« first day (17 days earlier)      last day (30 days later) » 
14:00 - 22:0022:00 - 00:00

22:00
Well, for starters, when you want to take a limit, obviously you want each integral to converge (or go to zero if possible)
Secondly, there are well-known choices to choose from. Pacman/keyhole, semicircles, arcs, rectangles, etc.
We also know where the poles of the function are, which helps as well. The branch cuts, for example, are not crossable. Hence, we avoided the positive real line by using the keyhole
You know, obvious indicators that come with practice/trial-and-error
Ahh could you go through a simpler example with me
???
But... how simple?
More simpler than the one at the top of this chat
By that, I mean, how to be simple, in complex analysis? :D
Ah, fun
Well, I suppose I could go through an entirely different simpler example:
Consider the following integral:
$$\int_{-\infty}^\infty\frac1{x^4+1}\ dx$$
just ask a background
I don't have /any/ knowledge of the residue theorem
I've read about it on wikipedia but I don't understand it...
however I have dealt with complex numbers before
22:08
Since it is negative infinity to positive infinity, the semi-circle contour is especially attractive:
@dydxx May we assume the residue theorem is true and work on it later?
@dydxx So the residue theorem is actually pretty basic to apply
Consider a Taylor series with negative exponents at a point
That's called a Laurent series
A residue of a function at a point is the coefficient of the $(z-a)^{-1}$ term
And with that definition, you can derive some formulas, etc.
A basic example: Do you know the Taylor expansion of $e^x$?
@dydxx
Yup
I am aware of it
Taylor Expansion of e^x
Then you are aware of the expansion for $e^{-1/x^2}$?
centered at $x=0$?
Wouldn't you just substitute in (-1/x^2) for x in the e^x series
(literally just plug $-1/x^2$ into the Taylor series)
yeah
22:15
We may then note that there is no $x^{-1}$ term, hence, the residue of $e^{-1/x^2}$ at $x=0$ is zero
That's residue by series expansion.
May we move on?
No I don't understand that
What do you mean by hence the residue...
The residue is defined to be the coefficient of the $(x-a)^{-1}$ term in the Laurent expansion (aka Taylor with negative powers)
Which is obviously zero here, hence the residue is zero
but there is no (x-a)^-1 coefficient term for our series
oh okay
yup
22:17
Probably should've used $e^{1/x}$
but whatever
also one more question why did u choose the semi circle contour
Now, we need to define funky integrals for complex analysis
is it explicitly based on the fact our int is from +inf to -inf
@dydxx Note the bottom part of the semi-circle goes from -a to +a... and limit as $a\to\infty$...
Yup, pretty obvious choice to choose
oh ok
22:19
So, do you understand what this means? $$\oint_C$$
An integral along our contour C?
Yes
Woo, so I get to skip that lecture XD
How would you have explained it tho?
So our contours are paths in the complex plane
@dydxx I've done it before XD
You obviously aren't my first customer
alrighty
22:24
So, the first step is to break the integral into two components. The circly part and the straight line part.
yup
Note that the circly part is bounded
It happens to be bounded by the arc length multiplied by the max value the function obtains on the arc
Anndd how do we know that?
$$|f_{max}|<\frac c{|x|^4}$$for some constant $c$
Contour integrals are like regular integrals, and they may be bounded by noting that
$$\left|\int f\right|\le\int|f|\le\int|f_{max}|$$
I don't understand where you got your f_max from
22:30
Note that $f=\frac1{x^4+1}\sim\frac1{x^4}$
(we are roughly sketching here)
Now, the next step:
are you assuming x is very large
so f can be approximated by 1/x^4
Since we are taking the limit as the radius of the circle tends to infinity
ah okay and also why did we add a constant?
'c'
22:33
Preferably, bounded. Which is why we have $c$.
(bounded is better than equivalence IMO)
ok
So we have an integral on the far right that has no argument in it, since $f_{max}$ is constant with respect to $x$.
Thus,
$$\int|f_{max}|=\pi a|f_{max}|$$
where $a$ is the radius of the circle
lol where did you get the pi and a from 0.0
Half of the circumference of the circle
okay because our contour is a semicircle
?
22:36
yup
and we are looking at the circle part first
alrighty
Now, $f_{max}\le\frac c{|x|^4}=\frac c{a^4}$
Combine all this and let $a\to\infty$, you get...
$$\int_{circle}\frac1{1+x^4}\ dx=0$$
0?
Which means that the straight line part of our contour is all that remains
Which also happens to be $\int_{-\infty}^{+\infty}\frac1{1+x^4}\ dx$
and how do we know that?
22:39
Geometrically.
A straight line is a regular integral from point a to point b, shifted up or down $yi$ units, depending where the line is
yup
A circle can be defined to be an integral from 0 to 2pi of $f(e^{ix})$, for example, again
(good example to use. Take the real or imaginary part, and you get a real integral that may or may not be hard to solve)
Anyways, back to our example?
yes
Once you've established the relationships between all of the parts of the contour, we solve the entire contour with Cauchy's residue theorem
(ugh, big words)
so in this case our integral is seperated into two parts of our contours
the cirrcle and the line
and we combine those two
to form the entire contour?
22:42
So, this means that the integral of the circle and straight line parts add up to an enclosed integral, which is dealt with by residues
Yes lol
So, we need to first find where we would have residues i.e. when the function is undefined due to division by zero
This will indicate when we will have expansions with negative powers
so we find the complex zeroes of x^4+1=0?
Yeah, pretty much, but with one more catch
We only look at points that are enclosed
i.e. upper half of the complex plane
which is the top part
of the circle
everything else we ignore
Pretty much I suppose
(x^2+i)(x^2-i)=0
22:46
The two points should occur at $x=e^{\theta i}$ for $\theta=\frac\pi4,\frac{3\pi}4$
(Skipping over and factoring with complex exponentials is the way to go)
ah okay
i get that
So next step is residues
This is actually easy
Just multiply by $(x-e^{\frac\pi4i})^n$ for the smallest natural $n$ such that $\lim_{x\to e^{\frac\pi4i}}(x-e^{\frac\pi4i})^nf(x)$ exists
This should obviously be $n=1$, since we only have one factor of each
Differentiate $n-1$ times (here we don't differentiate), then take the limit
Do this for each point, and add them up
Then multiply by $2\pi i$
And that is the answer
(lot to accept, lot to do, etc.)
Could u go through that for our example from "just mulitply"
lol
(this part I hate, so I'm gonna borrow a calculator)
Let $a=e^{\frac\pi4i}$
$$\lim_{x\to a}\frac{(x-a)}{f(x)}=\dots$$
please stand by
yup
22:52
Hm, wonder if we can apply LH on complex limits
would save me a LOT of trouble...
Complex numbers + limits in wolfram? It don't like it appears
are u asking me a question
$$R_1=-\frac{\sqrt2}4i$$
Nah, I was hand calculating some limits XD
The residue at the other point:
what is our f(x)?
$\frac1{1+x^4}$
Oh wait...
I think I got that right
>.>
how did you get R_1 ...
22:57
I computed the limit
$$a=e^{\frac\pi4i},R_1=\lim_{x\to a}(x-a)f(x)$$
how can we find the limit as x approaches a if a is just a number
6 mins ago, by Simply Beautiful Art
$$\lim_{x\to a}\frac{(x-a)}{f(x)}=\dots$$
I did that wrong
Should not divide by $f(x)$
okay so f(x) suppose to be at the top?
now we need to find
Yes, we need to find the limit of the correct thing XD
God, I'm bad
x approaches a for $(x-e^{\frac{\pi}{4}i}) \frac{1}{1+x^4}$
how do we do that lol
Factors cancel
And we take what's left
T^T such bad factoring problems
but better than evaluating hard integrals I suppose
AHA
$$R_1=\frac{\sqrt2}4i$$
ok
I hope I'm doing this right XD
X'D
uhh you can leave it there if you want
I understand the basic idea of doing a complex integral
23:03
$$R_1=R_2$$
by using contours etc
Wait, I want to actually finish
Alright
continue then xd
Nice symmetry happened XD
Add them, to get $$\frac{\sqrt2}2i$$
R_2 was found by taking the other root right
3pi/4?
23:04
Multiply by $2\pi i$ to get $\pi\sqrt2$ (let's ignore the negative, since I probably messed up somewhere)
Yes
lol alright
And that ends up equal to our integral I think
$$\pi\sqrt2=\int_{-\infty}^{+\infty}\frac1{1+x^4}\ dx$$
Darn, close, we messed up somewhere in the factoring XD
lmao
oh well
But the correct answer should be divided by $\sqrt2$, not multiplied
atleast you taught me the basis
of doing these types of integrals
23:06
Yeah, good enough XD
3
A: An integral of a rational function on the real line

Simply Beautiful ArtIf you take the counter-clockwise contour of a closed semi-circle with radius $R$ in the complex plane with a contour along the real line from $-R$ to $+R$, we have $$I_0=\int_{-\infty}^\infty\frac{1-x}{1-x^5}\ dx=P.V.\int_{-\infty}^\infty\frac{1-x}{1-x^5}\ dx$$ $$\lim_{R\to\infty}\oint_{\gamma...

@dydxx An example of when I did do it correctly
But you will notice the factoring steps are skipped cough cough
what is pv
principal value?
Yes
$$\lim_{n\to\infty}\int_{-n}^n=P.V.\int_{-\infty}^{+\infty}$$
you know how we split the integral into the arc+line
is the Arc usually zero
when we take the limit
Haha, the funny part is we usually choose it so that it comes out zero
but its not always zero, which is why we use other shapes
could you link me to one where it wasn't zero
23:12
or symmetry of a function to deal with finite limits, etc.
Well, say I wanted to look at the following integral:
Hm....
Lemme think first
$$\int_0^\infty\frac{\sin(x)}{e^x+1}\ dx$$
This one diverges for most arcs, since sine is exponential in the complex plane
Yet it is doable (not easily) with rectangular contours
ah okay
You will also note that $e^x+1=0$ infinitely many times along said rectangle
Which becomes a series problem
XD
The SUPER ADVANCED way to deal with a series lmao
Oh god, please don't try that at home
Hmm has that integral been solved here
on se?
Yes, of course. Namely, in the effort to solve a series that matches the residues
XD
Could u link me to that one
im not able to find it
lol tbh stackexchange is an amazing place
i came here asking about homework questions
got amazed by the raw talent of ppl here
and high level maths
23:18
12
A: How do I calculate the value of this series?

Jack D'AurizioYou may notice that: $$ \frac{1}{n^2+1} = \int_{0}^{+\infty}\sin(x)e^{-nx}\,dx\tag{1}$$ from which$^{(*)}$: $$ S=\sum_{n\geq 0}\frac{(-1)^n}{n^2+1}=1+\int_{0}^{+\infty}\sin(x)\sum_{n\geq 1}(-1)^n e^{-nx}\,dx \tag{2}$$ and: $$ S = 1-\int_{0}^{+\infty}\frac{\sin(x)}{e^x+1}\,dx =\color{red}{\frac{1}...

Oh god, here we go...
same
Ah, so that integral was a combo of integration by parts plus residue
lol, as he leaves the integral for viewers
hmm
hm?
No, we are not doing that problem
Sorry , this is way too much to comprehend HAHA
XD
Yeah, no problem
sadly this is the last of my spare time learning this stuff
in highschool i had so much time
but starting college this year
so gonna b to busy to learn extra math stuff
23:23
Yeah...
what are your plans for college?
I'm a year younger than you then :P
Math degree?
Hopefully
Have you tried differentiation under an integral?
I've heard of it yes
I heard it can be quite useful in certain integrals
23:25
It's really really basic. Want a quick run?
yeah sure dude
Consider the following somewhat hard integral:
$$\int_{-\infty}^{+\infty}\frac1{(x^2+1)^2}\ dx$$
it is hard to immediately solve, but the following is much easier:
$$\int_{-\infty}^{+\infty}\frac1{x^2+a^2}\ dx$$
Just solve, then differentiate with respect to $a$, then let $a=1$.
Done
:P Tricky little integrals can be reduced to a differentiation problem of a much simpler integral
Differentiate under the integral I assume right?
Brb playing a game of league
23:28
(Under the correct conditions of course)
no problem :P
(I shall be eating by the time you get back probably)
23:57
If we differentiate with respect to a under the integral we get
-4a/(x^2+a^2)^3
Nah, you did that wrong
I meant to differentiate the second integral
After solving that integral
im so confused
If we solve $\int\frac1{x^2+a^2}\ dx$
Then we can differentiate this and let $a=1$ to get our original integral
14:00 - 22:0022:00 - 00:00

« first day (17 days earlier)      last day (30 days later) »