\pi{16}$
@robjohn I got the same
@TedShifrin So I can not derive willy nilly?
@N3buchadnezzar I can post mine here if you wish...
$$
\begin{align}
\int_0^\infty\frac{x^{3/2}}{(1+x)^4}\,\mathrm{d}x
&=\int_0^\infty\frac{2u^4}{(1+u^2)^4}\,\mathrm{d}u\\
&=\int_{-\infty}^\infty\frac{u^4}{(1+u^2)^4}\,\mathrm{d}u\\
&=\int_\gamma\frac{z^4}{(1+z^2)^4}\,\mathrm{d}z\\
&=\int_\gamma\frac{z^4}{(z+i)^4(z-i)^4}\,\mathrm{d}z\\
&=\int_\gamma\frac{(w+i)^4}{(w+2i)^4w^4}\,\mathrm{d}w\\
&=\frac1{16}\int_\gamma\frac{(1-iw)^4}{(1-iw/2)^4w^4}\,\mathrm{d}w\\
&=\frac1{16}\int_\gamma\left(\begin{array}{c}1-4iw-6w^2+4iw^3+O(w^4)\\\times\\1+2iw-\frac52w^2-\frac52iw^3+O(w^4)\end{array}\right)\frac{\mathrm{d}w}{w^4}\\
did it anyway
user116900
Yay, I have reached 300, time to retire (but I will keep my account), lol.
@JasperLoy 300 for 4 lhfs .... hmm.
@robjohn =)
17:32
@N3buchadnezzar $\gamma$ circles the upper half-plane counter-clockwise
user116900
@robjohn Are you here?
Glad I'm not the only one retiring, @Jasper!
@JasperLoy yes
@TedShifrin don't tell me you are leaving!!
user116900
@robjohn I am not sure why two minuses got deleted when I edited.
17:39
We all leave eventually, @robjohn:) I'm pretty sure I'm retiring as prof a year from now.
Oddly there are people in my calculus 4 class that are graduating this semester..weird
@TedShifrin Oh, are you staying with MSE, then?
That shouldn't happen, mr eyeglasses. Weird
user116900
@robjohn If the minuses should be in there, they should be put back, but I swear I did not delete any minus sign.
My presence is reducing, @robjohn, but we'll see ...
17:41
@eyesperson Are you among them?
@JasperLoy I can accept and improve your edit...
user116900
@robjohn It has been accepted. But now I no longer know whether asker intends for the minus sign there or not. The edit history is strange.
@JasperLoy hmm
user116900
@robjohn He might have been editing at the same time I was, or something like that, so that something strange happened.
@JasperLoy yes and perhaps edited during the initial 5 minutes
user116900
17:43
@robjohn I left a comment on the post to ask him.
@JasperLoy good idea
@robjohn Is the initial 5 minutes a bit different?
@Sawarnik any subsequent edits someone makes within 5 minutes of an initial edit get folded in, they don't show in the history
That's not good.
That allows people to fix up things without a bunch of edits being listed
17:46
@robjohn Well fine, but it should show up somewhere. Otherwise confusions will arise like this one.
@Sawarnik I don't know if it is folded in all records, but I bet it is.
user116900
@robjohn OK, he has reedited it himself.
@Sawarnik It just seems a bit pedantic to keep every single edit. The suggested edit histories need other fixes before anything of this nature is fixed. They should be rendered properly for one
@JasperLoy that is good
user116900
300 is a nice number to stop. Because I only need 200 to get 300. 100 comes from association bonus.
17:54
Oh, I'm great at analysizing edit shenanigans within the grace period.
user116900
Anyway, matter resolved.
@GraceNote Ah, so it is recorded in gory detail...
Not really
We don't record the stuff that happens in the five minute grace period, otherwise that'd defeat the purpose of it in addition to creating a lot of bulk data.
@GraceNote that's why I thought it was there
I don't know if I should even ask what you mean then ;-)
I am, however, a connoisseur of temporal relations, with access to things like when people edit, so I can derive for example that in this case Jasper started editing during the grace period, before Marko added the minus signs during the grace period, and so indeed his revision makes it look like he removed the minus signs.
Basically I can confirm what people were pretty sure was the case, and tell you that indeed that is precisely what happened here.
user116900
17:58
Is there an auto downvote on a post we flag as spam? It seems to happen.
@GraceNote okay... that makes sense
@JasperLoy Flagging as spam or offensive applies a downvote, yes.
Hi, a quick question: what is the method to use on this one $\int \frac{dx}{(x-3)^3\sqrt{x^2+3x+y}}$?
user116900
@GraceNote There is now a very persistent spammer on ELU, love astrology, he's at it right now, lol. And has been for a few days.
Oh, this fellow
18:20
If I have a one form $\omega = \omega_\mu dx^\mu$ and a vector field $Y = Y^\nu\frac{\partial}{\partial x^\nu}$, can someone explain to me what $\omega(Y)$ means. Like what is the action of a one-form on a vector field?
Off to lynch some lunch... BBL
I understand that a differential form is part of the dual space of the vector space that the vector field $Y$ is a part of so $\omega(Y)$ should map to some real number. I'm just not sure how writing out what $\omega(Y)$ should look.
user116900
I am wondering why I am getting so many votes for lhf these days, never happened before, lol.
user116900
18:43
user116900
16 zombies in this chat.
19:19
How come the limit of $\frac{1}{xy}$ as $(x,y)$ approaches $(0,0)$ is not $0$?
I mean infinity
user116900
@ಠ_ಠ x and y can be positive and negative...
Nvm I figured it out
user116900
@ಠ_ಠ I sort of told you the answer.
@JasperLoy: stop now while you are half of 666
user116900
@robjohn 999 is the police here.
19:26
@JasperLoy even more reason to stop at 333
user116900
Interesting how some people write the entire post in LaTeX, even all the words, lol.
@JasperLoy it's better than entirely image
When computing $$ \sum_{i=3}^n 5i^2-3i+2 $$ Do I split the sum in into three or 2?
Sorry @Jas the chat doesn't update automatically
user116900
@ಠ_ಠ Are you on mobile?
19:32
I have to manually refresh so then I see tons of new messages at once
user116900
@rubito You can split into 3.
Yes @Jas
user116900
@ಠ_ಠ Luckily I don't use internet on my phone
@JasperLoy I'm not sure what to do with the 2 then. Like, do I just add that at the end after I'm done finding the sum of the first two terms
user116900
@rubito Yes, you add up the 3 sums. You just need to know how to sum a constant, and that is trivial.
19:35
@Jas I hate the school computers, I swear theyare slower than my phone
@rubito Do you know formulas for $\sum\limits_{i=3}^ni^2$, $\sum\limits_{i=3}^ni$, and $\sum\limits_{i=3}^n1$?
user116900
@rubito Do you know what the last sum is?
@JasperLoy correct me if I'm wrong but the sum of the constant will just be the constant right?
@JasperLoy I do know the formulas for that
user116900
@rubito Nope, it is k times the constant where k is the number of times you add the constant. Here k=?
@JasperLoy I was never given a $k$ constant
user116900
19:37
@rubito OK, the last sum is (n-2) times of 2, you know why?
@JasperLoy mmm not really? Like is $n-2$ one of the formulas? Because I can't find it in my notes
user116900
@rubito OK, the sum goes from i=3 to i=n, and that is (n-2) times altogether.
So I have the two sums for the two previous terms and then I have a $n*2$?
user116900
@rubito The final sum should be 2(n-2)
@JasperLoy Ohh I see, because that's our $upper index$ right?
user116900
19:43
@rubito The lower index is 3 and the upper index is n. From 3 to n inclusive of the endpoints that is a count of n-2. You sum 2 a total of n-2 times, so you get 2(n-2)
user116900
I think I will delete an answer because the asker still does not seem to understand it.
@JasperLoy Yeah, I'm having a hard time figuring out why is $n-2$. I know you explained it a couple times already
user116900
@rubito How many pages are there from page 3 to page 5 inclusive?
@JasperLoy Ohh! 2!
user116900
@rubito No, 3.
19:48
@JasperLoy Because you're including three?
user116900
Page 3, Page 4, and Page 5.
user116900
@rubito Yes
@JasperLoy Ok, it's starting to make sense. So like a rule of thumb would be to use the lower index minus 1?
user116900
@rubito Take the upper index minus the lower index and then plus one.
user116900
5-3+1=3
19:52
@JasperLoy Ok, that makes sense now! So that's the formula for constants?
user116900
@rubito Yeah remember to multiply that number by the constant.
@JasperLoy Awesome, will do! Thanks!
user116900
@rubito No problem, that would be 1000 USD, lol.
@robjohn @jasper @Mike @Pedro Any idea what they're finding wrong with my answer here?
Subtraction is hard @Jasper
@JasperLoy I'm afraid all I have left is a half pack of gum and a mini stapler
19:56
i'll take the half pack of gum :D
Not you, @skull.
tough crowd...
...i would have shared
20:12
@Ted Your comment to him seems like it explains what he's confused about
Thanks, @Mike. I just wanted confirmation. He wasn't asking for a correct argument, just for an explanation of whether his was right or not. shrug
Actually, there was a really good question asked by a faculty member yesterday. A not so well-known fact is that if the derivative is differentiable as a map, then you get symmetry of mixed partials (as opposed to the usual hypothesis of $C^2$). This is something you and @Pedro might be interested in thinking about.
Is this not so well-known fact in any of your books Professor @TedShifrin?
Nope, @skull. A bit too technical for my book. But important nonetheless, particularly in understanding calculus in infinite dimensions.
If I'm asked to find the length of the curve $(2a-x)y^2=x^3$ for $0\le x \le \frac{5}{3}$, can I make my life simpler by parametrizing $y=xt$? I don't think so, because the relationship between $x$ and $y$ is not linear, but otherwise the integral is impossible :/
@TedShifrin Suppose $f \colon \mathbb{R}^2\to\mathbb{R}$ has partial derivatives in a neighbourhood of $0$, and both partial derivatives are differentiable in $0$. Does that imply that the mixed second partials agree?
(Note: I'm not assuming that $f$ is differentiable at any point other than the origin.)
20:27
Hmm, now I'm not sure
Can someone verify that there is an error in the proof of en.wikipedia.org/wiki/Hilbert_projection_theorem when $H$ is a complex vector space? I'm interested in the part which shows that the minimizer is orthogonal to all vectors in $C$.
The expansion of $|(y+ta)-x|^2$ assumes that $\langle y-x,a\rangle$ is real
@DanielF: That seems to be the gist of the question I answered. I'll have to double-check Dieudonné in my office tomorrow.
@MarioCarneiro The proof treats only the real case. For the complex case, you can view the underlying real Hilbert space (effectively that is, take the real parts of the inner products), or mimick the proof using the complex inner product. Not much changes, except that instead of $2\langle u,v\rangle$ you have $\langle u,v\rangle + \langle v,u\rangle$.
Do you still have $t\in\Bbb R$? I don't see how this argument extends to the complex case, because all quadratics are negative somewhere on $\Bbb C$.
@TedShifrin Heh :D
20:41
No, I'm thinking it's not @DanielF ... I think I'll try to find a counterexample now. The hypotheses I had yesterday, I believe, were local.
@MarioCarneiro In the complex case, you don't get a holomorphic polynomial, you get $a + b\operatorname{Re} z + c\lvert z\rvert^2$.
Yeah, the difference between an orthogonal structure and a hermitian structure on $\Bbb C^n$ ...
And also $a=0$ and $b,c\in\Bbb R$, where $b$ is proportional to $\Re\langle y-x,a\rangle$, so that means that $\Re\langle y-x,a\rangle=0$. But then you still don't quite have the result.
Hi, I have another question on the convergence of sums:
Do you guys think 40 hours/week is enough studying to maintain a good grade in a real analysis course
20:54
When I got $ \frac{\sqrt{k+1}- \sqrt{k-1}}{3^k}$, can I use the ratio criteria to look for convergence?
i'm seeing somewhere that proof by contradiction can be summarized as the logical equivalence between ( p -> q ) and ([p and (~q) ) -> [r and (~r) ] ) for any formula r . But saw i'm thinking it should be a logical equivalence between ( p -> q) and ([p and (~q) ) -> F ) , where F could stand for any contradiction.
So, the initial logical equivalence would only be true and general if all contradictions could be expressed in the schemme of [ r and (~r) ] , that is, could be expressed as a conjunction of a formula and its negation .. but i think this is not the case.. is it ?

« first day (1379 days earlier)      last day (3936 days later) »