« first day (2847 days earlier)      last day (2168 days later) » 

11:00 AM
what are you trying to do exactly
 
Well, I'd like to be able to figure out which root of unity I need to multiply by in order for the result of the simplification to be correct.
 
maybe you should look at the arguments of everyone ?
 
That's certainly the right idea. I understand that the source of discrepancies like The simplest example of where it comes up: -1=(-1)^(1/2)*(-1)^(1/2) while (-1*-1)^(1/2)=1 is that the "identity" treats arguments that differ by an integer multiple of 2pi as the same, while the definition of nth roots for complex numbers treats them as different.
The trouble is that I need to be able to handle arbitrarily complicated radical expressions, and I don't know if I'll be able to explicitly calculate their arguments in general. I have considered using numerical approximation for the arguments.
My understanding is that the case where I need to worry about multiplying by a root of unity is when the radicands of the two surds I'm multiplying have arguments that sum to 2pi or greater.
I could mostly get by with approximating the arguments, but if a complicated expression happened to have an argument of zero in a nontrivial way, I don't see how I could ever verify that for sure using approximation. I'd only know that it was either between zero and some small epsilon or between 2pi and 2pi-epsilon.
 
11:18 AM
so your problem is that your principal nth root function is discontinuous
 
There's no avoiding that with complex numbers, is there?
 
that depends on what you really want to do
 
Something I might not have a rock solid grasp of. Could you give an example or two of what one might want to do where the answer to that question goes one way or the other?
I understand that things would get much simpler if I could settle for considering nth root extraction a multi-valued function, but I don't think that would be practical in the context.
Which, for what it's worth, is a programming exercise where the user can type in an arbitrary radical expression and my program does its best to simplify the expression. I premise that leaves the treatment of complex roots ill-defined, I know, but whatever I choose should behave in a way that is as familiar to, say, a high school algebra student as it could be while still addressing the full generality of the problem.
 
that sounds like a pretty difficult thing to do
 
11:33 AM
There's a lot to it, for sure. I've been working for a while now on finding the minimal polynomial of any expression, which is necessary for denominator rationalization among other things, but this complex root problem has been a gaping hole in my foundations all along.
 
@mercio, is (D) the correct answer? How do we decide between 'and' / 'or'?
 
well, can $a$ be twodifferent things at the same time
 
It may be that the way I build up expressions will ensures that no expression can ever by non-trivially equal to zero...I would certainly hope for my algorithms to have that property.
 
how much galois theory have you done
 
@mercio so, that's why d is correct. thanks!
 
11:41 AM
no that's not why d is correct, it's why c is false
 
I find the wording ambiguous
 
The abstract algebra I took in college just barely got to it at the end. I'm not comfortable with it, but know broadly what it involves.
 
you poor soul
 
sorry :)
 
It depends on $a$. It should be: what is the good end to make a valid general statement.
 
11:43 AM
well it doesn't specify where $a,b,c,d$ are taken from
 
The programming project, at least, has been self-imposed! I figured it would make a good non-trivial piece for a programming portfolio.
 
It depends on $a, b, c, d$ *
I think* the wording is ambiguous. Anyway...
 
It sounds like I'm not just being stupid and missing something obvious. Not as helpful as a solution, but still helpful.
 
@AlexKindel do you know the cartan formulas for roots of cubic equations
imagine if i start with, say $x^3+x+1 =0$
compute the minimal polynomials of $x^2+x$ and $x^2$
use the cartan formulas on those three polynomials to get a huge mess
then innocently ask you to prove that $(x^2+x) = (x) + (x^2)$
where the three things are replaced with their respective huge messes
 
Ell
Hi folks
 
11:51 AM
would your program be able to handle that
hello
 
Is this the equivalent of the quadratic formula for cubics?
 
yeah
 
The program only handles inputs from the user that are constant expressions, though it uses polynomials internally.
 
I mean, this would end up with a huuge complicated expression with square roots and cube roots of integers, possibly nested
 
Oh, you mean would it be able to manipulate the answer produced by that formula. Assuming I can solve the roadblock or two I'm dealing with currently, yes.
 
11:56 AM
Is there a solution using only dot products to $A + sB = tC$? (I.e I want to figure out s so that $A + sB$ is parallel to $C$? (A, B, C are vectors)
 
actually, how would it fare against the classic $\sqrt[3]{5\sqrt 2+7} - \sqrt[3]{5\sqrt 2-7}$ ?
 
12:12 PM
\o
 
I would test whether either of the terms is in the field consisting of the rationals extended by the other term, and if it is, find a linear combination of powers of that other term equal to it. I'm not quite enough of an algebra whiz to see it directly, but I suspect that would suffice t find the simplification?
 
but how do you check if they are in the same field ?
even determining $\Bbb Q(\sqrt[3]{5\sqrt 2 + 7})$ is nontrivial
 
plouffe.fr/simon/math/… Algorithm 4.5.4 and all its prerequisites, pretty much.
 
my internet sucks
 
that's a good book
 
12:22 PM
The book doesn't cover how to produce the minimal polynomials necessary as far as I can see, but I think I can do it.
I know the minimal polynomials of rationals. Given a number whose minimal polynomial I know, the minimal polynomial of its nth root is a factor of the radicand's minimal polynomial with the degrees of each of its terms multiplied by n. I can factor rational polynomials, and I can test which factor is the annulling polynomial using numerical approximation (I don't know if there's an algebraic way).
If I can find a number's minimal polynomial, I can rationalize it, so that takes care of quotients. That just leaves sums.
For those, I can apply the field membership algorithm for each term added, and ultimately get back a linear combination of powers of some number whose minimal polynomial I know, which is also a thing I know how to find the minimal polynomial of.
The original complex root question, incidentally, was partly motivated by the desire to never have to deal with products - given multiple surds multiplied together, I'd like to consolidate them into a single surd.
 
o..o'
how do you skip the products ??
 
12:38 PM
Like, if I have 2^(1/2)*3^(1/3) or something, I'd replace it with 72^(1/6).
For products of sums, distribute.
This is that a^(1/n)*b^(1/n)=(ab)^(1/n) identity I wanted to be able to extend to complex radicands.
 
you should probably just keep numerical approximations to everyone
 
If I can't do that extension, I can't eliminate products after all. I would still have my ways of getting the minimal polynomial, but I would be missing out on some very basic simplifications I feel I ought to be able to do.
And yeah, it turns out that at least as far as the Cohen book mentions, the last step of solving the primitive element problem strictly requires approximations, so I've been adding all the infrastructure.
When I got to the point of computing the argument of a sum is when the ambiguity with respect to complex roots crisis hit.
 
can you detect if some expression is a real number ?
 
I don't know of a way to do that, but that would solve my problem.
 
you probably should
you take its minimal polynomial
and uuh
if you see a root near the real axis
if it's simple it is real and if it's a double root then it's complex ?
:s
 
12:50 PM
I can certainly already tell if a number is almost real.
 
(because if you see a double root near the axis then it has to be a pair of complex roots with very small imaginary parts)
 
Oh, does the thing that holds for roots of unity where complex roots come in pairs reflected over the real axis hold for roots of minimal polynomials in general?
 
if you keep minimal polynomials over Q, yes
 
Probably a foolish question to anyone who knows Galois theory better than I do.
 
the roots of a real polynomial are always real roots and a bunch of pairs of complex conjugates
 
12:54 PM
So yeah, I don't think I should have any roadblocks to getting arbitrarily accurate approximations of anything I want, and given that, I should be able to determine realness.
I might also need to determine whether a number is purely imaginary. I guess I could multiply it by i then see if it's real.
 
yeah
or square it
and check if it's a negative real
 
hello, so a staff ask me a great maths questions
there are 36 students, and during the ice breaking, they must meet each other. But for some reasons, there can only be 8 people at 1 group, what is the minimum numbers of groups will be formed?

simplify it, the rules are:
1. all 36 students must meet each other
2. only 8 students in a group
i've tried it, first i thought fixing one person, like 1 _ _ _ _ _ _ _, then get all possible from 1 to 18, but there are way many groups
 
Squaring could potentially be expensive, since it might trigger solving the subfield problem multiple times. :D I have n idea how slow all of that's going to be once I have it working.
 
then i tried to separate 36 students into 9 subgroups, each has 4 students, and i get the minimum from here (36 groups total)
i wanted to know is there any theory/theorem related to this sort of questions? and is 36 the minimum?
 
idk how the subfeild problem works
i would expect a lot of time spent on factoring very big polynomials
 
12:59 PM
Yeah. You have to factor a polynomial over a number field, which in turn involves factoring rational polynomials.
 
well multiplying by i can multiply the degree of the minimal polynomial by 2
while squaring doesn't
 
True. I suppose I could test both ways and see what performance difference there is.
 
also you don't need to factor the result completely because you can detect multiple roots with gcd(P, P')
 
Though whether, given the minimal polynomial of a number, there is a direct way to get the minimal polynomial f that number times some root of unity is something I've wondered abut.
If there isn't, I think I'll technically still have to deal with products when calculating minimal polynomials, because using complex exponential representations of roots of unity is the only way I know to distinguish between nth roots in the general case.
Like, if I need to represent one of the roots other than the one I've decided the radical sign denotes.
 
1:22 PM
@Sawarnik Hey, how's it going (didn't read the deleted messages)
 
@mercio Anyway, thanks for the help. This is my first time in a chat, and it has turned out to be far more approachable than the forum, and probably more fruitful too.
 
you're welcome
 
2:18 PM
Good morning!
Let $I$ be an interval in $\mathbb{R}.$ For $k \in \mathbb{N}$, let $g_k, h_k \in L^2 (I)$ such that
$$\sum_{k \in \mathbb{N}} \left\Vert g_k \right\Vert_2^2 < \infty \quad \text{and} \quad \sum_{k \in \mathbb{N}} \left\Vert h_k \right\Vert_2^2$$
Define $K : I \times I \rightarrow \mathbb{R}$ by
$$K(x,y) = \sum_{n \in \mathbb{N}} g_n(x)h_n(y)$$
and $T : L^2(I) \rightarrow L^2(I)$ by
$$Tf(x) = \int_I K(x,y)f(y)dy.$$

Suppose that there exists $C \in (0, \infty)$ such that
$$\sum_{n \in \mathbb{N}} \left| \langle f,h_n \rangle \right| ^2 \ge C\left\Vert f \right\Vert ^2 \quad (\forall f \in L
I know that when the nullity is zero, then it is injective. By the Fredholm alternative, then it is injective if and only if it is surjective, thus, $I - T$ is bijective.
I guess the help I need is by showing that $\mathcal{N} (I - T) = \{0\}.$
 
3:14 PM
[Random]
Generalising square root operators.
Let $S$ be some algebraic structure obeying some identity
$f(a,b)=c$
We are interested to find $f^{\frac{1}{2}}$ such that $f^{\frac{1}{2}}\circ f^{\frac{1}{2}} = f$
 
Hello, if $E_1\times E_2$ is complete how to prove that $E_1$ and $E_2$ are complete?
 
I wonder the fact that fractons cannot move until when other bits are together will help inspired this via abstraction:
That is for example:
 
@Daminark hi
@PolineSandra define complete
 
$a^2 = 0$ is nilpotent, thus $ab$ does nothing, but $a^2b = 0$
$a^2 = 1$ is idempotent, thus $ab$ does nothing but $a^2b=b$
 
every Cauchy sequence converge @LeakyNun
 
3:19 PM
Hi. A challenge exercise: Let $A$ be a convex compact set of a finite dimensional space $E$. Let $f$ be a linear function such that $A$ is stable by $f$. Show there exists $x$ in $A$ such that $f(x) = x$ @LeakyNun @mercio
 
@PolineSandra use the inclusions $E_1 \subseteq E_1 \times E_2$
 
@Secret @LeakyNun do you have any ideas?
 
That's beyond my knowledge coverage of bounded linear operators thus unfortunately no
 
are you sure that $E_1\subset E_1\times E_2$? @LeakyNun
 
@Secret a fun exercise is to define a "square root" as a section $s$ of the map $\cdot^2\colon k\to k\cdot k\colon x\mapsto x^2$ such that $s(ab) = s(a)s(b)$. What can you you say about the existence and uniqueness of square roots for different fields $k$ like $\mathbb R,\mathbb C, \mathbb Z/pZ$?
 
3:21 PM
@PolineSandra as long as $E_2$ is nonempty
 
@anakhronizein I cannot quite parse $\cdot^2\colon k\to k\cdot k\colon x\mapsto x^2$. Are you saying the map is $\cdot^2 : k \to k\cdot k$ and then $k \cdot k : x \mapsto x^2$?
 
After the second colon is the transformation rule
So x-->x^2
$k\cdot k := \{x^2 \mid x\in k\}$
Poor notation, but I figured the context of the map was enough
 
Ah I see
 
3:37 PM
I wish Prof Ted Shifrin were here
 
Hi
0
Q: A nontrivial solution for $ f(f(x)) = exp(exp(x)) $

mickConsider the equation $$ f(f(x)) = exp(exp(x)) $$ Valid for all real $x$ , $f(x) \neq exp(x)$ ( not identically equal everywhere ) , $f(x)$ is analytic and $$ f(x) = a_0 + a_1 x + a_2 x^2 + a_3 x^3 + ... $$ Where $a_1,a_2,a_3,... > 0 $. Find solutions for $f(x)$. —- Remarks : I wonder ab...

 
Every fields has identities 1,0
$s(1)=s(11)=s(1)s(1)$
$s(0)=s(10)=s(1)s(0)$
$s(0)=s(00)=s(0)s(0)$
$s(0)=s(x0)=s(x)s(0)$

So $s(1)$ is idempotent, and $s(0)$ is an absorber. This means $s(1)=1$ and $s(0)=0$ thus $s$ is a homomorphism.

$s$ thus exists for $1,0$ hence existence is guarenteed. For these cases, $s$ is also unique
As for generic $x,y$, need to check...
 
I don't think that proves it existence.
Nor does it prove uniqueness.
 
ok maybe not uniqueness, but existence has to, because we always have $s(1)=s(1)s(1)$ for any field since every field has the unique identity 1, thus $s$ is well defined for at least the identity?
 
That doesn't mean s has to exist.
 
3:49 PM
hmm...
 
Hi, is a curve which has a cut somewhere a submanifold? For example a parabola on $[-1,1]^2$ in $\Bbb R^2$ which just ends at the points $(\pm 1,1)$.
 
So like a parabolic arc? No, it's not a submanifold. It's a manifold with boundary embedded in $\Bbb R^2$.
It's a submanifold at every point other than the endpoints of the arc.
 
Yeah like an arc with finite length. Does this mean a set needs to be open to be a submanifold?
Since excluding the endpoints would make it a submanifold?
 
No, the full parabola $y = x^2$ is not open, eg. But yes, the open parabolic arc is a submanifold.
 
Hi
 
3:59 PM
Okay, thanks. But why is the full parabola not open?
 
@Secret try it for C in particular.
 
Yesterday there was a good long Mathematica answer for this quesiton: math.stackexchange.com/questions/2789094/…
 
And see what you find out.
 
I can't find it anymore, do you know if it has been deleted?
 
You are right that it would have to be a homomorphism, but what does that mean with respect to the other properties?
 
4:00 PM
@philmcole Well, the open parabolic arc is not open as a subset of $\Bbb R^2$ either :D
It's just homemorphic to the open interval $(-1, 1)$.
For the full parabola $y = x^2$, that's actually a closed subset of $\Bbb R^2$ because it's the zero set of the function $f : \Bbb R^2 \to \Bbb R$, $f(x, y) = y - x^2$. $f$ is continuous, so $f^{-1}(0)$ is closed.
 
Oh it's not open because it's not "thick". If it would be like a band without its edge, then it could be open, right?
 
hmm... $i^2 = -1 = (-i)^2$
 
@philmcole Correct.
 
Okay thanks! :)
 
The open parabolic arc is a neither-open-nor-closed subset of $\Bbb R^2$.
 
4:03 PM
okay
 
vzn
@AlexKindel a very ambitious project! the author says at the beginning he has implemented nearly all algorithms in the Pari software pkg. think that may be a promising angle/ direction to go in afaik its still actively maintained.
 
I'm going to work on some algebraic topology while listening to this Agalloch album.
I have never heard that band before
 
Anyone, @BalarkaSen?
 
Any recommendation on algebraic topology textbooks for self-study? Im taking a course in the fall and would like to look at some stuff over the summer
 
It's not bad, @BalarkaSen
Thanks for showing
@NicholasRoberts I think Greenberg & Harper is not too bad for self-study
 
4:07 PM
@anakhronizein "It" being the band or the album? :)
 
It's not exactly modern. T
 
Have you tried Hatcher's book, @NicholasRoberts?
 
@HarryEvans I don't understand your question.
 
The album in particular, but the band by corollary. @BalarkaSen
 
I have looked at Hatcher, i dont really like the wording. Plus, the physical copy has a lot of errors im told. And i Dont want to read off the PDF
 
4:07 PM
@anakhronizein Ahh, nice.
 
I will look into that @anakhronizein
 
@NicholasRoberts are you good with category theory?
 
@NicholasRoberts Sad, man, really sad. Hatcher is life.
(I consider myself to be a Hatcher elitist)
 
@anakhronizein I have basic knowledge of category theory. Co-contravariant functors, universal initial/terminal objects, natural transformations. Thats about it
 
I posted it above, hoping you could answer, @BalarkaSen. Just in case you don't see it: math.stackexchange.com/questions/2790125/…
 
4:09 PM
@BalarkaSen Lol
 
0
Q: Using the Fredholm Alternative to Prove that a Transformation is a Bijection

Harry EvansLet $I$ be an interval in $\mathbb{R}.$ For $k \in \mathbb{N}$, let $g_k, h_k \in L^2 (I)$ where $$\sum_{k \in \mathbb{N}} \left\Vert g_k \right\Vert_2^2 < \infty \quad \text{and} \quad \sum_{k \in \mathbb{N}} \left\Vert h_k \right\Vert_2^2$$ and $$\left\Vert g \right \Vert_2 = \bigg [\int_{I...

 
@HarryEvans Ahh, I don't know any functional analysis.
Maybe try @Daminark?
 
If you would much rather a more category theoretic approach, there is either Spanier (more concrete) or May's "concise" notes.
 
Hmm, perhaps. I will check it out
 
Thanks, @BalarkaSen...I wish Prof Ted were around
 
4:12 PM
@anakhronizein Will the greenberg book go into simplicial complexes, triangulation, k-chains? stuff like that
 
Actually I don't think Greenberg & Harper do simplicial.
If you want simplicial, try Munkres.
(his algebraic topology book, not his topology book)
 
Ok, cool
Ya, i was looking at the table of contents and didnt see any of that
 
Greenberg & Harper do: fundamental group & covering spaces, singular homology (chains complexs are inherent to homology in general), cell-complexes, orientation and duality on manifolds, and products and the Lefschetz fixed point theorem.
 
Munkres is a good author on Topology; I haven't seen his Algebraic Topology book, though
 
Hello everyone
I have one question
 
4:15 PM
I like singular better than simplicial, but some people like simplicial better (for some reason!)
 
What monthly instalment will discharge a debt of $2455 due after one year at 5% per annum simple interest?
 
Daminark will announce that he likes $[X, B^nG]$ better, surely.
 
Any hint please
 
Will @Daminark be here? Or @TedShifrin?
 
@Faust can you help
 
4:18 PM
whens the intrestrest compounded
yeah
year*?
0.05 x= 2455
there
 
@anakhronizein Simplicial is where the intuition lies. Also you can build a simplicial model of a given topological space and transfer the whole singular theory there: Take a top. space $X$. Build a simplicial complex $S(X)$ whose vertices correspond to points in $X$. For every singular 1-simplex $\Delta^1 \to X$, attach a 1-simplex to the 0-simplices in $S(X)$. For every singular 2-simplex $\Delta^2 \to X$, attach a 2-simplex to the appropriate skeleton.
At the end, you're remained with a simplicial model $S(X)$ which is weak homotopy equivalent to $X$.
 
Hi @Faust
 
Morning
 
$H^{sing}_\bullet(X)$ is tautologically just $H^{sim}_\bullet(S(X))$
 
good luck visualizing singular homology :P
 
4:22 PM
Well, it can be done.
 
not by me :P
induced maps on singular are noice though
 
The idea is that singular $k$-chains can be realized as maps from certain simplicial $k$-complexes to $X$.
Kreck has a book where he develops a totally geometric homology theory where the $k$-chains are maps from $k$-pseudomanifolds-with-boundary to $X$. Mike loves that a lot.
 
hm interesting
 
@BalarkaSen ryan budney is insane
 
4:28 PM
Mike showed me Kreck's book.
It's good.
 
@Faust Yeah lol
 
@BalarkaSen i'll take a look later, thanks for the link.
 
@BalarkaSen hes a prof at my university worst teacher ever :\
 
looool
 
aaw
 
4:29 PM
I have found him to be frequently helpful in MSE and MO
Great guy
 
i like the guy and don't get me wrong hes a genius but if your not a smart grad student don't bother asking him a question
 
I guess good research mathematicians can be bad teachers tho
Yeah I know what you mean
 
Well it depends. Good mathematicians are good for either being good teachers or good researchers.
 
i can understand some of what hes saying but most people that ask him a question end up in a staring contest with him
 
@anakhronizein Edited, yup.
 
4:31 PM
This is probably a very dumb question, but I have noticed that several mathematicians, famous mathematicians, studied things that has their name assoicated with it. Like Riemann studied Riemann sums, Gauss studied Gaussian curvature, Euler studied Euler gamma, and so on. Is it a coincidence that those mathematicians studied functions and concepts that bear their name? Like I mean are mathematicians drawn to study things that already has their name associated with it?
 
its hilarious to watch him teach 1st econd or even third year classes
@MatsGranvik stop trolling
 
oO
 
lol the names were associated to those objects after they studied them
honestly can't tell if a troll
 
Sounds like it is straight from 4chan
 
lmao
4chan/sci/ does have some gems
 
4:52 PM
Why are all the hbar people invading here
what are you doing in my swamp
get out of my swamp
 
Yer. I sit here silently and sometimes try people's maths problems without comment when I get bored. My land now.
Such an invasion
**strains hard not to make joke about invasions**
 
5
Q: A nontrivial solution for $ f(f(x)) = \exp(\exp(x)) $

mickConsider the equation $$ f(f(x)) = \exp(\exp(x)) $$ Valid for all real $x$, $f(x) \neq \exp(x)$ (not identically equal everywhere), $f(x)$ is analytic and $$ f(x) = a_0 + a_1 x + a_2 x^2 + a_3 x^3 + ... $$ Where $a_1,a_2,a_3,\cdots > 0 $. Find solutions for $f(x)$. —- Remarks : I wonder ...

 
@BalarkaSen You know if the mod message about celly is out yet or is that for celly only?
 
@mick is that you in your profile picture
 
You asked me b4 @mercio
 
4:54 PM
really ?
 
Yes
 
how many years ago was that lol
I don't remember
 
4
 
@CooperCape It's been sent to him, yes. I know the content of the message, but can't publicly communicate it, I don't think.
 
@BalarkaSen Alright, no problem :p. Was just intrigued.
 
4:58 PM
If you read the Physics Meta room you'll see that the central moderation philosophy is now to measure size of the raw suspension log
If someone has been suspended a lot (auto-ban and mod ban alike), they're under the radar.
 
I see... I started reading but then realised that many other things in my life were more important.
 
Lol same
I've given it more time than I should have
 

« first day (2847 days earlier)      last day (2168 days later) »