« first day (4575 days earlier)      last day (740 days later) » 
00:00 - 21:0021:00 - 00:00

00:01
@Seiya I don't understand what books you can or cannot get. This is a paperback book, not a hardback. But it is a fabulous exposition based on vectors and gets to all sorts of beautiful classical stuff, including systems of circles, transformations, projective geometry, the inversive plane, conics and quadrics, and the beginnings of some beautiful classical stuff in algebraic geometry. In particular, they do one of my favorites.
How many lines meet four lines in general position? (Over $\Bbb C$ there is a universal answer; over $\Bbb R$, there is an "at most.")
could that be framed as a combinatoric or counting problem? similar to the 3 colours theorem or whatever it is called.
hi all
0
Q: Density of extended Mersenne numbers?

mickConsider the subset of odd positive integers defined and constructed as follows by these rules : A) $1$ is in the set. B) if $x$ is in the set , then $2x + 1$ is in the set. C) if $x$ and $y$ are in the set then $xy$ is in the set. I call them extended Mersenne numbers because rule A and B alone ...

my question of the day :)
@D.C.theIII It’s an enumerative problem, yes, but an algebraic one, not a topological question.
What's the difference?
Because lines are rigid structures given by specific algebraic equations. A saddle surface has two families of true lines. Very few surfaces can make that claim.
00:09
I will take that for now because there is definitely a lot there for me to unpack. 😄
00:48
@TedShifrin yeah okay that range of topics sounds right up my alley. I'm gonna get it
I've been exploring projective geometry anyway
You know if I ever actually did a PhD in math it could be in some field of geometry. Its easily my most favorite part of math
01:00
what is the more abstract group of SU(2)?
i.e. SU(2) is a group of matrices, but can it be abstracted from being concrete matrices and so on
I just noticed I earned another tag-specific bronze badge, this time for the tag "euclidean geometry"
ugh there is nothing more painful for me than trying to do writing from scratch
writing on MSE? easy, prompt built right in
01:42
I am out of my depth here, but I believe the "big boy" way to think about groups like SU(2) is as Lie groups (manifolds)
$-\lambda^3 + 2\lambda^2 + 3\lambda - 2 = 0$. Any way to solve this for eigenvalues in a reasonable amount of time without resorting to software? $\lambda = \pm 1, \pm 2$ do not work
@ペガサスSeiya Geometry as you know it is not much a part of research. But there definitely are differential geometry and algebraic geometry.
@SillyGoose You can talk about isometries of $\Bbb C^2$ with its standard hermitian metric. This will be $U(2)$. And then the $S$ signifies that they have to have determinant $1$ (which makes sense for linear maps, independent of matrix representation).
@D.C.theIII No. If it is one of my problems, I know that you can find at least one rational root. Are you sure it's really the correct characteristic polynomial?
I can attest that the roots are not nice
At a certain point, I don't see any reason not to use Wolfram or Matlab to give you eigenvalues/eigenvectors. You have to know the mechanics, but then why bother laboring over it if you're not in an actual class?
01:57
I agree, from this being a question in the text I figured they would've wanted me to be able to do it manually to get a feel for what's happening.
You can easily double-check. Is the trace $2$? Is the determinant $-2$?
And I have a shortcut for the $3$, too. I call it "Fred."
The trace of the original matrix is $2$
this is supposed to be the matrix:
I'm double-checking your polynomial with these questions.
$$ \begin{matrix}
1-\lambda & 2 & 2 \\
1 & -\lambda & 2 \\
0 & 1 & 1- \lambda
\end{matrix}
$$
I don't want the $\lambda$s, but no matter.
So Fred should be $3$, yeah. Determinant is $-2$.
02:03
what trace idea did you use? I do recall doing a question showing the trace and it being used in the characteristic polynomial
Yup, it appears you have the right polynomial. Ugh.
who is Fred?....Lol
Fred is $\pm$ the sum of the principal $2\times 2$ minors — one of my linear algebra students named it 20+ years ago, and I stuck with it.
The non-leading coefficient is $\pm$ trace because it will be the sum of the eigenvalues.
The linear (in this case) coefficient will be the sum of the principal $2\times 2$ minors because it will be the sum of the products of pairs of eigenvalues. And determinant is obvious.
All these notions are independent of basis, so hold for any matrix representation of the linear map.
I'd never heard of the principal minors up to this point. But just glancing at the description it is something I could understand now.
Oh, it just means you look at the ii, ij, ji, jj entries. Principal means you use diagonals to determine it.
02:13
can a first order linear diff eq have multiple solutions?
if so, is there a limit
You mean with a given initial value?
Anyhow, you know there is an existence/uniqueness theorem. If its uniqueness hypotheses fail, you may have very infinitely many solutions with a certain initial value.
@TedShifrin Not necessarily, I just read that $\frac{dy}{dx} + P(x)y = f(x)$ has two solutions $y = y_c + y_p$ where $y_c$ is the solution to the homogeneous equation and $y_p$ is the particular solution to the non homogeneous equation
just wondering if there could be multiple unique solutions to the homogeneous equation
What do you mean two solutions?
You can’t say multiple unique!
the diff equation $\frac{dy}{dx} + P(x)y = f(x)$ has the property of having two solutions $y = y_c + y_p$
This is a linear ODE.
That is NOT two solutions!
Pay attention, man!
02:23
WOlfram is a smart engine....those are some atrocious eigenvalues it spit out too
This is linear algebra stuff. Comparing solutions of $Ax=b$ to the homogeneous case $Ax=0$.
okay so the homogeneous case is the trivial $dy/dx = 0$ solution then?
No. Why should it be?
Say the ODE is $y’-y=0$.
Then make the RHS $x$.
instead of $y = y_c + y_p$ so that $\frac{d}{dx}[y_c + y_p] + P(x)[y_c + y_p] = \frac{dy_c}{dx} + P(x)y_c + \frac{dy_p}{dx} + P(x)y_p = f(x)$ where $\frac{dy_c}{dx} + P(x)y_c = 0$ is the solution to the homogeneous form,
what if we had $y = y_c + y_p + y_z + ...$ would that make sense
aye, but I was trying to construct my unitary matrix $Q$ from all the eigenvalue stuff. Just saw wolfram has an eigenvalue calculator too...
02:31
Obliv, you are writing a lot of nonsense. A first-order linear equation has infinitely many solutions . When you fix an initial value, there is a unique solution.
I'm not quite sure what the text was expecting in those calculations becasue even the eigenvectors are atrocious.
This is why authors should write an answer manual, whether it’s available or not, before publishing the book.
@TedShifrin so when the book says the solution to a first order ode is the sum of two solutions where one is the solution to the homogeneous case, it's talking about infinitely many solutions
Or proofread more carefully.
That is not the sum of two solutions of the given ODE. Did you read my comment about the linear algebra? You should know this from there.
i.imgur.com/TEUPVGC.png I can't tell if I'm misquoting at this point
02:35
Well I could play around and read up on QR factorization now since I at least understand it.
I think I get what you mean, it's moot at this point I don't think there was a need to confuse myself
Obliv … It is badly written. They are adding a solution of the homogeneous (a DIFFERENT equation) to a particular solution of the given equation.
I got confused because I thought the book was saying there were only two solutions for any given linear first order ode
it is an extremely introductory text I think, it also just slaps on integration factors without explaining the motivation or where it comes from
Think about solutions of $x+y=1$ in terms of solutions of $x+y=0$.
(waiting for Ted to mention he has a book on Diff Eq)
02:40
Introductory is no problem, but the way they wrote this is what led to your misunderstanding!
@TedShifrin isn't that your research area as well?
I do have one question though. If I know $A$ is unitarily equivalent to some diagonal matrix, why even bother with getting $R$ in the $QR$ decomposition if I can just go: $Ax = b \rightarrow Q^*DQx = b \rightarrow x = Q^*D^{-1}Qb$?
@ペガサスSeiya Yes, in my former life. But I’ve never done Euclidean geometry with your passion and trickery. It just doesn’t interest me.
dc3: why bother with any of it? everything depends on context. depending on context, matrix factorizations may be a distraction from whatever you are hoping to get from considering Ax = b. if you are solving the equation you may also care about numerical issues, or (both in numerical and non-numerical contexts) what things about A that you have already computed or had given to you, versus things about A that you do not have given and need to compute.
QR is totally general.
02:44
Nvm I see, $x = -y$ which when added to $x = 1 - y + (-y)$ is a solution. So the homogeneous case of any linear ODE can be added to the solution like that?
The $Q$ has nothing to do with eigenvectors.
e.g. do you know that A is diagonalizable. do you have the eigenvalues or eigenvectors. [and as ted has just pointed out, the matrices "Q" in your two factorizations are generally not the same]
@TedShifrin there must be something wrong with me (or my brain).
Obliv, a particular solution is $(1,0)$. Now add any vector on the line through the origin, yes.
@Seiya Nothing wrong with enjoying geometry, it's the tangible visual aspect of math I really enjoy the beauty of it too
02:47
I wasn’t saying you’re bad. I’m saying your Euclidesn geometry prowess isn’t needed for advanced geometry.
I still think geometrically, unlike leslie . :)
I like thinking geometrically too. Rotating and manipulating shapes in my brain. Its fun
The downside to that is sometimes I don't watch where I'm going. My eyes are open but I'm basically blind at that moment
Did you recognize my avatar?
@TedShifrin you mentioned its name but I forgot. It looks like the top of a circus place or something
@leslietownes THings for me to ponder.........it isn't lost on me that you threw in the word "context" to capture all the happenings of the day in one phrase....I see you....👀
Obliv, you understand that they’re saying the sane thing about ODE?
@seiya It’s what you get when you rotate a common object.
02:51
Yes, this is very deep stuff though. Not because it's any different but infinitesimals make stuff more deep
@ペガサスSeiya consumed by your passion, not a bad thing. 🙂
@TedShifrin which common object?
You tell me!
@D.C.theIII well true but, when you constantly walk into doors or walls then its not so fun
dc3: linear algebra is a particularly tricky thing to be introduced to, because some of the notions you're introduced to are fundamental to a lot of things, or the theorems you see are in some sense 'the final word' in a given setting. and then other stuff is somewhat case-specific, or involves arbitrary choices, or can be theoretically easy but computationally difficult in a way that can't be taught in an introduction. and whatever the status of these notions, they arrive all at once.
which maybe gives the impression that they're all of equal importance/applicability, and are all in simultaneous use all the time.
you could have a one-semester class entirely devoted to matrix factorizations.
02:54
@Obliv Not really deep. It is just linear algebra in a different setting.
although in any given context you may not want/need to use one, or only one of a wide variety of choices has any hope of helping you.
@leslietownes Strang endorses that!
he's cornered the market on that.
@TedShifrin I have a shape in mind (probably not the right one) but I don't even know its name. From what I can tell, this shape is protruding sideways at the top and bottom but slightly "empty" around the midsection, if that made any sense
Nope. It’s a common shape you know well.
03:00
How do we prove formally that the cone on a closed manifold is a manifold with boundary iff that closed manifold is a sphere
Ok. Done for the day, time for steak, veggies and watch Match of The Day.
robjohn has such a beautiful animation about that rotating shape. don't click through if you don't want spoilers. math.stackexchange.com/questions/115743/…
@leslietownes That was neat
@TedShifrin a cone?
He had the name of the shape in his bio. I'm not sure if it is there anymore
03:07
I don't want spoilers that's why I'm not looking anything up
nvm
Dang you can't undelete
gonna repost how does $\frac{d}{dx}[e^{\int P(x)dx}y] \to e^{\int P(x)dx}\frac{dy}{dx} + P(x)e^{\int P(x)dx}y$
the last term doesn't make sense to me shouldn't it be derviative of $e^{\int P(x)dx}$ be $e^{\int P(x) dx}$ so the term should be $\int P(x)dx e^{\int P(x) dx}y$
oh wait It's a chain rule isnt it. I just know d/dx e^x is e^x but its really 1*e^x huh
nvm
03:24
Yeah, same as how $\frac d{dx}e^{x^2}=2xe^{x^2}$
03:42
Btw I slandered the textbook, it totally goes over integration factor, I misread.
04:33
How to find the shortest distance between $\frac{x-1}{2} = \frac{y+1}{3} = z$ and $\frac{x+1}{5} = \frac{y-2}{1}; z = 2$?
I know there is a formula... but $\frac{x+1}{5} = \frac{y-2}{1}; z = 2$ this equation is a little bit different. $z = 2$ is given separately. How to convert this in the form $\frac{x-a}{A} = \frac{y-b}{B} = \frac{z-c}{C}$?
You cannot. $C=0$, but you need to know which plane $z=c$ the line is in.
xy plane?
No. They told you $z=2$.
What is the direction vector of second line? (5i + j + 0k) maybe?
04:47
oh okay... Now I can continue. Thanks a lot!
05:17
What is opposite category?
something you can google? :)
my confusion is because of contravariant functor: is it always a functor? Because by definition, it reverses the composition which a functor by definition doesn't do.
Leslie: I did. But I want some example on that. Here is what I understood from wiki about opposite category: Suppose that C is a category. Then C* is the opposite category (dual) if obj C*=obj C and if Hom (a,b) is a set in C, then Hom(b,a) is in C*.
"is this a functor" is not enough context, you need to specify the category
So take two categories C and D. F: C-->D is a functor.
a "contravariant functor" from a category C to a category D is not a "functor" from C to D
or maybe is, but only in weird degenerate cases
05:24
@leslietownes yes :). Thank you so much!
That's exactly what I also thought.
thanks for confirming that.
"Functors as just defined are also called covariant functors to distinguish
them from contravariant functors that reverse the direction of arrows."
what is 'reverse the direction of arrows'?
some category theory folks like being as general as possible, and love realizing things as examples of other things
Hom (a,b) becomes Hom (b,a)? Is that what is being referred to there?
so you can build notions of "covariant functor" and "contravariant functor" from a single concept of [covariant] "functor," if you don't mind changing a category
people really get off on this and i don't know why
koro, side note, who is making you study categories and why are they evil
what turned them from the righteous path
I'm trying to understand algebraic topology from Rotman's book. In the book, there is a proof of Brouwer's fixed point theorem using categories in the first chapter and then there is some intro to categories and functors. I'm trying to understand that.
They use it later in homotopy theory.
If you have the category of vector spaces, then $\text{Hom}(\cdot, Z)$ is a functor and so is $\text{Hom}(Z,\cdot)$. Which is covariant and which is contravariant?
05:33
something something temperatures at two points on the globe are exactly the same....
apart from that, category, functors were also speedrun in our functional analysis class. I don't understand why though.
Leslie: We have Bessel's inequality in Hilbert spaces- $\sum_{i=1}^\infty|\langle x,e_i\rangle |^2\le \|x\|^2$. In my class, I think it was generalized to any sum on LHS (not just a countable sum).
it was done using nets. I don't yet know the details of the proof. I didn't understand it at that time.
@TedShifrin the later one is covariant.
(as per the definition that I have studied.)
this is true, although maybe not as surprising as it first sounds. for any x in a hilbert space, it will have at most countably many nonzero components with respect to an orthonormal basis, regardless of the cardinality of the basis.
Yup. Right.
If a set $C$ is a proper closed subset of an open set $U$ then will $C$ be closed in cl($U$)? Everything is in $\mathbb R^n$.
Homology is covariant, cohomology is contra.
Closed in $U$ or closed in $\Bbb R^n$?
05:39
@TedShifrin We are given that C is close in U.
I think the statement is false.
Example?
Like $U=(0,1), C=(0,\frac12]$
Good. Or $C=U$ :)
proper
05:41
:(
I have one more confusion. This is regarding congruence on a category.
Never heard of such.
I still don't get it why we can choose such $y$.
I thought that the argument is $C,\partial \Omega$ are compact and disjoint so they have positive distance in between them so we can choose such $y$.
A congruence on a category $\mathscr C$ is an equivalence relation ~ on the class $\cup_{(A, B)}$ Hom(A, B) of all morphisms in such that: 1) f in Hom (A,B) and f~f' implies that f' is also in Hom(A,B), 2) if f~f', g~g', then gof~g'of'.
But $C$ may not be closed in cl($\Omega$).
@leslietownes the only motivation i've found for category theory is so i can halfway understand why the diagrams i like making instead of writing equations work
05:49
I don’t think that’s it. You don’t need distance between $C$ and $\partial\Omega$. Why are you doing that?
Usually, an equivalence relation is defined on a set, but here we do not have a set in general, rather a class. Fine, acceptable for now. ~ partitions $\cup_{(A,B)}$ Hom (A,B) into equivalence classes.
semi: yeah, its very helpful as a language for simplifying the presentation of high level stuff. and for concisely summarizing lots of details at once.
my problem with is that once people step into the world of category theory, you can also immediately go down every rabbit hole in existence, without useful examples by your side, or really any notion of what concepts are being abstracted from.
the web, in particular, is really bad at this.
Then the book defines an another category $\mathscr C'$: obj $\mathscr C'=$ obj $\mathscr C$, Hom$_{\mathscr C'}(A,B):=\{[f]: f\in Hom_{\mathscr C}(A,B)\}$
so my sympathies are with koro and i hope his instructor turns away from the dark side.
05:53
to blatantly steal a philosopher's metaphor: a bird might think it could get through the sky all the more easily if there wasn't any air to resist it...forgetting that without the air there's nothing for the bird to push against
So how can these two categories (clearly visually so different) have the same class of objects? Why is obj $\mathscr C'$= obj $\mathscr C$?
ahh, I think I understand... It doesn't matter if the objects are the same or not.
By definition: a category C has three ingredients- 1) class of objects obj C, 2) sets Hom (A,B) , 3) composition and some more properties.
@TedShifrin Then How to prove that?
The question was: Give an action of $Z_2$ on a torus so that the orbit space is a cylinder.
I think I did it correctly. But still, I lost some marks there.
apparently there was some objection to 'rotating about an axis' as not being a mathematical term or something like that.
was I supposed to write the explicit maps? (that is, for the last homeomorphism as well?)
If yes, then how would one do that using explicit maps?
rotating around the first axis would be $(x,y)\mapsto (x+X,y)$ presumably
06:08
hard to know for sure, i'm guessing they wanted more information about why the quotient was a cylinder. i don't think it's necessarily about the specific realization you used, or a lack of explicit maps.
@PNDas Maybe think about points in $U$ far away from the boundary. Then approach $C$.
So I take a circle in xy plane, centered at (3,0,0) with radius 1, rotate it around y-axis to get a Torus. Then I define an action of Z_2 on the torus T, that reflects about yz plane.
This indeed identifies the whole Torus with its left half (or right half).
Actually, I saw a similar example in Armstrong's also.
Cutting the torus in half gives a cylinder up to homeo. That sounds fine.
I talked to the teacher. He agreed eventually that it is correct. It was some misunderstanding due to 'rotate about' terminology.
He also suggested to use maps on $S^1\times S^1$ to show that. But I don't understand yet how to do that.
Use two angular coordinates to parametrize the torus. Maybe use $[-\pi,\pi)$.
06:20
ohh
theres an art to giving examples in topology, where you want the example to have enough structure so that you can talk about it clearly, but maybe not so much structure that when you discuss it, you're spending a lot of time in particulars about its realization that do not directly affect the topology.
@Koro, good question for your abstract algebra knowhow:
https://math.stackexchange.com/questions/4637533/what-is-the-minimum-a-b-such-that-for-any-such-a-b-in-bbbz-we-have
involves linear independence and prime numbers
@DLeftAdjointtoU I'll take a look at that. Thanks :).
Thanks @Koro
You da man
:D
Please ask here if you have a question
I can show you what I mean - it's not too complicated, but it is fresh math area that hasn't been talked much about
Applying linear independence to arithmetic functions ie.
$(d\mid \cdot)$'s actually form a basis of all arithmetic functions, you can prove that using mobius inversion formula
So you have the usual Dirac delta basis like always, but these divisor functions (whether or not $d$ divides the argument, are also a basis
A schauder basis i.e. not Hamel
unles all your arithmetic functions have finite support
Well I'm talking about a finite length vector of values of those divisor functions
in that post
There's also a homological chain complex involved (see related post close to this one)
But I haven't found a use for it yet
@Koro I can tell you're having trouble with it
What's the first stumbling block?
Maybe I can help
07:05
@DLeftAdjointtoU Sorry, I was away. I have seen the messages now.
I have not yet seen the post in detail though. I'll look at it but not right now.
I see that the post is deleted now. Perhaps, you figured out the answer already. :-).
Hey @Koro what do you study these days?
07:25
@onepotatotwopotato algebraic topology, functional analysis
etc.
It is true that in a normed linear space X, a cauchy sequence converges to at most one point.
But I seem to have found an NLS, where this is not true: Suppose that $Y$ is a complete subspace of a normed linear space X. I can form the quotient X/Y. X/Y is an NLS with the norm $\|[x]\|=\inf_y \|x+y\|$.
Suppose that X/Y is complete. Now, I claim that X is complete, and that a Cauchy sequence in X converges to more than one point.
Take any cauchy sequence $(x_n)$ in X. Given $\epsilon>0$, there exists N in $\mathbb N$ such that for all n, m>N, $\|[x_n]-[x_m]\|=\|[x_m-x_n]\|=\inf_y\|x_m-x_n+y\|\le \|x_m-x_n\|\lt \epsilon/2$
So $([x_n])$ is a Cauchy sequence in X/Y, which is complete, so there exists $[x]\in X/Y$ such that $[x_n]\to [x]$.
Now, I claim that $x_n\to x'$, where $x'\in [x]$. The proof is as follows: $\|x_n-x'\|=\|x_n-x'+y-y\|\le \|x_n-x'+y\|+\|y\|$. Taking infimum on both sides: $\|x_n-x'\|\le \inf_y \|x_n-x'+y\|=\|[x_n]-[x']\|=\|[x_n]-[x]\|\to 0$ as n tends to $\infty$.
So $\lim x_n =x'$ for any $y\in Y, x':=x+y$.
Why is this happening?
@onepotatotwopotato what have you been studying these days?
if $f$ and $g$ are scalar valued functions on $Y$, are you implicitly using the (generally false) "identity" $\inf_{y \in Y} (f(y) + g(y)) = \inf_{y \in Y} f(y) + \inf_{y \in Y} g(y)$?
in "taking infimum on both sides"
or the same with $\leq$
triangular inequality, assuming f and g are positive valued.
i.e., $\inf_y(f(y)+g(y))\le \inf_y f(y)+\inf_y g(y)$, where f, g are non negative functions.
(must be a serious matter that Leslie had to write mathjax. :D)
I should be more careful around this inequality I guess.
07:43
so like $\inf_{x \in \mathbb{R}} (|x-2| + |x|)$ is not zero, while $\inf_{x \in \mathbb{R}} |x - 2| = \inf_{x \in \mathbb{R}} |x| = 0$
i would not generally infer anything from my switching into or out of chatjax but in this case i think it does help to think about what the inf is infing over and maybe my usual text silliness isn't the best for that
Alas, the inequality is wrong.
the correct inequality is other way round.
thanks Leslie. I understood my mistake.
@Koro Analysis and algebra mainly. Maybe I would study functional analysis this summer.
 
1 hour later…
09:11
@TedShifrin Happy birthday! 🥳🎂🥳 thank you for all your time, patience, and room ownership duties here. We really appreciate your help.
 
1 hour later…
10:32
is 'provided' same as 'if and only if'?
statement 1 is true provided statement 2 is true. = statement 1 is true iff statement 2 is true ?
 
1 hour later…
11:42
2
Q: Is it necessary to use the Hahn-Banach theorem to show that $(X/M)^*\simeq M^\perp$?

user122916Let $X$ be a Banach space with dual space $X^*$, and let $M$ be a closed subspace of $X$. Then $M^\perp=\{x^*\in X^*: x*(m)=0 \text{ for all } m\in M\}$ is a closed subspace in $X^*$. I read the following statement in Garnett's Bounded Analytic Functions: The Hahn-Banach theorem gives us the ...

How to show here that T preserves norms ?
11:56
slightly not so important question, but why is integration wrt to $u$ when we can do wrt $t$?
12:07
4
Q: Isometries between dual factor space and annihilator

Marios GretsasLet $X$ be a normed space and $Y$ a closed subspace of $X$ and $Y^0=\{f \in X^*|f(x)=0,\forall y \in Y\}$. Prove that $Y^0$ is isometricaly isomorphic with $(X/Y)^*$. I have to find a function specifically between $Y^0 $ and $(X/Y)^*$. One idea is the function $S: Y^0 \longrightarrow (X/Y)^*$...

How does this answer the question?
12:28
It is not true that norm 1 maps are isometries so how does the post answer the question?
12:51
quite some activity here :p
3
Q: Density of extended Mersenne numbers?

mickConsider the subset of odd positive integers defined and constructed as follows by these rules : A) $1$ is in the set. B) if $x$ is in the set , then $2x + 1$ is in the set. C) if $x$ and $y$ are in the set then $xy$ is in the set. I call them extended Mersenne numbers because rule A and B alone ...

13:19
Can anyone please help me with this math.stackexchange.com/questions/4637693/… ?
@Franklin did you see my answer to the question about the subgroups of $\Bbb{Z}/21\Bbb{Z}$
13:38
@SineoftheTime Yes! I saw it. I also got a simpler understanding. Nevertheless it was a great help from you. Accepting your answer in a jiffy!
14:11
7
A: The dual of subspace of a normed space is a quotient of dual: $X' / U^\perp \cong U'$

user127096It is mostly correct now, but could be written better. Consider $l:X'/U^\perp\to U'$, which sends each coset of $U^\perp$ to its restriction to $U$. This map is well-defined because all elements of the coset agree on $U$. The map is surjective, because every functional $f$ on $U$ can be exten...

What does this post mean by l does not decrease the norm. It seems false.
 
1 hour later…
15:21
Now, I am having problem with this: math.stackexchange.com/questions/4624141/…. Can anyone please help me?
P.S please ignore my previous comment as I have understood it now more or less!
(I meant this:"Can anyone please help me with this math.stackexchange.com/… ? " which is to be ignored. As I can't delete it now)
 
2 hours later…
17:21
0
Q: Dual of a closed subspace of a Banach space is isometric with a quotient space

KoroThe dual of subspace of a normed space is a quotient of dual: $X' / U^\perp \cong U'$ I am trying to prove that $X' / U^\perp \cong U'$, for $U$ being a closed subspace of the Banach space $X$ in the following way: Define $T: U'\to X'/U^\perp: T(f)=[\tilde f]$, where $\tilde f$ is Hahn Banach ext...

I am surprised to discover that Hahn Banach extension theorem doesn't talk about uniqueness even upto isometry.
 
1 hour later…
18:28
I see a heart-shaped identicon floating around here. Hmmm, who might that be, @robjohn ?
does anyone have recommendations on resources for applied graph theory 0.0
Anyone see any orange hearts below?
:D
^^^ @robjohn Can you pull off the one above?
@amWhy: they look like candies...
koro: for a general subspace of a general normed space there is no uniqueness of the extension. you need to assume more about the normed space (or at least the relation between the space and the subspace) to get that.
@Koro Well, despite the mean square and mean orange heart, deep down, @robjohn is a sweet teddy bear!
18:44
Suppose that E is reflexive Banach space. I know that for all x* in E' (the dual of E), there exists x in E, x*(x)=$\|$x*$\|$.
But if I add one more condition: $x\in E$ is such that $\|x\|=1$, then can I get this result?
It is true for x*=0 map. So suppose that x* is non zero. Then consider the subspace W= span (x*). Define f($\lambda $x*)= $\lambda f(x$*). Extend it to $\tilde f$ by HB.
f(x*)=d, say. There exists x in E such that the canonical map $j_E$: $j_E(x)=\tilde f$.
So x*(x)= $f(x$*)=d.
How do I get: x: $\|x\|=1$?
Hi.... Is there any site like coursera that offers algebraic geometry course with certificate? Or topics related to it...
Are there an infinite amount of definable unique operations?
koro: er, have you even defined f on W? what is f(x*)? could you make some simplifying assumptions (e.g. that ||x*|| = 1)? you will need to use that the HB extension has the same norm as the thing you are extending.
this is also on MSE if you get tired of thinking about it.
hmm, I thought of defining that later. f(x*)=||x*||.
I haven't taken abstract algebra, does there exist operations that aren't just forms of + or -
or like increasing/decreasing the ordinal rank of an element in a set I guess
18:59
@Obliv: yes. composition of functions for example.
Leslie: yes, I should also use the fact that the canonical map is an isometry.
given y as a bijection in R, does $\Delta y$ mean the difference between two values in y's image?
would y be the function whos independent variable is elements of its domain $x \in \mathbb{R}$
I almost forgot how good a full 7-8 hour sleep feels
@ペガサスSeiya have you watched Boku no hero academia?
@Koro yeah I've been watching it
Season 6 soon
19:15
I watched season 6 latest episode the other day. I skipped second half of season 5. :)
@Koro I feel like its coming to an end soon after season 6
I don't see where they can go after Shigaraki
have you also noticed that the laser guy (Yuga Aoyama) seems to be missing in the show?
19:37
just making sure, $x + y = 4$ can be $y = 4 - x$ y as a function of x, or $x = 4 - y$ x as a function of y
morning folks!
Hi @Koro
Hi @copper.hat!!
beautiful day in the bay area today, the super bowl (big sporting event) is on so things are quiet here.
20:05
Did the bike ride this morning?
just about to head to the pool, went for my little bike ride yesterday :-)
A little active recovery from yesterday's grueling ride
aujord' hui est tu anniversiare @TedShifrin?
Dump the lame French, but yes :)
ne peux pas practique?..................alors...............Happy Birthday. Just hit 30 right?
@nafisemodaresi I know nothing like this for advanced mathematics.
Your Google Translate is way sub-par. Yeah, 30. Right.
20:15
@D.C.theIII fortunately/unfortunately not a grueling ride. albeit as i age, any ride starts to become so :-)
Oh it wasn't google translate. I was actually writing it out. I do speak two other romance languages and did take a year or two of French
my daughter is 22 today, scary how time flies
next thing will be walking her down the aisle........scary thought
i walked my sister down the aisle, that's enough :-)
@Ted have you come across this miraculous answer before?
20:42
Yes, indeed. I’ve thought about the cocycle approach and a partition of unity standard proof. Either way, thecset must be closed.
where he does pull those cocycles from? it's kind of mystifies me at the moment, not sure if I'm overlooking something
20:56
@TedShifrin last night when you computed the principal minor of this matrix you said you got Fred to be $3$, I just computed it but got $-3$ what did I miss to do? $$\begin{matrix} 1 & 2 & 2 \\ 1 & 0 & 2 \\ 0 & 1 & 1 \end{matrix}$$
little known fact, archimedes actually wrote 'the sand reckoner' to work out a method for determining ted's age in years
Leslie could answer this as well...
Fred is ± the sum of the principal 2×2 minors
question: If we have a "full cylinder", namely $D^2 \times I$ and we remove a point from the boundary, the fundamental group is trivial right?
00:00 - 21:0021:00 - 00:00

« first day (4575 days earlier)      last day (740 days later) »