« first day (2339 days earlier)      last day (2980 days later) » 

06:01
Yes, a horizontal line has no tilt.
90° tilt for vertical
with respect to the horizontal line
makes still no sense haha
Or A vertical line has a tilt of 0 with respect to the vertical.
Positive to one side and negative to the other.
only positive would work too
mod 360
Don't be afraid of negative numbers :-)
They are just the opposites of the positives.
well, sometimes only positive makes things easier
06:06
Measured in the opposite direction.
must be a true pain to calc anything with degree celsius
Profit and loss too.
@Pissedofflayman with tilt it is fine
it is not ambigious what we mean with a 90° tilted line with respect to a horizontal line
Slope is measured with reference to the horizontal.
As having a value of 0.
Positive up. Negative down.
Vertical no value
since lines are infinite
06:16
Yup that too
in respect to lines, we really don't care wether they are vertical from the one side or the other
now in respect to vectors there are other means to shift their directions
matrices
Do you study math @Null?
not as hard as I wish
06:21
Meaning, at university I meant
yes
What year?
1st
1st half year to be precise
Ahh I see
So calculus and group theory?
analysis and linear algebra
06:22
What are you doing in linear algebra?
groups we had in linear algebra
So you know what a group is, order of groups, cyclic groups and group homomorphisms?
don't have much of an idea of calculus
cyclic groups, yes. group homomorhpism a little. tho i forgot what a group is. as far as i remember it is a binary operation on a set.
with some axioms it satisfies
Yes, closure, associa, identity, inverse
Group homomorphism just means f(a*b)=f(a).f(b)
group homomorphisms axioms are some, the most important would be identity goes to identity?
06:26
That's not an axiom^
ah ok, but a conclusion
really
should learn it, right now xd
You can show it in two/three lines
Do you want a hint?
nah, i proved it already sometime
so much to learn i forgot it
f(e_1)=f(e_1*e_1)=f(e_1).f(e_1) \implies
You'll forget it until you use it as a prerequisite
The top layer just decays away, so you build onto the top of it before it does
if f(a)=f(a).f(a), then f(a) is neutral for f(a)
in respect to .
06:32
So in a group, inverses exist right, so you can make it slightly cleaner (imo)
which is pretty much the definition for neutral element, i guess
$(f(a))^{-1}f(a)=f(a)$ (cancelation)
and then
1=f(a)
$1_H$
$f:G\to H$
yop
$$e_H=f(e_G)^{-1}\cdot f(e_G) = f(e_G)$$
That's much nicer
(imo)
i agree
06:36
Good work
And we've made use of the only property we care to remember for a group homomorphism, that is the $f(a*b)=f(a).f(b)$, so hopefully it's slightly easier to remember
this should be taught so much earlier
Yep, I think so too
For one thing it makes people less afraid of it
"Joe and Jane are making a group...."
like I would call myself of the upper half of educationed people. I really don't want to feel how it is to be in the lower half.
must be a horrifiying experience
user228700
06:53
Hi, everyone :-) I've a quick question. Range of polynomials = Real numbers, correct?
eh
user228700
Huh?
mmh
user228700
Huh?
i just mmh, don't know
can you prove it?
06:57
in The h Bar, 24 mins ago, by Sir Cumference
1. Ok, pick any number
in The h Bar, 24 mins ago, by Sir Cumference
2. Ok, now write out your number in english
user228700
Well, polynomials are defined to be expressions for which the variables can only ever have whole number exponents. So there is no way to get complex numbers. There's my answer, I guess. Nvm, then.
in The h Bar, 23 mins ago, by Sir Cumference
3. Count the number of letters in your number
in The h Bar, 23 mins ago, by Sir Cumference
Ok, now take the number of letters as your new number. Repeat steps 2 and 3.
in The h Bar, 22 mins ago, by Sir Cumference
If you keep doing this, you will always find yourself at 4, with an infinite loop, since "four" has 4 in it.
Challenge:
Given the string length function $sl(n)$ for any english name of any mathematical object $n$. It is known that $n=4$ is a fixed point. Prove or give cnounterexample that $n=4$ is an attractive fixed point for all $n$ for this algorithm.
@Kaumudi.H not necessarily
the range of f(x)=x^2 is [0,infty)
get some explicit bound, like sl(n) <= sqrt(n) + 20 or something, explicitly check for sufficiently small numbers
@Kaumudi.H Range of polynomials
Where are the coefficients from, are they real numbers, or complex numbers, and what are you evaluating at?
I.e. are you allowing polynomials 2ix^2+6x+i
You said you are evaluating only at integers, but if you have complex coefficients
07:03
is range the minimum codomain?
Range is the range, image is the minimum codomain
Range is whatever you want it to be, such that it contains the image
Well, the only thing I knew so far is $sl(n)$ seemed to be $\in \mathbb{N}$ hence finite
so range is the maximal codomain, if that makes sense?
It makes some sense,
07:04
(or can be any set that contains the image)
Yep
Any set that contains the image
If you have something with image $\Bbb R$, say $f(x)=x$ for $x\in \Bbb R$
actually screw that:
33
Q: Are there mathematical objects that have been proved to exist but cannot be described in words?

SecretThis might be a very stupid, and possibly philosophical question, but attempt to apply mathematics to everything plus inspired by this question caused me to ask this question Is there any mathematical object that has been proved to exist but cannot be described in words? If the answer is...

Then you can say the range is $\Bbb R$ or $\Bbb R\cup \{\infty\}$ or $\Bbb C$
Gregory Grant already had an answer
well, still, if $4$ is indeed an attractive fixed point, I suspect even uncountable sets will eventually converged to it by the algorithm...?
well, certainly the real numbers are not enough for all possible polynomials
07:06
But the image is a very specific thing, that is the set of outputs
Have you seen $R[x]$ before @Null?
@JackDon yes, but also $\mathbb{C}[x]$
so one would have to restrict the question in some way
otherwise there is no meaningful answer
Okay, so then, one could ask what the range is for evaluation of polynomials in $\Bbb Z[x],\Bbb R[x],\Bbb C[x]$
Exactly
Apples[x]
Hopefully Apples is a ring
@Secret I've seen a proof on reddit, showing 4 and -15 are the only fixed points
But since you can't have -15 letters, 4 is the only thing you can end up at
07:13
Well, that's true, but whether 4 is attractive or repulsive or neither makes a lot of difference. If 4 is not attractive, then there could exist $sl(n)$ that will never reach 4 after iteration
@Null Shush ;)
Don't ruin the magic
<- educated fool
2
It will be very interesting if given a mathematical object with an uncountable number of words (which an example is hinted in that question above) actually do end up at 4. That will mean 4 is a fixed point of infinite "radius(?)" (forgot the correct term)
even the term educated fool is not from me haha
What is sl(n) here, I thought this was the special linear lie algebra
07:15
O, we use that symbol in this context to mean the string length function
(That was caused my 'wut' above, I thought you put $\mathfrak{sl}(n)\in \Bbb N$ :P)
Sketch proof: Assume that whenever $sl(n)$ is unbounded (either countably infinite or uncountably infinite), we set $sl(n)=\infty$
Now $sl(\infty)=sl('infinity')=8$
Next $sl(8)=sl('eight')=5$
Next $sl(5)=sl('five')=4$

Therefore if $n$ has an unbounded number of words, the algorithm converges to 4 given the assumption.
that relies heavily on the spelling of infinity
So the question is then whether for all $n \in \mathbb{N}$ $sl(n)$ has an attractive fixed point at 4
why not take $\infty$ as one letter?
07:22
@Null well, you can argue there is a language $\mathcal{L}$ such that $sl(\infty)=\infty$. Then you have another fixed point, thus a counterexample
For english: 'one' does converge, because it maps to 'three' and then 'five'
ah
in pariticular, the fixed point $\infty$ in that above case is attractive for all $n$ with unbounded description, but not for finite objects
what is the thing you search? another fixpoint?
The aim is to prove that at least in the english language, the algorithm always converge to the fixed point 4 for any mathematical objects $n$.
because that will explain the magic behind sir cumference's example
or prove that no counterexamples exist?
07:28
or show that there is a counterexample that 4 is attractive fixed point for all $n$
I guess in terms of bounds as suggested by Mikemiller, one thing we can note is the trend: For all numbers $n \in \mathbb{N}$, $sl(n)$ increases much slower for each increase in $n$.
If this can be proved rigorously, then we will have the recurrence relation for large numbers $n > 100$ where $sl^{(2)}(n) < sl(n)$
and that will mean any $n > 100$ will end up in the set $\{0,1,2,\dots, 100\}$ which the results are knwon
you would need to start by giving a careful definition of how to express a number in the english language (e.g., how to write 10^n for arbitrary n)
also screw infinity, it will not help i think
Indeed, we can either stick to the illion system (which I think is going to run out after googol) or the english language description will be something that describes the operations e.g. "ten to the power of n"
the only thing that is certain is it is of finite length
so it may not be as easy as it sounds
one could read $10^n$ as 2+digits of n letters
 
1 hour later…
08:38
@ZachHauk what I meant was, you can easily check wether your website is reachable with a proxy site.
 
1 hour later…
10:06
Hello!! We have that $\mathbb{R}^{2\times 2}$ is a ring. We want to show that $K=\{\begin{pmatrix}1 & b \\ -b &a\end{pmatrix} \in \mathbb{R}^{2\times 2} \mid a,b\in \mathbb{R}\}$ with the matrix addition and multiplication is a field.

I thought to show that $K$ is a subring of $\mathbb{R}^{2\times 2}$ and it is also invertible is ths correct?
Hello @robjohn !! Is my idea correct?
Are you sure that "1" in the matrix isn't an "a"? Anyway your approach seems good
Oh yes, it is an "a".
Ok, thank you!! :-)
10:22
You might notice that this field is similar to something you know already at some point in this exercise ;)
Why can't the square root of 4 be -2 instead of 2, if -2 times -2 also equals 4?
@AlessandroCodenotti What do you mean?
@Ramanujan $\sqrt{4}=\sqrt{2^2}=\sqrt{|2|^2}=|2|=\pm 2$
@MaryStar why/how mod came?
@Ramanujan It holds that for $x\in \mathbb{R}$ : $\sqrt{x^2}=|x|$
Hey everyone, just a quick question "Find the order of the cyclic subgroup of $D_{2n}$ generated by $r$", wouldn't the order of that cyclic subgroup just be $n$?
10:37
Some findings from that algorythmic stuff I was doing yesterday, it's probably simple textbook stuff but it might be interesting, it was very interesting gradually axidently discovering it:
Two sequences can be convoluted together by a simplification of pretty much literally just multiplying the sequences as if there terms were the digits of two numbers, and when you code this up short sequences can be convoluted in one instruction!

Below, the convolution of (1, 2, 3, ...) with itself is revealed to be the tetrahedral numbers. The tetrahedral numbers entry on OEIS also asserts this.

On paper, you can use a simpler version of the usual multiplication heuristic, where carrying does not occur when adding, and the adding step at the end stops early.
This also occurs when you are multiplying any two polynomials, and it is one reason why I found multiplying polynomaisl so annoying, since convolution is a nonlocal function and hence depends on every single detail in the polynomial such as their coefficients, and make it very hard to visualise what each coefficient is doing for an arbitrary polynomail
If some kind of pattern can be found in this sort of convolution function, then it will become easier to predict where the roots of a polynomials are
Is it related to the (pascal triangle)/(n choose k) thing that happens with polynomials?
well, for powers of the same polynomial, by multinomial theorem you get the usual pascel triangle and its simplex generalisation on what the coefficients are. But what I am talking about is a more general case where two polynomials of different degree multiplying together
that is, all those cross terms that appeared whenever you expand something
if they actually obey some underlying rule, then predicting e.g. (a-b)(a+b) have no cross terms of the form ab or ba will become easier
Some long time ago, I asked the maths chat about a way to represent the discrete convolution as a matrix, but it does not help much because the resulting stuff is still kinda complicated
You could visualise the convolution they form like you would with a continuous convolution where it's much easier to imagine the resulting curve? Sorry if thats not that helpful.
Well, I think the gist is, since we realise multiplying any two polynomials (or as you have shown, sequences in general) is a discrete convolution, can more be learnt about this convolution function so that we can predict in advance that given two sequences with + and - terms, when you multiplying them together, which terms will vanishes
10:54
a = {1, 2, 3, 4}, b = {1, 2, 3, 4}, result = {1, 4, 10, 20}
In this case, does anything cross or vanish?
I guess vanishing would be getting a 0.
Hmm, I think I might be thinking about something different because $(1+2+3+4)^2=1(1+2+3+4)+2(1+2+3+4)+3(1+2+3+4)+4(1+2+3+4)=100$
'a' and 'b' above represent the polynomials your describing, and 'result' the resulting polynomial.
O, in that case, we will have a polynomial of degree ab. so:
e.g.
$(1+2x+3x^2+4x^3)^2=1(1+2x+3x^2+4x^3)+2x(1+2x+3x^2+4x^3)+3x^2(1+2x+3x^2+4x^3)+4x^3(1+2x+3x^2+4x^3)=\textrm{a deg(9) polynomial that I am too lazy to write out}$
but for this example there are no cross terms vanishing since all are positive
However consider:
$(-1-2x+3x^2)^2=-1(-1-2x+3x^2)-2x(-1-2x+3x^2)+3x^2(-1-2x+3x^2)$
ugh, I need to think of a better example. Give me a sec...
$(1-2x+4x^2)(1+2x+4x^2)=1(1+2x+4x^2)-2x(1+2x+4x^2)+4x^2(1+2x+4x^2)$
11:10
It's very complicated for me to understand already.
now e.g. the term $4x^2$ will vanish, but before expanding we would not knew
I get the central issue.
It's like a fractal
What I aim for in the future is some formulae that given any $a$ and $b$, it will give me which terms will vanish in the expansion before the actual expansion is carried out
If we wrote this convolution as a matrix (before every entry is summed up), it is clear there is some kind of symmetry within, but (at least for me) it is not sure how to better characterise that symmetry and convert that into a formula
Notice the red circled entries, $8x^3$ and $2x$ will go, and only one $4x^2$ will be left behind
It seems your polynomials have terms (seperated by +) in the form of ax^b, and the convolutions have terms (I wrote them seperated by commas) in the form of just one number per term.
oh, I get it, nevermind
The powers are the important part.
11:28
@Secret $$ \begin{array}{c|c} &1&2x&2x^2\\\hline 1&1&2x&2x^2\\\hline -2x&-2x&-4x^2&-4x^3\\\hline 2x^2&2x^2&4x^3&4x^4 \end{array} $$
Things cancel better :-)
Yeah, if only there's some universal formula that can predict this beforehand. Everytime when I try to help myself or others to check whether some proper rational function can be simplified, this problem pops up and I often have to multiply 10+ terms and get infuriated whenever I saw terms disappear that I don't previously suspect
Solving this problem might also have implication on factorisation, in fact
So what we knew is that whatever that answer is, lies in the property of discrete convolution
11:48
$$ \begin{array}{c|c} &1&2x&2x^2&\frac43x^3&\frac23x^4\\\hline 1&1&2x&2x^2&\frac43x^3&\frac23x^4\\\hline -2x&-2x&-4x^2&-4x^3&-\frac83x^4&-\frac43x^5\\\hline 2x^2&2x^2&4x^3&4x^4&\frac83x^5&\frac43x^6\\\hline -\frac43x^3&-\frac43x^3&-\frac83x^4&-\frac83x^5&-\frac{16}9x^6&-\frac89x^7\\\hline \frac23x^4&\frac23x^4&\frac43x^5&\frac43x^6&\frac89x^7&\frac49x^8 \end{array} $$
hmm... the matrix is not antisymmetric, but the terms that tend to cancel are antisymmetric entries in the matrix...
If we restrict the multiplication tables to the entries that give the antisymmetric terms, then we do have submatrices that are antisymmetric
not sure what that means...
Conjecture: Given a polynomial $P(x)$ and its square is asked $[P(x)]^2$. Then for any two corresponding pair of terms in P. If they are of the form (a,b) and (-a,b), then the cross terms vanishes
e.g. $(1+2x+2x^2+\frac{4}{3}x^3+\frac{2}{3}x^4)(1-2x+2x^2-\frac{4}{3}x^3+\frac{2}{3}x‌​^4)$
Pair 1: (1,2x) and (1,-2x). Property is fullfilled. Hence cross term of them vanishes
Pair 2: $(-2x,\frac{2}{3}x^4)$ and $(2x,\frac{2}{3}x^4)$. Property is fullfiled. Hence cross term of these vanishes
correction on typo: Conjecture: Given two polynomials $P(x)$ and $Q(x)$. For $P(x)Q(x)$, pick any two terms from each of P and Q. If they are of the form (a,b) and (-a,b), then the cross terms ab and ba vanishes
The next step is to wonder about a combinitorics problem. Given a polynomial P with n terms, the number of possible pairs of terms with no repetition can be picked from it is $\frac{{}^nC_2}{2}$ (halved to avoid double counting transpositions)
12:53
@alan2here so convolution is just multiplying the generating functions together
13:06
(just treat this as a commutative monoid for a second)
The structure is strange...), a bunch of transpositions
13:18
actually, this is an abelian group
but still, strange
hello, I was just wondering how can I simplify below expression:
expression below*
13:41
conjugation distirbutes over multiplication and addition, thus $\overline{(\overline{X1}(\overline{X2}+X3)})=X1\overline{(\overline{X2}+X3)}=X1‌​(X2+\overline{X3})$ Work from there
O_O
Pretty pictures
@Secret and no so pretty pictures...
That pretty picture is a 6 element abelian group (just treat those numbers as abstract labels) with some strange tranposition structure in it
Still trying to make sense of its generating set that give rise to that transposition structure propagating throughout the whole group
ok then.
@Secret so generally $\overline{X1 X2} = (\overline{X1}) (\overline{X2})$ is true?
no it cant be
yup. One can prove that easily with the polar form of the complex numbers
13:52
ahh
funny :D
actually im talking about boolean algebra
overline above means negation
O that... I am not very sure. I only remember there's demorgan's laws and negation is involutive
@Dogatemy hi :-)
14:00
So I'm thinking about permutations and stuff
Let's say that $(a,b)=(b,a)$, $(a,b,c)=(b,c,a)=(c,a,b)$, etc
if you identify them like that, it becomes a combination...?
so that cyclical permutations don't change them
but $(a,b,c)\ne(a,c,b)$ since that's not cyclical.
Ah I see
And we can plug these things into themselves, like $((a,b,c),(a,c,b))$.
So that one that I just typed is special, because any permutation preserves it.
So the questions are, what's the easiest way to do something like that with four letters, and is it impossible to do with five.
(Again for the next post, please just treat 0 and 1 as just labels, I am too lazy to change them)
If you restrict your stuff to just transpositions you get some abliean group like these, which fullfill your requrement that cyclic permutation don't change them?
14:06
Has identity, has inverse
@Secret What?
Commutativity
The things that preserve $((a,b,c),(a,c,b))$ are all permutations of the letters, so $S_3$, which is nonabelian.
ok, so your example need elements in $S_3$, I see...
But something like $(a,(b,c))$ is not preserved under transposing $a$ and $b$.
I'm thinking, for four letters, I have something much more complicated:
14:11
This is a nice question
Define $s_1=((a,b),(c,d))$, $s_2=((a,c),(b,d))$, and $s_3=((a,d),(b,c))$
Yeah that's the largange invariant thing
and then $((s_1,s_2,s_3),(s_1,s_3,s_2))$ should be preserved under all permutations of four letters
(s1, s2, s3) does it
oh.
yeah, agreed
@BalarkaSen Switching $a$ and $b$ wouldn't preserve that.
14:13
got it
I don't think it's impossible with 5.
I am thinking of stuff like this
But doesn't that just make it harder?
Actually that has 6 conjugates, so I guess
Assuming we can translate it to this problem (which I'm not sure because what's the analog of, say, $x_1+x_2+x_3$?)
@AkivaW So the most obvious thing to try is to get 4 permutations $s_1, s_2, s_3, s_4$ on those 5 letters which are preserved (not individually, but as a whole) under action of $S_5$. If you could do that, you could do it all as you know how to do it on 4 letters, by the same iterative thing. I think that'd give a surjective homomorphism $S_5 \to S_4$.
Which doesn't exist (kernel is a normal subgroup of order 5; there is no such thing)
Maybe that's not entirely correct. I think that's the circle of ideas you want to get into
14:35
Are the 6 conjugancy classes of $S_4$ $1_{S_4}, (ad)(bc), (dbc),(ac),(abcd)$ and $(ac)(bd)$?
Actually forget surjective; it'd give a homomorphism $S_5 \to S_4$ in any case. But kernel has to be $A_5$. That means the image fits in $\Bbb Z/2 \leq A_4$. So permuting the 5 letters would just produce at most 1 conjugate of $(s_1, s_2, s_3, s_4)$... what does that mean? shrug.
ok nvm, miscalculated, shoudl be 5
@MaryStar did you show that that's a field?
15:32
I have done the following:

$\begin{pmatrix}1 & 0 \\ 0 &1\end{pmatrix}\in K$

Let $A=\begin{pmatrix}a_1 & b_1 \\ -b_1 &a_1\end{pmatrix}, B=\begin{pmatrix}a_2 & b_2 \\ -b_2 &a_2\end{pmatrix}\in K$ then
$A-B=\begin{pmatrix}a_1-a_2 & b_1 -b_2\\ -(b_1-b_2) &a_1-a_2\end{pmatrix}\in K$
and
$A\cdot B=\begin{pmatrix}a_1a_2-b_1b_2 & a_1b_2+b_1a_2 \\ -(a_1b_2+a_2b_1) &-b_1b_2+a_1a_2\end{pmatrix}\in K$

From the above we have that $K$ is a subring.


We have that $\det K=a^2+b^2>0$, when $a$ and $b$ are not simultaneously $0$. So all the non-zero elements of $K$ are invertible.
Looks good
Now do you agree that if we consider the matrices with b=0 we get a subfield isomorphic to R?
What's the square of $\begin{pmatrix}0 & 1 \\-1 & 0\end{pmatrix}$?
15:52
@GFauxPas are you here?
yeah I am
just replacing your desmos graphs. nothing wrong with them, but I think we should have all graphs on the same page be from one software
agree?
I'm not talking about this
I don't mind if you replace the desmos graphs
sorry if I made you upset
I said I don't mind
I just found you for the generating function issue
because pm seems to have ignored it
I thought generating functions don't have to converge at all
15:55
Given a semigroup $S$ with possibly infinite elements , it is known that there is an element $a$ such that its left semigroup action on all other elements $x \in S$ either fixes it or map to a distinct element i.e. $ax=x$ or $ax=b$ where $b \neq x$. It is also known that a left inverse exists for $a$, thus $a$ is injective. How can I show it is surjective and hence the semigroup action $a$ and its inverse is bijective over $S$ and hence a permutation?
I am not talking about the convergence
though I agree with you
please respond to my point
what's the point of contention here
that it should be z/(1-z)^2 instead of 1/(1-z)^2
@DHMO this just shifts the power series
yes

« first day (2339 days earlier)      last day (2980 days later) »