« first day (2993 days earlier)      last day (2325 days later) » 

00:06
@kas!
@LeakyNun Have you tried thinking of it as an infinite product?
@KarlKronenfeld yes, I have
to no avail
00:18
Is ln((e^x-1)+1) any easier? There's a trick for proving the desired equality from this.
Are there good individual books(lots of theorems and solutions to problem exercises) for each of these Discrete Mathematics topics like

1)Logic and propositions
2)Relations,functions,cardinality and sets
3)Graph Theory
4)Counting and probability
I'd give that a try @LeakyNun, it looks like inclusion-exclusion somehow.
00:36
@KarlKronenfeld hmm, I still can't do it
01:14
@PrashinJeevaganth A Concise Introduction to Logic by Patrick J. Hurley is the book I'm using. For logic I mean.
@CaptainAmerica16 Noted, thanks :)
01:26
@LeakyNun what if you did something like taking its taylor expansion at 0 and show that for f(x) = exp(log(1+x)), the higher derivatives f^{(n)}(0) = 0 for n \ge 2?
 
2 hours later…
Pig
Pig
03:09
@LeakyNun Cant you just differentiate? Let $f(x) = exp(ln(1+x))$, and use chain rule to see that $f'(x) =f(x) \frac{1}{1+x}$. From this deduce that $f(x) = 1+x$ (all the above statements live in the ring of convergent power series)
03:22
He's working in p-adic power series, so I guess you want f' to be meant in a formal sense
how are the transcendental numbers distributed
nvm
03:38
@MikeMiller Now you have me proving the chain rule for derivatives of formal power series. :)
Can anyone suggest good book for self study Differential Geometry? I also want to know the prerequisites. What about the book written by John A Thrope?
@MikeMiller You can probably write the elliptic curve as a section of O(3), in which case you don't have to do that much work anymore
I parsed that iffily. I meant "I think one can write the elliptic curve in P^2 as zero locus of a section of O(3)"
Thats certainly true. Sure, I see your point.
vzn
vzn
03:56
@BalarkaSen o_O back from the dead!™ how is university treating you? (somewhat afraid to ask lol) :P
vzn
vzn
Sep 26 at 22:04, by anakhro
Clearly he's in some dean's basement in chains.
@BalarkaSen There was a great question the other day with an elegant answer, one sec
@BalarkaSen Hi, can you answer this question ?
4
Q: Immersions of vector bundles

JHFLet $E$ be a vector bundle of rank $k$ over a smooth manifold $M$ of dimension $m$. Then $E$ can be thought of as a $(m+k)$-dimensional manifold, so the (weak) Whitney immersion theorem states that $E$ can be immersed in $\mathbb{R}^{2(m+k)}$. Is it possible to immerse $E$ in $\mathbb{R}^{...

03:59
@taritgoswami Try do Carmo
I don't know any other books, or haven't read them thoroughly enough to recommend
@BalarkaSen Ok, is that an author name?
OK, in which Univ you are in ?
If you don't mind :-)
vzn
vzn
@BalarkaSen any interest in ? cant remember if we ever discussed it. tarit is an undergrad student and says his university has no program. wanted to introduce you 2.
@BalarkaSen I am also from India
04:01
I don't know number theory
@tarit ISI
@MikeMiller Very cool question
I am really glad OP found a proof
@BalarkaSen I just missed the chance to get into B.Math degree only for 3 marks :-( , I have covered elementary number theory for Entrance exam, your are in M.Math then?
And an elegant one, too - it's just arguing that the map on cohomology is an iso (plus fundamental group considerations, but whatever)
vzn
vzn
04:04
@BalarkaSen tarits interested in analytic number theory & suspect it might have connections to stuff youre interested in, dont know exactly...
The proposition is very strange, I have never encountered that before
But it makes sense. If $\varphi_{ij} : U_{ij} \to O(k)$ are the transition functions of $E$, then you should be able to extend that to the transition functions $\psi_{ij} : U_{ij} \to O(k+m)$ of $G$ by some block matrix shit
Yeah exactly.
Well, I guess preciselt what we are saying is that G's transition functions can be reduced to blocks
Right, there's no homotopical obstruction in that. Cool!
Love that
Well, there's something subtle - the dimension of the manifold has to matter
@BalarkaSen Do you plan to read it anytime soon? It would be good to bounce questions off of someone else learning it.
04:11
please help me solve $\ln(1/2)=\ln(x)-1/\ln(x)$
@MikeMiller Good point. It seems you can lift because of cellular approximation, and that $M$ has no higher than $m$-cells
it simplifies to $-\ln(2)=\ln(x)-1/\ln(x)$
@Symposium Unsure. I'm doing too many things at once and Galois theory is in the list of things I want to read. I have been looking into bits and pieces off of it because my friend is reading it as well, but I can't promise anything
then I tried factoring the right hand side but it didn't seem to help
another question: is $1/e^{\sqrt{2}}$ a transcendental number?
vzn
vzn
any RH fans in here? just ran across a neat ref, wanna share it with anyone who might actually look at it :)
04:18
I will look at it with probability .75
vzn
vzn
:) ok good enough for me. warning it talks about fractals + collatz + empirical work also =D
in Number theory, 27 mins ago, by vzn
@taritgoswami ok just ran across this ref mentioned on this RJL blog page, see my comment, its very neat, @MatsGranvik think you will like it also :) it relates Riemann Hypothesis and Collatz wrt fractals + experimental work (my fave!) + much else Experimental Observations on the Riemann Hypothesis, and the Collatz Conjecture / Chris King https://rjlipton.wordpress.com/2018/10/10/watching-over-the-zeroes/ https://rjlipton.wordpress.com/2018/10/10/watching-over-the-zeroes/#comment-9512‌​0
is $e^{\sqrt{n}}$
transcendental for $n=1,2,3,4,5,6,7,8,9,10,11,12$
what I know: $e$ is transcedenet and $\sqrt{2}$ is irrational but not trancvedantal
so putting them together in such a way that exponentiantion comes into play with the transedental part as the base and the square root part in the exponent, not sure what the result should be
05:00
answer: $ e^{\sqrt{n}} $ for $n\in \Bbb Z^+$ is a transcendental number
0
Q: Elements in the basis of Product topology determined by sub-basis other than sub-basis elements.

Unknown x I could prove the result for $|\Lambda|$ finite. Here $|\Lambda|$ is arbitrary. My attempt:- Let $\langle x_{\alpha}\rangle_{\alpha\in \Lambda}\in B \implies \langle x_{\alpha}\rangle_{\alpha\in \Lambda}\in \pi_{\beta_i}^{-1}(U_{\beta_i}), \forall i=1,2,3,...,n. $ So, $\pi_{\beta_i}(\langle x...

Can you verify my proof?
@BalarkaSen fair enough.
Are there good individual books(lots of theorems and solutions to problem exercises) for each of these Discrete Mathematics topics like


1)Relations,functions,cardinality and sets
2)Graph Theory
3)Counting and probability
05:43
While working on Heisenberg-Weyl operators (specifically working on their eigenvalues), I have come across the following problem:
Given an $n$, $n$th roots of some complex number $p, |p| = 1$ sum to zero. Let $A$ be the set that contains all these $n$ roots. My intuition was if we form a subset of $A$ whose elements also sum to zero, then this subset is necessarily a complete set of $m$th roots of some $p'$, where $m$ is a factor of $n$. However, this intuition turned out to be wrong
see: https://math.stackexchange.com/questions/490115/summation-of-roots-of-unity-equal-to-zero?rq=1
For example, for a prime $n$, can we show that there does not exist a subset of $A$ whose elements sum to zero?
 
2 hours later…
07:42
hi @loch
Hi @LeakyNun
@loch you did elliptic curves right?
Did someone say elliptic curves?
Not at imperial but yes
07:45
@Daminark unfortunately the first half of the course is about p-adics
(I do not have "aphantasia" but the concept is surreal)
I should learn more about p-adics tbh
@loch I'm being asked to prove exp(ln(1+x)) = 1+x. I don't really know how to do it. Even Ted said it's hard.
Also I guess we're all a little aphantasiac when it comes to higher dimensions
@loch for x in pZp for p>2, that is
07:46
@LeakyNun The series?
What's the series for $\ln(1+x)$?
x-x^2/2+x^3/3-x^4/4+...
i thought I mentioned a strategy earlier hmm
where you can look at its taylor expansion at 0
I didn't do the computation of course
we are looking at its taylor expansion at 0
07:47
as in
$\sum(-\sum(-x)^n/n)^m/m!$ I guess?
@AkivaWeinberger right
And then binomial theorem
differentiate f(x) = exp(log(1+x))
show f^{(n)} (0) = 0 for n \ge 2
Well hold on actually
Let's go a bit more general
What's the $x^c$ term of $(a_0+a_1x+a_2x^2+\dots)^n$?
07:49
something multinomial
@AkivaWeinberger Interesting. I think I'm on the extreme opposite end of this.
@loch not sure if that is harder or easier to show...
Sum of $a_ia_j\dots a_k$ where there are $n$ of them and they add up to $c$?
@AkivaWeinberger right
Counting different orders as separate
So here it's $a_n=1/n$
and let's do $\exp(\ln(1-x))$ just so we don't have to deal with the minus sign
and it'll be the same anyway
07:51
@loch I think I can show that f^(2) (x) = 0
thanks
alternatively i think what @pig mentioend earlier also straight up works, use the fact that f'(x) = f(x) 1/1+x,
get (1+x) f'(x) = f(x)

write f(x) = \sum a_i x^i, then f'(x) = \sum ia_i x^{i-1}, then compare coefficients
Hm OK I'll actually want paper for this
And I realize if I do $\exp(-\ln(1-x))$ by accident (eg by ignoring too many minus signs) I'll get $1+x+x^2+\dotsb$ instead
What I should do is compute the coefficient for $x^3$ by hand
@AkivaWeinberger stop throwing users down the FB rabbit hole :P
I think applying multiplication k times works too, but it looks laborious.
07:56
$\frac13-\frac12+\frac16=0$, yeah?
So that's$$\frac{\frac13}{1!}-\frac{\frac11\frac12+ \frac12\frac11}{2!}+\frac{\frac11\frac11\frac11 }{3!}$$
2
Conjecture:$$\frac{\frac14}{1!}-\frac{\frac11\frac13 +\frac12\frac12+ \frac13\frac11}{2!}+\frac{\frac11\frac11\frac12+ \frac11\frac12\frac11 +\frac12\frac11\frac11 }{3!}-\frac{\frac11\frac11\frac11 \frac11}{4!}=0$$
$\frac14-\frac{11}{24}+\frac14-\frac1{24}$ does in fact equal 0
Right OK so that's what $\exp(\ln(1+x))=1+x$ means
It should be zero for all $n\ge2$ (and one for $n=0,1$)
So is there a simple way to prove that
@LeakyNun
@AkivaWeinberger we just discussed it above
?
How to prove that it always equals zero?
15 mins ago, by Leaky Nun
@loch I think I can show that f^(2) (x) = 0
15 mins ago, by loch
alternatively i think what @pig mentioend earlier also straight up works, use the fact that f'(x) = f(x) 1/1+x,
get (1+x) f'(x) = f(x)

write f(x) = \sum a_i x^i, then f'(x) = \sum ia_i x^{i-1}, then compare coefficients
But is there a simple way to show that the monster equation I wrote always equals zero, I mean
Directly
I don't think so
08:09
leave monsters alone
It's such a ridiculous expression isn't it
I like that the number of terms in the numerators follow Pascal's triangle
('cause the number of ways of getting $n$ positive integers integers to add to $m$ is $\binom{m-1}{n-1}$)
(counting rearrangements as separate)
(Proof: stars and bars)
Here's a great trophy for whoever can do it :-)
08:51
That looks neat! @user1732
I think there's a formula for $X^k$ where $X$ is a power series. But I'm too tired to find it.
A formula that isn't iterations of Cauchy of products, that's.
09:18
@user1732 I must have that trophy. It's Ockham's razor! What's the question?
1 hour ago, by Akiva Weinberger
But is there a simple way to show that the monster equation I wrote always equals zero, I mean
Any comments on this?

While working on Heisenberg-Weyl operators (specifically working on their eigenvalues), I have come across the following problem:
Given an $n$, $n$th roots of some complex number $p$,$|p|=1$ sum to zero. Let $A$ be the set that contains all these $n$ roots. My intuition was if we form a subset of $A$ whose elements also sum to zero, then this subset is necessarily a complete set of $m$th roots of some $p′$, where $m$ is a factor of $n$. However, this intuition turned out to be wrong
1
Q: Summation of roots of unity equal to zero

Taekyo LeeLet $u$ be a integer which has an odd prime $m$ as a divisor. So we let $u=u_1m$ where $u_1=\frac{u}{m}$. Consider the set of $u$th roots of unity, i.e., $A=\{e^{j\frac{2\pi}{u}n} | 0 \leq n \leq u-1\}=\{e^{j\frac{2\pi}{u_1m}n} | 0 \leq n \leq u_1m-1\}$. For convenience, let $e^{j\frac{2\pi}...

1 hour ago, by Akiva Weinberger
Conjecture:$$\frac{\frac14}{1!}-\frac{\frac11\frac13 +\frac12\frac12+ \frac13\frac11}{2!}+\frac{\frac11\frac11\frac12+ \frac11\frac12\frac11 +\frac12\frac11\frac11 }{3!}-\frac{\frac11\frac11\frac11 \frac11}{4!}=0$$
^ You mean this? This is trivially zero by evaluation.
Unless you want me to generalize, expand, telescope and say the pattern expressed here is zero, I find no challenge.
Infact the pattern in the numerators are beyond me.
Neglecting that part, the integral for the alternating signed reciprocals of the gamma function is non-zero. So, this conjecture is doubly difficult to prove right or wrong.
You could try induction. I have a small idea for that. But I'll leave it to you to figure out.
@user1732 Good luck!
~toodles~
09:35
cya
10:00
@Nick The pattern is, the numerator of the $i$th fraction is the sum of fractions of the form $\frac1{a_1}\frac1{a_2}\dotsb\frac1{a_i}$ where $a_1+\dotsb+a_i=n$
and the denominator is $i!$
In the one you quoted above, $n=4$
@AkivaWeinberger are you here?
@AkivaWeinberger I'm wondering
Hm, I could also write it more uniformly
I'm thinking about the irrationality of sqrt(2)
and whether it follows from Gauss' lemma
and if so why do we bother proving it using infinite descent
10:14
as the sum of things of the form $\prod(-\frac1{ia_i})$ where $\sum a_i=n$
@LeakyNun Which one is that, again?
X^2-2 is irreducible over Z[X] so it is irreducible over Q[X]
Does the proof of that require unique factorization?
Hi guys, are there good individual books(lots of theorems and solutions to problem exercises) for each of these Discrete Mathematics topics like


1)Relations,functions,cardinality and sets
2)Graph Theory
3)Counting and probability
@AkivaWeinberger maybe
In any case, $\sqrt2$ is one of the first proofs you learn, most people learning about it won't have heard of Gauss's lemma yet
There's a geometric proof somewhere as well (also infinite descent though)
And the continued fraction one is nice
Remind me how you prove that $a\sqrt2+b\sqrt3+c\sqrt6$ is irrational for integer $a,b,c$ not all equal to zero?
10:35
@AkivaWeinberger that's insane XD
10:48
Ah, looked it up
$a\sqrt2+b\sqrt3+c\sqrt5+d\sqrt6+e\sqrt{10}+f\sqrt{15}+g\sqrt{30}$ is the actual hard one
but that's the one that generalizes most easily
Hi, in this book, Jost argues that two loops (possible based at different points) which are freely homotopic through loops are homologous. Does someone know where I can find a detailed proof? i.imgur.com/hDxbv8m.png
What does homologous mean to you?
You're thinking singular chains? And these loops give homology classes by triangulating them?
That the loops $\gamma_0,\gamma_1 \colon S^1 \to X$ determine the same class in $H_1(X;\mathbb{Z})$.
@abenthy That is a proof, isn't it?
Also, took me a while to realize that the parentheses weren't actually mismatched
Once you triangulate the cylinder, yes
10:56
But you have to triangulate it in a nice way, dont you?
I wonder if you can do continued fraction for Q_p
Oh, right, $A$ is a square and we want it to be a triangle
@abenthy You're worried about respecting given triangulations on the boundary?
Cut it diagonally
@Mike Yes, exactly.
10:57
But you already know that the triangulation doesn't matter: you get the same homology class no matter what.
On the loop.
Eg, this follows because H_1 is the abelianization of pi_1.
@AkivaWeinberger Thanks for posting the picture.
In fact it is not difficult to show that for triangulations of the circle, any two are related by a common subdivision.
Triangulate $A$, not $\Sigma$ (triangulate it in the domain space, not the range space)
And subdividing a triangulation preserves homology class.
10:58
Besides, you can always extend two different triangulations $\Delta_1$ and $\Delta_2$ on $S^1$ to a triangulation on $S^1 \times I$ that restricts to $\Delta_i$ for $i = 1, 2$ on the two boundary copies.
@BalarkaSen Too much work imo
It's "obvious"
But yeah
Wait, I'm confused, why are the loops triangulated
I would rather just triangulate the annulus and demonstrate that the triangulation of the boundary is inessential
The loops are just images of $[0,1]$
10:59
I'm also a little bit confused.
@AkivaWeinberger What is a homology class?
Is $\gamma_0$ a loop or its homology class?
He should write $[\gamma_0] - [\gamma_1]$ instead of $\gamma_0 - \gamma_1$, but what worries me more is that he writes $\partial H(A)$ for a boundary, where its not clear to me how he identifies $S^1 \times [0,1]$ with a $2$-simplex.
@MikeMiller It makes for some interesting questions nonetheless. If $(M, \Delta_1)$ and $(M, \Delta_2)$ are two different triangulations on a manifold $M$, are they always "concordant"? Namely, is there a triangulation on $M \times I$ which restricts to $\Delta_1$ on one side and $\Delta_2$ on the other side?
This is very difficult.
11:01
I suspect this has to do with Hauptvermutung
@abenthy The boundary as a manifold.
All the boundary pieces of the 2-simplices on the interior cancel out.
So you're only left with triangulations of the manifold boundary.
Ah, so he consideres the image of the homotopy in the manifold and takes the boundary?
I thought $\partial$ is the boudary operator in homology.
@abenthy He already says he identifies $S^1$ with $[0,1]$, so you just need to make $[0,1]^2$ into a simplex. And the answer is to cut it diagonally
I see. That makes sense.
@BalarkaSen You are correct. My understanding is that these are related to the homology cobordism group.
Concordance classes of PL triangulations are surely better understood.
11:03
If Hauptvermutung holds for $M$ (i.e., $\Delta_1$ and $\Delta_2$ have a common refinement), then this is true (because going from one triangulation to a subdivision can be done through a concordance)
So counterexamples are going to be uber complicated I suppose
I am somewhat certain that two oriented PL triangulations on M are concordant iff they are oriented PL homeomorphic
Huh
brb, gotta get coffee
@AkivaWeinberger let L/K be infinite separable algebraic such that every polynomial in K[X] splits in L
Then L is algebraically closed
I can't prove it rigorously
which means, I can't prove it at all
11:08
So you have the polynomial $a_0+a_1x+\dotsb+a_nx^n$ with $a_i\in L$, and you want to show that it has a root in $L$
And since the $a_i$ are in $L$, each $a_i$ is the root of a polynomial $b_{i,0}+b_{i,1}x+\dotsb+b_{i,n}x^n$ for some $b_{i,j}\in K$
Oh, wait, hold on
oh and K is perfect
$L$ isn't the algebraic closure, $L$ is just something containing the algebraic closure?
What does "infinite separable algebraic" mean
Also "perfect"
@BalarkaSen Ok, part of the old Galewski Stern result on triangulations is the following.
Let $\mu: \Theta_3 \to \Bbb Z/2$ be the Rokhlin homomorphism.
@AkivaWeinberger yes, it follows that L is the algebraic closure of K
infinite = infinite field extension
We have the Kirby Siebenmann invariant in $H^4(M;\Bbb Z/2)$, which is an obstruction to PL triangulation.
11:11
separable = separable extension
algebraic = algebraic extension
perfect = every polynomial is separable
OK, so the last thing I wrote follows from $L/K$ being an algebraic extension
What's separable again, though? Something about repeated roots?
We may apply the Bockstein of the above surjectivr homomorphism / short exact sequence to get a class $\delta \Delta(M) \in H^5(M;\ker \mu)$. This vanishes iff $M$ is triangulable.
Further, if a triangulation exists, they are classified up to concordance by classes in $H^4(M; \ker \mu)$.
So we need to understand $\ker \mu$ better.
It's infinitely generated. Does it have torsion? Is it free? Are there elements divisible by arbitrarily large integers? Etc
@AkivaWeinberger every x in L is separable over K, i.e. the minimal polynomial of x has no repeated roots in a splitting field
So I remember reading this proof but not how it goes
OK so suppose $\sum a_ix^i$ doesn't have a root in $L$
then we can do $L[x]/(\sum a_ix^i)$, or equivalently, we can do $L(\alpha)$ where $\alpha$ is its root
Hm maybe this doesn't work
If $M$ is algebraic over $L$ and $L$ is algebraic over $K$ then $M$ is algebraic over $K$, the question is why
Ah wait
Linear algebra
maybe
Linear algebra solves all our problems, we just need to find out how
Like, assume $\{1,\alpha,\alpha^2,\dots,\alpha^N\}$ is linearly independent for a certain $N$, get a contradiction somehow
Probably the product of the degrees of $a_i$
11:23
Hi
@AkivaWeinberger ok I have a complicated proof
hey @AlessandroCodenotti you're also an algebraist
18 mins ago, by Leaky Nun
@AkivaWeinberger let L/K be infinite separable algebraic such that every polynomial in K[X] splits in L
18 mins ago, by Leaky Nun
Then L is algebraically closed
@LeakyNun Oh, I wasn't aware
Happy birthday, here's some algebra
So I think the deal is, I think $\{\alpha^ia_0^ja_1^k\dots\}$ is a base for the $L(\alpha):K$
where $i$ ranges from $0$ to the degree of $\alpha$ over $L$ and $j$ ranges from $0$ to the degree of $a_0$ over $K$ etc
and $\alpha$ is the root of $a_0+a_1x+\dotsb+a_nx^n$
(so I guess $i$ ranges from $0$ to $n$)
What is the rank of $f^*\mathcal{F}$ if $\mathcal{F}$ has rank $n$ and $f:X\to Y$ is a degree $q$ map of Riemann surfaces?
11:32
And since that's gonna be a finite set, we get that it's finite dimensional, so $\{1,\alpha,\dots,\alpha^N\}$ is gonna be linearly dependent over $K$ for large enough $N$
specifically for $N=n\prod\deg(a_i)$
I've computed the degree, but not immediately sure how the rank of a bundle behaves under the inverse image of sheaves
@user600999 the pullback of a bundle always has the same rank!
Really?
That's dank
That's dank af
Perhaps the degree of the pullback is also easy to give?
That's rank
11:35
I computed it using the adjunction of $f^*$ and $f_*$ and using Riemann Hurwitz formula
@AkivaWeinberger I don't see how that gives you a contradiction
It's not, I ended up not doing a proof by contradiction
$$\text{deg}(f_*f^*\mathcal{F})=\text{deg}(f^*\mathcal{F})+(1-g_M)-\text{deg}(f)‌​(1-g_N)$$
Where $f:M\to N$ is a nonconstant holomorphic map of compact Riemann surfaces of genera $g_M$ and $g_N$
We just need $\{1,\alpha,\alpha^2,\dots\}$ to fit inside a finite-dimensional vector space over $K$
and $\text{deg}(f_*f^*\mathcal{F})=\text{deg}(\mathcal{F})$, so I can just move stuff over to LHS
11:37
which means we need a finite set that spaces a vector space that it sits in
Is that a.o.k @loch
Can someone verify whether the suggested answer is correct? I have e terms for some reason
@PrashinJeevaganth You confirm mine and I'll confirm yours
@user600999 I dont even understand your problem XD
11:39
Ripped
My Calculus aint that high level yet
It's some higher order calculus that I am doing
Need to take Calculus IX
could tell from the word Rienmann
well all my math modules are "elementary"
That man invented calculus XIII
11:40
oh was it not Newton?
lul
hmm I'm unfamiliar with your version of Riemann Hurwitz

but you should find $deg(f^*\mathcal{F}) = qdeg(\mathcal{F})$
Newton made Calculus -IX
He's joking. There's no such thing as Calc XIII
That's too many
What?
Surely there is calc XIII!
@loch oh
No, XIII! is even more
11:41
@AkivaWeinberger I don't see how that gives the result I want
I dont even recognise Roman numerals that well lol
So $\alpha$ is algebraic over $K$
@loch You are sure of this?
@PrashinJeevaganth Calc 13
What did we want to show again
@loch I thought that was for pushforward?
11:42
@user600999 gg who came up for the numbering of calculus classes
Mine's prolly at 3 I guess ...
@PrashinJeevaganth At calculus XXI you start doing derived topos theory on spectral shtukas
topos doesn't sound good ...
I bet you could find calculus XXX if you looked hard enough
Oh geez, oh gosh Akiva
@AkivaWeinberger we want to show that L is algebraically closed...
11:44
Right OK so all polynomials in L have roots in L
and we also know all polynomials in K have roots in L
Hi,

I would like to understand how to project a point onto a line.

The first method that comes to my mind would be to say that the line is defined by two points $q_1, q_2$. Point $p$ is projected onto that line by finding a point $p'$ on the line that is closest to $p$.
the vector $\vec{pp'}$ is orthogonal to $\vec{q_1q_2}$. Using these facts there can be solved an equation system.

However in my lecture I have read this:

$$\frac{1}{||v||}v \left( \frac{1}{||v||}v^{T}p\right) = \frac{1}{||v||^2}(vv^{T})p=\frac{vv^{T}}{v^{T}v}p$$
Give the elementary calculus kid a chance guys, is the answer correct cuz I got the same expression with e^-3 term inside
Therefore we need to turn a polynomial in L into a polynomial in K
@loch My form is $\text{deg}(f_*\mathcal{L})=\text{deg}(\mathcal{L})+(1-g_X)-\text{deg}(f)(1-g_Y)‌​$
and I am fairly sure $\text{rank}(f_*\mathcal{L})=\text{deg}(f)\text{rank}(\mathcal{L})$
Slightly skeptical of your claim for deg(f^*F)
tough crowd
@loch Okay I believe you, it's in my hearts horn
12:11
@TedShifrin Wow, I didn't know this!
@BalarkaSen ^^^
@loch I am super confused now. p306 of Hartshorne tells me that $f_*\circ f^\times:Pic_Y\to Pic_Y$ is given by multiplication by $deg(f)$
(That $\times$ is just $*$ but the math-mode was getting rekt)
But the unit of adjunction at component $\mathcal{F}\in\text{Sh}(Y)$ should tell me that $f_*f^\times\mathcal{F}\cong \mathcal{F}$, and degree is preserved by isomorphism surely right?
12:24
$f_*f^*\mathcal{F}$ is very rarely isomorphic to $\mathcal{F}$!
Right, I goofed, I just get some morphism form the adjunction
Not an isomorphism
12:56
@BalarkaSen I can write $\langle\nabla_XY,Z\rangle$ without $\nabla$ (this is the Koszul formula)
Can I write $(\nabla_XY)f$ without $\nabla$? (Where by $Xf$ I mean the directional derivative of $f$ in the direction of $X$ at each point)
For example, $\nabla_XY\langle Z,W\rangle$
@AkivaWeinberger it's very hard to prove that if every polynomial in K has a root in L and L/K is algebraic then L is algebraically closed. And to prove that, the last step is to prove that if L/K is separable and K is perfect and every polynomial in K splits in L then L is algebraically closed.
It seems to me that you've trying to prove the former question.

« first day (2993 days earlier)      last day (2325 days later) »