up to you how you attempt to define the product of all elements in a subset of R, but however you do it, unless you break what finite products mean or the usual notion of convergence that gives you infinite products, it will probably fail to give you a result unless the set is at most countable and clumps in a certain way around 1.
complex analysis books sometimes do this a little bit.
you could say "haha if 0 is in there then the product is 0" but that isn't really engaging with what the notion of product is. sets containing 0 are also a special case for countable products.
haha I didn't think of 0 ! (not factorial). yeah, I understood that product of finitely many terms is well known; and the product of countable (infinite) terms is usually defined as limit (just as sum of countable many terms).
koro note that in the case of conditionally convergent series the notion of 'sum of all elements of a countable subset of R' will sometimes make sense with extra structure beyond the set itself (i.e., a linear ordering of that set) but not in general.
something like "sum S = sup {sum F: F subset S, F finite}" is only gonna give you finite sums for absolutely convergent series where the choice of ordering would not matter.
books in measure theory sometimes prove that or something slightly more general than that.
and that notion doesn't give you sums of multisets, e.g. sum {1,1,1} = sum {1} = 1 under that definition.
If V is a vector space and S is the set of all subspaces of V. Then considering the binary operation as 'sum of subspaces', what elements of S have additive inverse?
Sum of subspaces A and B is defined as the set {a+b, a in A and b in B}.
I think only the zero subspace has the additive inverse.
Hi :) I'm learning about hyperreal numbers $*\Bbb R$. Just as a preliminary question, in what sense does $\Bbb R(x)$, the set of rational functions of functions in $x$ with real coefficients, "instantiate" $*\Bbb R$?
In mathematics, specifically in commutative algebra, the elementary symmetric polynomials are one type of basic building block for symmetric polynomials, in the sense that any symmetric polynomial can be expressed as a polynomial in elementary symmetric polynomials. That is, any symmetric polynomial P is given by an expression involving only additions and multiplication of constants and elementary symmetric polynomials. There is one elementary symmetric polynomial of degree d in n variables for each positive integer d ≤ n, and it is formed by adding together all distinct products of d distinct...
@Shaun i looked at hyperreals a long time ago but don't recall this. where are you learning hyperreals from?
I'm teaching myself about hyperreal numbers. My main motivation for doing so is that they include infinite numbers, whose existence I hear disputed & doubted often as "quantifying the unquantifiable".
In this YouTube video, around twelve minutes in, it states that
$$\Bbb R(x)=\left\{ \frac{f(x)}...
Assume $\sum\limits_{n=1}^{\infty} \dfrac{1}{a_n}$ is a convergent positive term series and
$p>0$. Prove $$ \sum_{n=1}^{\infty} \frac{n^{p+1}}{a_1+2^pa_2+\cdots+n^pa_n}$$ is
convergent.
Since
$$a_1+2^pa_2+\cdots+k^pa_k\ge \sqrt[k]{a_1\cdot2^pa_2\cdots k^pa_k}=\sqrt[k]{a_1a_2\cdots a_k}\cdot \sq...
You need $\sqrt{|x|}$ to be in your set or else the transfer principle fails
and also $x^x$ and basically everything else
In fact, every function on the reals can be extended to a function on the hyperreals, no restrictions. This means we can extend, for example, the floor function $\lfloor x\rfloor$ to the hyperreals. The solutions to $\lfloor x\rfloor = x$ are called the hyperintegers.
Some infinitesimals $x$ will satisfy $\sin(1/x)=0$ and some will satisfy $\sin(1/x)=1$ etc. This (the fact that these are not infinitely close to each other) is the nonstandard analysis way of saying "$\lim_{x\to0}\sin(1/x)$ does not exist"
Do you know that infinitesimals satisfy this as opposed to just hyperreals in general? I.e., we have actual non-infinitesimals satisfying it, so how do we know actual infinitesimals do?
$\Bbb R^{\Bbb N}/{\sim}$ where $\sim$ is an ultrafilter
I think that's the best way to think about it. If $n$ is the equivalence class of $(0,1,2,3,\dots)$, then $\lfloor n/2\rfloor$ is the equivalence class of $(0,0,1,1,2,\dots)$
A hyperinteger is a hyperreal that's the equivalence class of a sequence of integers
The ultrafilter properties are basically exactly what you'd want for this to work. For example, $(0,1,0,0,1,0,\dots)$ has to equal either $0$ or $1$ (by which I mean it has to be equivalent to either $(0,0,0,\dots)$ or $(1,1,1,\dots)$)
so given any two subsets of the integers, the ultrafilter picks one and discards it
I should say "nonprincipal ultrafilter"
A principal ultrafilter would simply pick one of the components and say two sequences are equivalent if they agree on that component. Quotienting by this would give the usual reals
@leslietownes we had a lot of coyotes from January through March (breeding time), but the number of sightings has gone down dramatically since then. Now we have lots of bunny sightings.
robjohn: that's cute. we were still seeing bunnies here up until about a week and a half ago. but we've been seeing the coyote the whole time, so shrug
there's a large warren of thick bushes nearby that i think they live in, and if they can flee in there, a coyote wouldn't get 'em.
Can we express the distribution of a coordinate of the $n$-sphere in any known distribution?
In formal terms, consider $S^n = \{x\in\mathbb{R}^{n + 1}: \|x\|=1\}$ (i.e. the usual $n$-sphere). If we sample $x$ uniformly from $S^n$ what is the distribution of $x_1$?
By "sampling uniformly" I mea...
if it's the unit sphere, think about the area of the sphere contained in 0.9 < z < 1 vs. the area of the sphere contained in 0.0 < z < 0.1
there used to be a sci.math thread or faq or something, directed toward algorithmically generating uniform points on the sphere, that might have answered this question in as good detail or better than that
if indirectly (since its goal was not to answer that question)
one time in grad school we had a problem set with a slightly unclear problem, where the ambiguity was very difficult to notice. i think everyone in the class got slightly different solutions.
a slightly weird problem i ran into a week ago was along these ilnes
a common problem is to find the ground state of a hydrogen-like atom. the equation for that, modulo constants, is $-u''(r)+(1/r) u(r)=\epsilon u(r)$
small adjustment: replace $1/r$ with $2/r$. then the solution with smallest $\epsilon$ is given by $u(r)=re^{-r}$ with $\epsilon=-1$
the problem of interest here was to perturb that ODE with a term $f(r)$ which vanishes for $r<R$
the weird part here is: what exactly is the perturbation parameter?
the only one that really makes sense here is $R$ itself, i.e., an expansion in powers of $R$
but usually you can write the preturbation as $\lambda \cdot V(r)$ where $V(r)$ is order-1 and $\lambda$ is small. That description doesn't seem to work here
so the usual description of perturbation theory doesn't really make sense here
the ambiguity was something like, it was asking for coefficients of something in a basis, but not posed that way. and depending on how you read the problem there were at least two bases worth choosing, or you could come up with basis-free formulations of the problem that were much harder
which makes me wonder about the source i'm comparing with. they only worry about the angle $\theta$, and while that's not exactly wrong, it does make me worry they've managed to get the wrong distribution
(though the physics habit of writing expressions like $d \sigma/d\Omega$ always makes me paranoid)
yeah
differential cross-sections in scattering always make my eyebrow twitch
roughly speaking, i have a function $F=A+B\cos\theta$, where $\theta$ is the scattering angle of some process, and we're told that every scattering direction is equally ilkely
so I think that means that $F$ should be uniformly distributed over $[A-B,A+B]$
the solution i'm looking at gets a much worse answer by a more circuitous route
My teacher had a Ph.D. (in mathematical logic) for what it was worth. She was not a good teacher — she punished me for my curiosity and natural questions. She also chose to stop teaching the calculus course because she wanted to teach students who needed her (laudable ... but ...) and then told one of those "regular" classes that she didn't expect them to understand X because they weren't honors.
That made me livid. It's OK to tell students stuff is hard and it's ok to say that you don't have to understand it completely. It's not OK to say "you're too stupid for this."
i had a number of unpleasant teachers in hs, but none who were academically pompous. they'd make you hate them for reasons unrelated to your abilities or performance in the subject matter
Well, most of our math teachers were dumb as rocks and coaches. The guy that took over the calculus class — whom I'd had junior year for math analysis — was a nice guy, but he literally had never had calculus in college many years earlier. So that was a disaster. He dismissed me and two other people and we did it on our own. Ultimately, I came back and taught the class for the last month or so.
This was 1969-70, before they had AP classes, but I took the AP BC test. It even had a $\delta$-$\epsilon$ proof on it.
I don't remember doing any practice tests or anything. Those were the days before "civilization."
it also seems different in college, where students have more time and room to plan a course of study. there are many options and figuring out what works is part of it.
I did get annoyed with a very bright student who skipped almost all my Calc w Theory classes and did well. I didn’t ban him from office hours, but I should have. When he got to diff geo a few years later, I warned him and it didn’t go well.
i think hs teachers obviously should help students meet/exceed graduation requirements with a view to what they might do next, but it isn't their place to decide what those next steps are or aren't going to be. if someone is underperforming that is an advising issue for planning next hs steps, not a "the door to X future is closed" issue.
and obviously fine for hs teachers to tell kids that what they're getting away with in hs might not work in college. we heard a lot of that.
although it does raise the question of how they're able to 'get away with it'.
i guess now a lot of formulaic curricula can take the blame and not the teacher.
in courses without a lot of check-in mid semester (or where students can copy assignments freely with one another), not attending classes can make it difficult to spot disasters before they happen. which can waste a lot of the student's own later time and money, even if they're at fault.
this may be US-specific. in other jurisdictions it might not waste much of the student's money.
i would also distinguish between expressly penalizing students on the basis of attendance (which i think is counterproductive, although is not uncommon in the US) vs. attempting to intervene in nonattendance to avoid potential student disasters (which ought to be done).
some students just drop off the face of the earth, get an F on the final, and you get an email "can you regrade that? i need at least a D to graduate." and taking another unexpected term means another X months of unplanned rent + enrollment expenses, or interfering with some job. well, whoops.
best to catch that stuff early, but you can't always catch it if they aren't there.
i wonder how all of this worked out with remote learning during the pandemic. my wife has horror stories about this.
i can guess. europeans who come to the US often say, why is there homework. why is there attendance. if you can't do the exams, then go away. there are a million competing explanations for why that hasn't been the system in the US.
Whenever I attend a class, either the lecture was "easy" as hell and I wasted a lot of time, or there were crucial points which I didn't understand and the professor just skipped it, the result was to get lost and not understand anything from that point on
The most productive thing for me was to just self-study, so I never set foot on lectures ever again
similar with grade inflation. one of the competing explanations, as hinted above, is the relative expense of college in the US. there might be others.
in the 60s being in college could mean not getting sent to vietnam. some people locate the beginning of a lot of what we might think of as an intensified paternalism to that era
colorful: i'm exaggerating. the idea i get from some europeans is "why does it matter if the student doesn't understand the material, they're the only one affected by it."
if people in my classes were doing fine on midterms and such i didn't care where they were, but if people would not show up at lecture and get middling grades, i'd usually schedule at least one meeting and say, hey, you're on track to get X grade, is that consistent with your expectations, and would a lower grade throw them off,
and if not, attending lecture was always part of the recommended solution
I also spent 7-10 office hours a week with a largely full office, helping students. Also not the European model (or the US model at most research universities).
Exams are just to check you’ve learned the basics, not to push the challenging boundaries. So they were just a part of my courses.
Hmm I see, idk how much the homework is worth, but in my universities it's about 20%. We have the choice of "exam counts for 100%" or exam counts for "80%" (+ the 20% homework)
i sometimes used formulas where a lowest midterm exam score could sometimes be adjusted up if there was improvement on the final, but HW was always a pretty big chunk.
for people who don't find HW annoying, they really like it, because it's a piece of the grade they have the most control over
I really like homework because it provides me with a lot of problems. But I consider it unfair to have it to count towards your final grade without having the choice not to. My argument is that, for some students (like me), it's not always possible to submit the homework on time, and we have better lack to show our skills long term, i.e. in the final exam, not weekly.
I have a serious case of OCD and sometimes I can't study for like 2 weeks
And then for 1 week I will study 12 hours per day to cover everything
By the time I've caught up, I've missed like 3 homeworks
yeah. one thing i noticed a bit when teaching was that substantial homework could be very tough on people who had a lot of other stuff going on, but easy for people who didn't. it wasn't always measuring what you'd want it to measure.
someone might be doing well on the homework just because that class is the only 'real' class they are taking that semester, even if they are spending a worryingly long time to do it. and busy people could miss assignments or not just get stuff done, irrespective of understanding.
it's hard.
similar thing with exams, too, limited time vs. take home. can be really unfair depending on someone's final exam calendar, which at most schools is random.
the way to solve this at the upper levels is just to give everyone As and have grades not matter. but you gotta do something to filter and provide feedback at the lower levels. at least i think so.
some of my wife's friends think that colleges shouldn't give grades. it might be easier to do that in a field that isn't math or physics or something that needs a lot of prerequisites to jump into.
i don't have any conception of what understanding makes a C in a history class vs. an A in a history class, for example. but with any of the sciences or engineering, i can form a pretty good guess.
Somewhat unrelated, I wanted to ask experienced mathematicians if the following is considered "normal": I don't know basic linear algebra (doing calculations) but I know "advanced" algebra (proving theorems etc). For example i just started learning Lie Algebras and I'm doing fine, but I'm not familiar with properties of matrices, eigenvalues etc
And I think it's important to note that I'm a theoretical physicist
that seems pretty normal to me. if the 'calculations' you don't know are things like matrix diagonalization, eigenvalues, whatever. you can do a ton of stuff without the calculational side coming into play.
and indeed in physics sometimes it doesn't even help
it was pretty common in my math grad program for people never to have gotten the 'calculational' version of an elementary subject because they never had to take it in college.
one time my officemate asked me something, and it was basically, what is the rank of some 4x4 matrix. and i said "can't you just compute it?" and he didn't know how to do it from the matrix. he was trying to realize it as the dimension of something that he understood better. i'm not kidding. he knew what the rank was conceptually, but had never learned the algorithm, row operations, anything.
he picked it up very quickly, of course, and would have had to teach it if he had been a TA in the right class. but he hadn't done that.
colorful: i wonder if that depends at least somewhat on the school? my impression of physics undergrad in the US is that it varies widely in mathematical sophistication, sometimes depending on the interests of the faculty or size of the department. even at research schools.
this would have been a while ago. like the hardy-littlewood era. i think both of them only had masters degrees, and this was not unusual for that time.