@TobiasKildetoft What I find strange about that question is that $[A,B]=\lambda I$ does show up in physics a lot in a form that amounts to $[\frac{d}{dx},x]=1$. But that's an exception that proves the rule, since these can't be represented by finite-dimensional matrices.
(There's a theorem that says that every continuous function with domain $[0,1 ]$ — that is, the closed interval — is bounded. Thus, if you have an unbounded function on $(0,1)$ — the open interval — then it can't have a continuous extension to $[0, 1]$.)
I have a question. Given $a,b>2$ $a,b\in \Bbb{N}$ I have to prove that $2^a+1$ is not divisible by $2^b-1$. My idea is to take cases. The first being $b>a$. It becomes very easy because $2^b-1>2^a+1$ and hence $ 2^a+1$ is not divisible by $2^b-1$. But how should I proceed further?
(For example, take $\frac1x$, with domain $(0,1)$. It's unbounded, and it can't be continuously extended to $[0, 1]$ because it can't be defined at $0$.) @usukidoll
@robjohn these days I derived an amazing family of infinite series, and also some very interesting integrals. Thinking to propose some new stuff in more journals.
if nothing else, one has (stealing from Wiki's page on the integral test) the bounds $$\int_N^\infty f(x)\,dx \leq \sum_{n=N}^\infty f(n)\leq f(N)+\int_N^\infty f(x)\,dx$$
@Semiclassical Sure, it's good to think of all variants. Then, I have under development a tool for calculating infinite series which has some little connections with what you propose. The idea is pretty sophisticated at this point, but as I said it's under development.
@Semiclassical one can also try this variant $$\sum_{n=1}^{\infty}\frac{H_n}{n}\left(\zeta(2)-1-\frac{1}{2^2}-\cdots -\frac{1}{n^2}\right)\left(\zeta(2017)-1-\frac{1}{2^{2017}}-\cdots -\frac{1}{n^{2017}}\right)$$
@user1618033 it's all about computing series and integrals in closed form or you also ask for asymptotic estimates, inequalities and other real-analysis stuff ?
@LeGrandDODOM There are also other interesting problems, not only computing series and integrals in closed form. I have some inequalities added and also limits that involve finding asymptotic estimates.
@Semiclassical ask the best mathematicians that are friend of yours, and see what they say. They require special techniques to calculate in closed form. I'm not aware of the existence of any such paper.
It might be a great suprise to me to find one though. Perhaps I would be glad to mention that paper in my book too.
@MikeMiller Nothing new, sorry to disappoint. I worked through a couple of concrete calculations from Ted's exercises. I feel like I am missing out the big picture still.
what seems especially nice there is that $e^{-N x}$ term in the integrand. if i write down such representations for both, that gives $e^{-N(x+y)}$ in the combined integral, which is readily resummed with $\sum_{n=1}^\infty \frac{1}{n}$
The derivative is the infinitesimal change at the point; how much the function is going up at that point. If you think of this like a vector field, it's how much the function is moving to the right, infintesimally. Integrating across the interval is adding up all those infinitesimal changes, which is the total change.
I am more interested in the first question, about why being swirly is related to $\partial X^1/\partial x_2 = \partial X^2/\partial X_1$. That I want to spend some time on at some point of time.
Which is why in 2D I suggested thinking about curl as being what you get when you draw a small vertical (or whatever) line and seeing how much of the vector field points left, and how much points right.
That's a bit weird, because I was thinking of integrating $\partial X^1/\partial x_2 - \partial X^2/\partial x_1$ as sum of all the "infinitisimal swirls" of the vector field, because yesterday I figured it measures local swirls, not globals.
ok, so we're summing up local swirls of (the vector field corresponding to) $\omega$ which would give me $\int_{\partial S} \omega$ (which, I think, measures how much swirly - globally this time - the vector field is on $\partial S$?)
let me try to understand an example. e.g. take $(0, -x^2)$ - that's like a river flowing. so take a small vertical line near the bank $x = 1$. then the curl measures how much of the current escaping through the right of it than the left?
that makes sense, because that's the force which makes things closer to the bank rotate.
yeah, in my previous case i think the force is the negative of (things flowing the right) - (things flowing the left). yep, we can make the rigorous using the orientation of the curve.
I am afraid it's not intuitively clear to me why the total integral is amount of energy leaving the surface though.
no, i will say swirly to take revenge upon physicists who handwave mathematics. there is no way to handwave physics, so the next best is to abuse physical terms.
In continuum mechanics, the vorticity is a pseudovector field that describes the local spinning motion of a continuum near some point (the tendency of something to rotate ), as would be seen by an observer located at that point and traveling along with the flow.
Conceptually, vorticity could be determined by marking the part of continuum in a small neighborhood of the point in question, and watching their relative displacements as they move along the flow. The vorticity vector would be twice the mean angular velocity vector of those particles relative to their center of mass, oriented according...
@MikeMiller Yeah, I am pretty sure this should be the picture: I take a bowl full of some liquid, I apply some force at each molecule of the liquid so it all starts to turn and twist. Now take some small enough particle and place it at random on the liquid. It will start to move, but focus on how much it "spins around it's own axis", and the observation should be done on an infintisimally small interval of time.
You could have placed this particle anywhere on the bowl and it would have given you some number ("it's angle of spin around it's axis"). Sum these up. Now forget about the interior of the bowl and focus on it's boundary. the liquid moves there, and if you place some particle there and make it not to enter the interior, it'll go round the boundary of the bowl by some angle. I think Green says these are the same.
I don't know how to make the "escaping of energy" thing work.
The reason this works out is, I guess, because when you sum up all those "infinitesimal angle of rotation of a particle around it's axis", many things cancel out.
$\int_{}^{}\int_{R}^{}\frac{sin(x)}{x}dx\,dy$, To solve this integral, with the conditions, Let $R$ be the triangle in the $xy$-plane bounded by the $x$-axis, the line $y = x$,and the line $x = 1$, would the limits of the integral for be from 0 to 1 and for y be from 0 to x?
@Balarka The picture to me is much clearer with divergence in 3D. Maybe I'm wrong that it should work as well in 2D. But it seems strange that there should be a visible difference.
@AkivaWeinberger There was a guy in Ted's youtube lectures who asked what $\partial$ is called when partial derivatives were defined. Another guy googled while Ted was trying to remember, and announced that the official name is "curly dee" in the wikipedia article.
No I didnt write it backwards, I actually saw this technique once from a YouTube video, can you give some some links to easy resources on change of order computation?
@BalarkaSen I had said "would the limits of the integral for be from 0 to 1 and for y be from 0 to x?" which you said to be correct , Now you said "Our integral was $\int_0^1 \int_y^1 \sin(x)/x dx dy$, yep?"
(You should note carefully that the order of the integral just got changed due to setting the independent variable in the limit to be something different. This is the strategy of changing order of integrals)
It's worth noting why Akiva's thing here does not make sense, as Akiva pointed out. You need the outer integral to be independent of any other variables. The variable in the outermost $d(\text{variable})$ should be independent of everything else.
it seems worth mentioning that while there's really nothing nice about the indefinite integral $\int \frac{\sin x}{x}\,dx$, the definite integral over the entire real line is just $\pi$