« first day (5010 days earlier)      last day (11 days later) » 

12:00 AM
A classic Leslie zinger...
 
they'd heard something somewhere about 1, 2, 4, 8 and i began going into adams's proof and they stopped me and said "actually i'm not that curious about the details"
haha
side note, i don't know of a proof of that result that does not use algebraic topology
 
@monoidaltransform he gives the argument that two equioriented balls are ambiently isotopic in the second-to-last paragraph. it's a pretty clear sketch, in my opinion.
 
That means I will not be seeing a proof of this for many years hence
oh well
 
@monoidaltransform a smooth embedding can be isotoped first into a chart and then into a linear one via $(x,t)\mapsto\frac{f(tx)}{t}$, then it comes down to $SL(n)$ being path-connected
in the smooth category, this is known as Palais' disk theorem
combining this with the generalized Schoenflies theorem should yield the annulus theorem in the smooth category also, if I recall correctly
the annulus theorem in the tame category, however, is very hard
 
@Thorgott I see. Yes. Ok. Now that is cleared up, but I have no idea how that answers my overflow question. Second, what makes the annulus theorem in the tame category hard?
Apparently, this result of Palais should imply my question
 
12:10 AM
I mean, take your favorite embedding $M\hookrightarrow\mathbb{R}^N$. this yields an embedding when restricted to $B$ (if $N>n$, there is nothing to worry about; if $N=n$, you have to ensure the orientations are right), so you can find a homeomorphism of $\mathbb{R}^N$ taking this to the standard embedding and that's it
 
@leslietownes When and where was that?
 
as to what makes the annulus theorem in the tame category hard: I don't know, simply that I have no clue how it works and that it took Quinn to prove it
 
xander a while ago, and here in southern california
 
@leslietownes My father would typically make an attempt to read the thesis / dissertation of anyone he was on a search committee for. It seems that, perhaps, there was someone else like that in SoCal.
Or another lawyer with a math degree.
 
@Thorgott Two questions. Second is mainly out of curiosity. (1). Can we always ensure that the orientations are right? Why? (2). According to Moishe, one can get a local diffeomorphism using Hirsch-Smale theory, can one complete the argument using that?
 
12:14 AM
@monoidaltransform Sure. If they are left, just turn all the arrows around the other way.
 
Ah right because we're working with connected manifolds
OH I see what you did there. @XanderHenderson haha
 
I'll just try and post a quick answer
 
okay thanks @Thorgott
 
12:33 AM
@monoidaltransform you don't mind assuming that the original ball $B$ in your question is collared, do you?
 
@monoidaltransform I was worried for a second...
 
No. I think I want it to be any. Is it false in this case?
@XanderHenderson hahahah. That made me laugh
 
I don't know
I mean, for large $N$, it's always true
but for $N=n$ and arbitrary $B$, I don't know
 
@Thorgott the construction of $\alpha$ in page 18 in maths.ed.ac.uk/~v1ranick/papers/ks.pdf seems kinda like what i'm looking for, but I don't know and this is beyond me
The more I think about it the less i'm convinced that it is haha
 
12:50 AM
yes, that's one explicit construction of an immersion of the punctured torus into Euclidean space of the same dimension, there's more elaborate versions elsewhere
but note how their CAT = DIFF or PL, so their balls are certainly collared!
 
In DIFF, a ball is always collared?
 
that's the tubular neighborhood theorem
well, a special case thereof
 
I see. In this case, could you please write the details of their argument in the answer then? If all of this is spelt out, I would certainly find it very useful, enlightening and be extremely thankful ad would accept it
 
1:28 AM
I posted an answer addressing the question at hand. If you want clarification on some step of the mentioned answer of Lee Mosher, I believe that should be the subject of a separate question.
 
1:45 AM
@Thorgott Thank you very much! Just one last point, the embeddings of the large ball and smaller ball (B and B') will be equioriented because B is collared and the other other is contained in it?
 
@leslietownes @Thorgott can one of you give me an overview of Clifford algebras?
 
@monoidaltransform the argument does not really care about orientation at that point, but this is true in any case. you can construct these balls explicit by taking a homeomorphism B = B^n and having B' correspond to a Euclidean ball of radius a sufficiently small epsilon
@Jakobian I've only encountered the definition once and it did not leave a lasting impression on me
 
Oh. Do you know where to find some info on them?
Normally I wouldn't bother but I want to help a friend with understanding them
 
2:06 AM
I've encountered them in the context of Lie groups/algebras, so that's where I would look, but I don't know if that is relevant to the context your friend may be operating in
 
2:38 AM
@Thorgott same
or maybe that was properly directed to jakobian :)
 
Bob
2:52 AM
Does Ted, the retired math prof, still come here?
 
3:03 AM
he's taking a break currently
 
Bob
thanks
my impression is that the skill in math of US students continues to go down
I am thinking of high school students and non-math majors in college
does anybody have a good feel for it?
I do not
 
it is very difficult to talk on behalf of millions of people, and maybe that is the first thing to keep in mind
math education remains an unsolved problem, in the US and elsewhere
 
Bob
okay, thanks
good night
 
 
4 hours later…
6:48 AM
$n(1-z_0) = t(p_2-p_1)$ is the constraint. In fact, this has some relationship with the twin primes conjecture
Starting with solving the equation $p_1x+p_2y=1$
If p1,p2 are twin primes and n is even, then the equation has solutions $(1,1)$ and solves the goldbach's conjecture
For the general goldbach conjecture which ask if every integer can be written as the sum of two primes, this only occurs if the prime gap is a multiple of the given integer
So the whole question resolves to, is there an integer for every gap between primes
 
 
5 hours later…
11:30 AM
Suppose I have a sequence of subsets $A_i$ of some universal set $C$ and suppose that $\bigcup_{i=1}^\infty A_i$ exists. Is it then always possible to construct an increasing sequence of sets by $$B_n=\bigcup_{i=1}^n A_i$$ whose union is also $\bigcup_{i=1}^\infty A_i$? In other words $$\bigcup_{n=1}^\infty B_n=\bigcup_{i=1}^\infty A_i.$$ The reason I'm asking is because it is claimed a $\lambda$-system is a $\sigma$-algebra only if it is $\pi$-system, and this argument is kind of used.
 
11:41 AM
Let me rephrase my question...
The precise argument goes like this. Let there be $\lambda$-system $\mathcal C$, which is a monotone class. If in addition it is a $\pi$-system, then it is closed under finite intersections and consequently also under finite unions. Now, given a sequence $A_k$ in $\mathcal C$, then each finite union $B_n=\bigcup_{k=1}^nA_k$ is also in $\mathcal C$. As $B_n$ is increasing, the union $\bigcup_nA_n=\bigcup_nB_n$ is in $\mathcal C$, showing that an arbitrary countable union is in $\mathcal C$.
What I was doubting was $\bigcup_nA_n=\bigcup_nB_n$...
But showing both inclusions is maybe trivial
 
this should not be unintuitive, try working it out
 
 
1 hour later…
1:02 PM
@psie the union exists?
 
@Jakobian it always exists, or? :)
 
It does
 
ok, then I was trippin
 
Since $A_n\subseteq B_n\subseteq \bigcup_i B_i$ you have $\bigcup_i A_i\subseteq \bigcup B_i$
Agreed?
 
yeah, agreed
 
1:11 PM
And since $B_n \subseteq \bigcup_i A_i$ you have $\bigcup_i B_i \subseteq \bigcup_i A_i$
 
great
 
1:56 PM
"god created natural numbers. everything else is the work of man"
-Leopold Kronecker
 
2:10 PM
Ohhhh, never mind. Will delete
 
Forgot how to do that remove thing
 
2:29 PM
@RyderRude man created natural numbers, everything else is a work of man
 
Did man make trees? The Earth? The galaxy?
 
@YourLordJoyBoy yes, sure. Whatever you want
 
@Jakobian Just asking a question XD
BAM! 300 votes, civic duty earned.
 
@Jakobian man created man :)
 
Man created nothing without taking something that already existed to make it :) And I'm gonna stop right there.
 
2:38 PM
@YourLordJoyBoy I disagree because changing structure of something classifies as making something
 
Cars were something else that already existed before becoming cars. Common sense tells us that correct?
Same with houses, heck even computers.
 
Having a bit of a brainfart moment still...hoping to get some help
So one has the fact that $n>m$ for naturals n,m and $0<a<1$ implies that $a^{1/n} > a^{1/m}$
This I am trying to prove (using properties of square roots and exponentiation in the rationals which I've already proven)
I seem to be proving the exact opposite though...
I've looked at it a few times and can't see where I am going wrong. Any tips?
Oh my god
horrific mistake
I see now...I should probably delete this too. My grade school recollection of basic arithmetic going out the window once I formalize
 
2:54 PM
@EE18 There there bruh
 
@YourLordJoyBoy yeah actually I don't disagree. I had some weird way of interpreting your comment
 
@Jakobian :D
 
$(-m)^{-1} = (-1)m^{-1}$ in a field right?
Ya I guess so: $(-m)^{-1} = ((-1)(m))^{-1} = (-1)^{-1}(m)^{-1} = (-1)m^{-1}$ in a field
 
3:10 PM
indeed
 
Bout to dance happily over polynomials making SENSE!
 
3:46 PM
@YourLordJoyBoy cool
 
4:01 PM
@Jakobian Now if they make sense Thursday, freedom :D
That AA degree is ALMOST MINE! CAN JUST ABOUT TASTE IT!
 
4:16 PM
@YourLordJoyBoy the taste of paper, eh?
 
4:27 PM
Fun fact: I finished my phd in spring of 2020 (I was, I think, the first person at the university to present a phd defense over Zoom---I was certainly the first in the department). Because of the prevailing conditions, I was not able to walk or be hooded by my advisor. However, in 2022, I was hooded by the vice president of my current institution---the degrees conferred at that commencement were mostly associates degrees.
 
@XanderHenderson so you did your PhD in your 40s.
 
@XanderHenderson Old man Xander proves he's old.
@Jakobian -speaks in Scott Mcneil's Piccolo voice- Funny
 
@LuckyChouhan I finished at about that time, yes.
It took me more than a decade to complete a bachelors degree, then 2.5 years to finish a masters. I took a few years off between the BA and the MS to teach, and a little time off between the MS and the PhD to lecture.
 
@XanderHenderson Impressive regardless.
Hell it took 4 tries to finally get my ged.
Guess which portion gave me the biggest problem.
ALSO!
$(b^4)(5b^5)$ Is this truly simple as I think?
 
4:44 PM
@YourLordJoyBoy what is it that you think
 
@Jakobian Would it be $6b^9$?
 
no actually
how did you get that?
 
@Jakobian I think I had a brainfart of my own...
 
don't guess, tell me how to obtain the solution
 
Hrmmmm......$5b^9$?
 
4:47 PM
I don't feel like writing new exam questions today... maybe I'll just use an exam from three years ago... :/
 
@YourLordJoyBoy: did you get ChatJax working?
 
@Jakobian Since its multiplication, I'm pretty sure you add the exponents. @robjohn I've had it working since last week but yes.
 
@robjohn we had to troubleshoot few times :P
 
@Jakobian But got it in the end.
 
@XanderHenderson change question order, or just the same exam?
 
4:48 PM
@robjohn I'll mix up the order a bit.
Change a 2021 to 2024.
 
I figured that had to be changed, so a change of order seemed reasonable.
 
@YourLordJoyBoy well tell me exactly how would you write it down. Step by step
 
@Jakobian do you remember what the problem was?
 
@robjohn yeah. @YourLordJoyBoy didn't have their bookmark tab enabled
 
Hmm. I thought we had checked that. Any way, glad it’s working.
 
4:51 PM
@Jakobian $(b^4)(5b^5)$ Add the exponents, and since there's an invisible one with the $b^4$ that ultimately gives you $5b^9$.
 
@YourLordJoyBoy You want to simplify $(b^4)(5b^5)$ so you would have to write some equalities. Don't write it in words to me, write some equalities
 
@robjohn I ultimately had to log in with my other gmail to finally make it work. It won't work with my school gmail, that is what I learned.
 
@YourLordJoyBoy that is odd. I don’t know why that would be.
 
@XanderHenderson why a decade?
 
I dicked around. Didn't go to class. Took a couple of years off.
 
4:54 PM
@robjohn Stupidly secure. @Jakobian Trying to do that. um, how do I show the adding exponents bit?
 
Greetings @robjohn ! How are you?
 
@XanderHenderson The child never skipped class XD
 
@YourLordJoyBoy okay maybe its better to just write it down for you. You go like $$(b^4)(5b^5) = b^4\cdot 5\cdot b^5 = 5\cdot (b^4\cdot b^5) = 5\cdot b^{4+5} = 5\cdot b^9 = 5b^9$$
 
@XanderHenderson Oh, so what your batchmate used to say to you and what did you tell your parents when they realized "Isn't Xander taking too much time to complete his bachelors?"?
 
@LuckyChouhan Just getting over a bit of Covid. My wife picked it up on our way back from Texas and then I picked it up from her.
 
4:56 PM
@LuckyChouhan "Batchmates"?
 
$$(b^4)(5b^5) = b^4\cdot 5\cdot b^5 = 5\cdot (b^4\cdot b^5) = 5\cdot b^{4+5} = 5\cdot b^9 = 5b^9$$ ?
 
@robjohn Oh I'm feeling so sorry for you. I know you recently have been through the surgery. Wishing you good health. Take care,
@XanderHenderson yeah!
 
@YourLordJoyBoy why did you copy paste what I wrote
 
I have no idea what that means...
 
@robjohn SUCKS! Covid was a pain in the ass.
@Jakobian Making sure I can do it too is all. In the end, I was right. :D
 
4:59 PM
@XanderHenderson May be some of them were doing PhD when you were still doing your bachelors.
 
CONFIDENCE IS RISING!
 
@LuckyChouhan I don't know what "batchmate" means.
That is a nonsense word.
 
@YourLordJoyBoy uhhh. By copy-pasting it?
 
@Jakobian So I'm a lil lazy
 
@XanderHenderson classmates , friends. Then what do you call a person who studies with you in your college class?
@XanderHenderson then please suggest a better word.
 
5:00 PM
@YourLordJoyBoy writing it down verbatim doesn't really do anything. I posted it for you to understand
 
My point: I wanted to make sure $5b^9$ was correct. And it was, which says my thought process was right.
 
@LuckyChouhan "classmates" or "friends" would work...
@LuckyChouhan Well, I spent my first two years at a hoity-toity small private liberal arts college. At the time, I was a Russian Studies major. After I failed out of that institution, I pretty much lost touch with most of those folk.
 
@Jakobian Get what I'm sayin man? :D
 
since you guessed that the answer is $6b^9$ at first which shows you lack understanding
 
My confidence is beginning to rise higher and higher
 
5:02 PM
I then spent a few years at Large State University as an anthropology major. I was not super close with most of the other people in that program---I was interested in doing science, while most of the rest of the department wanted to tell fairy tales. I left after a couple of years.
 
@XanderHenderson I see, but finally you succeeded haha, Xander is a PhD now.
 
@YourLordJoyBoy what is $z^2(z^2+z)$ ?
 
@Jakobian You actually explained what I was trying to say.
 
And then a few years later, I went back to school and finished a BA in mathematics with a minor in education (so that I could teach). I am still close with a couple of people from that program.
 
@XanderHenderson so you had changed lots of subjects.
 
5:04 PM
@XanderHenderson so you ended up not doing science :P
 
@Jakobian No. I did math.
 
@XanderHenderson did you like math and science in your high school?
 
@LuckyChouhan Sure.
 
@XanderHenderson your replies are faster than ChatGPT's
 
@XanderHenderson are you actually pretty knowledgeable when it comes to some of the science subjects? Physics, chemistry, maybe biology?
 
5:08 PM
@Jakobian Nope.
 
@Jakobian what about you?
 
Nope
I just know some things about biology from the stuff I watch on youtube
 
@Jakobian $z^4+z^3$. You have to distribute the $z^2$ outside with all in the parenthesis. Simply put, $z^2 x z^2 + z^2 x z$ Simplifying to $z^4+z^3$.
You get what I mean
 
@YourLordJoyBoy is $x$ supposed to be multiplication?
 
Yes
Couldn't figure out how to do the dot
 
5:11 PM
you can write \cdot to have $x\cdot y$ or \times to get $x\times y$
 
I have taken basically no sciences in my life. As an undergrad, I took one semester of chemistry, and one semester of geology.
In high school, I took a year of "earth science", a year of chemistry, and a semester of physics.
 
@YourLordJoyBoy its correct
 
And in grad school, I took a quarter of quantum mechanics.
 
@Jakobian See, that's what I didn't know how to do. If I lack understanding, it is in the coding.
 
I think @YourLordJoyBoy is playing with @Jakobian
 
5:11 PM
That is all the science I know.
 
@LuckyChouhan I'm not. I want to ensure that I'm doing this stuff correctly.
 
@XanderHenderson do you discuss math, science, literature with your kids?
 
My math confidence isn't as it should be so I'm trying to boost it so I'm ready for the final.
 
@LuckyChouhan Sure.
 
@XanderHenderson oh geology sounds super cool. I've watched a video recently about how in the oceans there are those underground "rivers" that carry sediment
thats how there can arise rock formations that look as if there was a river but actually there wasn't
 
5:13 PM
@Jakobian Would be nice to have time for these vids.
 
@XanderHenderson Is your real name is Xander or Alexander? Because I remember there used to be a file and app sharing famous app called Xander.
 
@XanderHenderson Imagining you as a green ranger with a beard.
 
sorry, not underground but underwater
 
TIME FOR THE GAMBLE RUMBLE
 
5:39 PM
can someone help me with this part of the dirichlet problem?
0
Q: How can I show that $h(z)=\mathbb{E}(u(B_{\tau_D}))$ satisfies the mean value property?

Summerday Let $D$ be a complex domain and $u$ a bounded function on $\partial D$, let $B$ be a complex Brownian motion and denote by $\mathbb{P}_z$ the distribution of $B$ if $B$ starts at $z\in D$. Define $h(z)=\mathbb{E}_z(u(B_{\tau_D}))$ where $\tau_D:=\inf\{t\geq 0: B_t\notin D\}$. I want to show that...

 
The war of change
Can we only edit some posts and others are stuck?
The next stage :O ITS WITHIN REACH
 
5:56 PM
What happens if I find a point of non-differentiability in a point that is not in the domain of the main function?
The domain of the main function is $(-2,0) \cup (0, +\infty)$
The domain of the derivative instead is $(-\infty,-2) \cup (-2,0) \cup (0,+\infty)$
I found the point of non-differentiability at -2 plus and minus.
Main function = $\log(\frac{x^2}{x+2})$.
Derivative = $\frac{x+4}{x^2+2x}$.
 
@Pizza The professor's pizza returned
Just in time to be REMINDED
 
6:29 PM
The Borel sigma algebra on $\mathbb R^n$ is generated by the open sets, but since a sigma algebra is closed under complements, it also contains the closed sets and they too generate the Borel sigma algebra. This makes sense. But I don't understand why the compact sets also generate the Borel sigma algebra. The motivation given in my notes is simply:
> Since $\mathbb R^n$ is $\sigma$-compact (i.e. it is a countable union of compact sets), its Borel $\sigma$-algebra is generated by the compact sets.
I'd be grateful if someone could provide a bit more explanation as to why the Borel sigma algebra is also generated by the compact sets.
 
@psie If $\mathbb{R}^n = \bigcup_n K_n$ where $K_n$ are compact, and if $F$ is a closed set then $F = \bigcup_n (F\cap K_n)$ where $F\cap K_n$ are compact
 
ah ok, this makes sense
 
see this post for some examples of when sigma-algebras generated by compact and closed sets don't agree
a similar notion is Baire set which is an element of the $\sigma$-algebra generated by compact $G_\delta$ sets. For nice spaces this coincides with Borel set
 
ok 👍
 
 
1 hour later…
7:52 PM
I have a basic question. What is the pushforward of Lebesgue measure $\lambda$ on $\mathbb R$ under the mapping $x\mapsto rx$, where $r>0$? I'm reading about the monotone class theorem and an important corollary that follows the theorem that highlights some conditions when two measures are equal on a measurable space. The author says it should follow from the above corollary that the pushforward is $r^{-1}\lambda$. How?
I'm not sure I even understand the notation $r^{-1}\lambda$.
 
is anyone familiar with chern-simons theory from the mathematical side of things :0
 
@psie $(r^{-1}\lambda)(A) := r^{-1}\cdot \lambda(A)$
 
ah ok :) that part makes sense then
 
The pushforward where $f_r:x\mapsto rx$, $((f_r)_\# \lambda)(A) = \lambda(f_r^{-1}(A))$
 
the question is of cohomological nature i think...anyways here it goes. in chern-simons theory we have so called "on-shell" (satisfying a particular differential equation called the equation of motion) Lie algebra valued connection $1$-forms $A$ on a principal bundle $P$. Assume that $A$ can be globally defined (which amounts to restricting what the structure group is).
 
8:00 PM
they are saying you should look at $\mathcal{C} = \{(a, b) : a, b\in \mathbb{R}\}$
 
So when the structure group is non-abelian, then the equation of motion is $dA + A \wedge A = 0$.
 
@Jakobian ok, I will try
 
WHEW! What a day. Studying, housework plus dogsitting.
 
We have so called "gauge transformations" $A \to A + dX + A \wedge X$ where $X$ is a Lie algebra valued $0$-form
 
$f_r^{-1}(a, b) = \{x\in\mathbb{R} : rx\in (a, b)\} = (r^{-1}a, r^{-1}b) = r^{-1}\cdot (a, b)$
 
8:02 PM
when $G$ is abelian we get a nice identification that "gauge equivalent" and on-shell connection $1$-forms is precisely an element in the first de Rham cohomology class. This is because in the abelian case, the equation of motion is $dA = 0$ (A is an exact $1$-form) and $A \to A + dX$, which only differs by an exact $1$-form.
 
its Lebesgue measure is $r^{-1}b-r^{-1}a = r^{-1}\cdot (b-a) = r^{-1}\cdot \lambda((a, b))$
 
The question is: when $G$ is not abelian is there an analogous identification?
 
ah, cool, thanks a lot
 
Can someone check my msg above pls
 
@Pizza what does "-2 plus and minus" mean?
 
8:13 PM
isn't K radial in the hypothesis?
if yes, then what is happening in the 2nd line of the proof?
 
I swear, that algebra 1 final is laughing at me! >:(
IT WILL BE BEATEN!
 
@Koro $k$ is decreaasing so it has at best jump discontinuities. We can approximate it from below by some sequence of functions $k_j$.
 
Hey everyone, does anybody know of a quantity related to "sum of all probabilities below a certain threshold" for probability distributions?
There's an easy upper bound of the Shannon entropy, but that does not depend on the threshold
 
@Jakobian yes, that is fine. But my question is that the first line of the proof says-we prove 2.1.9 when K is radial. But K is already radial, no?
K(x)= k(|x|)
 
Yes. But you said that if yes, then what is happening in the 2nd line
 
8:22 PM
I asked it incorrectly then, my bad.
So if K is radial, then there is no problem.
 
I believe that the first line saying that $K$ is this and that, is only a statement of what we are assuming right now
 
Why do they say in the proof that 'we prove 2.1.9 when K is radial'?
 
so yes, we are proving this for $K$ radial but its like saying... let $X$ be a compact Hausdorff space when in the statement of your theorem its stated that you are proving this for compact Hausdorff spaces
 
I get it now.
The first line in the proof says: when K is 1) radial, 2) satisfying 2.1.8 and 3)compactly supported and continuous.
Once proven for such K, then by density we'll prove it for general K in the theorem.
 
what's the definition of a holomorphic function on a complex analytic variety?
 
8:40 PM
In mathematics, and in particular differential geometry and complex geometry, a complex analytic variety or complex analytic space is a generalization of a complex manifold that allows the presence of singularities. Complex analytic varieties are locally ringed spaces that are locally isomorphic to local model spaces, where a local model space is an open subset of the vanishing locus of a finite set of holomorphic functions. == Definition == Denote the constant sheaf on a topological space with value C {\displaystyle \mathbb {C} }...
this answers it, though perhaps you have a different definition in mind?
 
9:13 PM
How algebraic can you make theoretical cryptography? As far as I can see, although theoretical cryptography employs and blends many different fields, it uses most of these on a surface level. The algebra used in cryptography seems less advanced than actual algebra research (well, duh, I guess).

My question is: Is it possible to have your cake and eat it too? Can I do cryptography research that reads and feels like commutative algebra / group theory research? Do you know anyone who does this sort of thing?
 
9:41 PM
ephe i don't know what you have in mind specifically, but a lot of cryptography research is as "algebraic" as algebraic research gets. you might look at the submissions from various research groups to NIST in its contest for "quantum-resistant" encryption standards, and the people who submitted them.
here's one i found at random - david jao at the university of waterloo. djao.math.uwaterloo.ca some very mathy math there. i guess when you get down to breaking specific algorithms, maybe practical details of implementation and performance take precedence over theorems. but still really big-m Math, imvho.
 
10:18 PM
@leslietownes I've looked into a few of those papers now and I must say: Thank you very much for this! I thought that crypto only borrowed some algebraic structures and hardness assumptions but didn't really care about the structures themselves. Those thoughts have been pleasantly debunked.
 
10:49 PM
Currently I'm reading two books on the same subject. In book 1 they say that $\max(f,g)$ is measurable (where they say that $f,g$ have codomain the extended reals) and in book 2 they say $\sup(f,g)$ is (here, the codomain is simply $\mathbb R$). I wonder...what's the difference between $\max(f,g)$ and $\sup(f,g)$? I don't know really what to make of $\sup(f,g)$. Nowhere is this notation specified in the book.
I interpret $\max(f,g)$ to be $h(x):=\max(f(x),g(x))$, but I don't understand $\sup(f,g)$.
 
psie: absent any sign that it means other than what you would expect it to mean, sup(f,g) = max(f,g) in this setting.
 
yeah ok, that makes sense
 
it is more generally true that the supremum of a sequence of measurable functions is measurable. in this more general setting (taking a supremum over a sequence of values, not a pair of values) the supremum is not in general the max. maybe they proved that more general result somewhere?
or maybe the authors of the world's textbooks are just preoccupied with other things. if i had written the book, i might appreciate an email saying "btw, you never defined this" if indeed that is true. that is the kind of thing someone might put on a list of errata, at least.
 
yeah, if I had the email, I would email :)
 
@psie Functions $f:X\to\mathbb{R}$ are naturally partially ordered by $f_1\leq f_2$ iff $f_1(x)\leq f_2(x)$ for all $x\in X$. In this setting $\sup(f_1, f_2)$ is a function $x\mapsto \max(f_1(x), f_2(x))$. But $\max(f_1, f_2)$ doesn't have to exist. Hence notation $\sup$ is less ambiguous
 
10:59 PM
oh, i hadn't considered that they would be using an abstract-order-theoretic definition of "max," and not just doing pointwise max with values on the codomain.
i think we can all agree that the author still should have defined the usage
 
I might agree if I were to see the book
Just to double check if what psie is saying is true
 
yes, all of this is conditioned on an assumption that what psie is saying is true
for what it's worth, i don't find it particularly hard to believe
 
yeah. It happens all the time that people use notation they never defined. And its not very important anyway (to double check it)
 
i would like to see a competition for funniest list of errata. sometimes really good books have surprisingly long errata. particularly in the first edition
my advisor published a book with an exercise that was just a mess, and someone emailed him about it, and he put a fix in a list of errata on his webpage, and the fix wasn't quite a fix, and i forget exactly how i told him that he needed to fix an error in his errata
but it was probably the first time i heard my advisor say the "f-word"
 
fixing errors in erratas...sounds annoying :)
@Jakobian when is an instance that $\max(f_1,f_2)$ would not exist?
 
11:15 PM
@psie when $f_1, f_2$ are not comparable with each other
 
44 mins.......before voting can begin anew........
 
For example if $f_1 = f, f_2 = 0$ but $f(x)$ can be both positive and negative
 
@leslietownes You should say to em "LANGUAGE"
Then the whole class shouts, "LESLIE DON'T LIKE THAT KINDA TALK!"
 
as in $f(x) < 0$ and $f(y) > 0$ for some $x, y\in X$
 
hmm, but we are taking the pointwise max, right?
 
11:20 PM
no
 
How many Jakobians fit into the equation
1? 2? And also how many psies?
Like say, the variables are replaced with a Jakobian and a psie
 
@psie what I'm talking about is why notation $\max(f_1, f_2)$ is ambiguous and so I'm interpreting it as maximum in the poset of functions
 
ah ok, yeah $\max(f_1,f_2)$ is really ambiguous now that I think about it
 
What about a maximum number of @Jakobian or @psie?
 
I don't find those jokes to be funny if I'm going to be honest
 

« first day (5010 days earlier)      last day (11 days later) »