02:00 - 06:0006:00 - 00:00

2:08 AM
what's the longest amount of time you've managed to keep working/studying without burning out

hi chat
how's everyone doing

2:33 AM
shin, if you take breaks you can avoid burnout. i usually need one every few hours, even if it's just for 10 or 15 minutes. or if you have multiple tasks, switch between them so you're not spending all day on one thing.

@Semiclassical can you recommend a good primer on quantum mechanics?
something quick, to get acquainted with the basics

Have you browsed the Feynman lectures?

no, actually
probability and uncertainty?

Yup, he does the quick and dirty approach.

I'll watch it, thanks.
I am looking for some linear algebra in there
like I guess I need someone to phrase quantum mechanics in linear algebra I guess, quick and dirty, so I understand what they're getting at

2:41 AM
@leslietownes i'm thinking long-term, multiple weeks at a time sort of thing
feels like, eventually, even taking weekends off doesn't cut it

Enjoy a long weekend, once in awhile.

3:06 AM
Alas, time flys when you're having fun.

Time flies... are they related to boring beetles?

o/
@JoeShmo the tricky thing is that it depends on what you're using it for

@leslietownes "It's better to burn out than fade away" - N. Young

shin, without anything to break it up, i think the longest i've gone is maybe 9 months on one thing. i did get tired of it but did not quite reach the point of burning out.

like, there's the approach most physics textbooks use which emphasizes the progression from classical mechanics to quantum mechanics
by contrast, quantum information/computing books tend to go right to the formalism
so if all you want is the latter, QC books aren't a bad starting point

3:16 AM
yeah so the latter

i can get sort of mini-burned out on things that don't take up all of my time. i was spending about 1/2-2/3 of my time on the same matter in my current job for almost three years, and by the end i was insanely sick of it, even though it wasn't all of my time.

but everybody ought to be familiar with quantum physics

try this JoeShmo

the standard text seems to be Nielsen and Chuang though it can be a bit intimidating

why intimidating? the math?

3:18 AM
it does quite a bit
for a lighter entry point, take a look at google.com/books/edition/Quantum_Computing_for_Everyone/…

All I'm looking for is for somebody to phrase QM in linear algebra
$|0> = e_1, |1> = e_2$ sorta thing

ah
the trouble is that most books will use physics notation
you're not going to be able to avoid bra-ket notation

that's fine I'm not particular about it

you're much more likely to see $|0\rangle =\binom{1}{0}$ as well
hmm, alright

but like, without mumbo jumbo

3:20 AM
well, the Bernhardt book i quoted as the 'lighter' entry point spends the entire second chapter on linear algebra

whatever I can skip it
I just need somebody to show me the motions

if you're confident in your linear algebra, Nielsen and Chuang may work

@Semiclassical One problem is that $1+\frac52+\left(\frac52\right)^2+\left(\frac52\right)^3+\dots$ does not converge.

yes
formal nonsense arguments have a tendency to work even though the steps in between are not justified :P

lol, I know my linear algebra

3:22 AM
@Semiclassical This one does not work. The actual answer is $\frac14$ of the answer that that approach gives.

mmkay
right. i knew there's a reason why i went back to computing $(D-2)e^{5x}$ instead

joe, the idealized book you're looking for may not exist. my knowledge is 10 years old i did not sense a lot of crossover between the QC folk notation/terminology and traditional linear algebra notation/terminology.

it's alright
I'll just jump in see what happens
we'll do it live

it does bother me a little, though. $(D-k)e^{\lambda x}=(\lambda-k)e^{\lambda x}\implies (D-k)^{-1}e^{\lambda x}=(\lambda-k)^{-1}e^{\lambda x}$ makes sense regardless of the values of $k,\lambda$

one of the drawbacks of a sufficiently new field where many of the luminaries just know everything and don't write out the translation into earlier language.

3:25 AM
huh, excellent stats

but the summation you get from 'expanding' $(D-k)^{-1}$ doesn't converge unless $\lambda$ is small enough

whatre you trying to solve semi?

trying to see whether expanding $(D-k)^{-1}$ in a Taylor series makes sense :P
(given that D is just the d/dx operator, it's not at all guaranteed)

but what's the problem statement?
I see..

3:30 AM
Suppose you have the inhomogeneous differential equation $(D-k)y=e^{\lambda x}$, and you want to find the particular solution
the simplest (and really best) way to do it here is method of undetermined coefficients

but?

but, tossing common sense to the wind, we could blithely write $y=(D-k)^{-1}e^{\lambda x}$
and ask what a sensible inverse for $D-k$ is

oy vey

lol
one way to make it seem not quite so nonsensical is to note that $(D-k)e^{\lambda x}=(\lambda-k)e^{\lambda x}$

might make sense by some functional analysis nonesense

3:32 AM
so formally, at least, we have $(D-k)^{-1}e^{\lambda x}=(\lambda-k)^{-1}e^{\lambda x}$

probably does, in fact

and indeed the last expression is the particular solution
but i was also trying to remember how one sometimes does it by writing $(D-k)^{-1}=-k^{-1}(1-D/k)^{-1}=-k^{-1}+k^{-2}D-\cdots$
@JoeShmo yes. the functional analysis way is likely to apply the Fourier transform
since that converts differentiation into multiplication

sounds plausible..

so one route is to Fourier transform, solve for y in Fourier space, and then transform back to get the solution
but meh

@Semiclassical To solve $(D-k)y=f$, it is better to use $y=e^{kx}\int e^{-kx}f(x)\,\mathrm{d}x$. The constant of integration gives the homogeneous solution.

3:35 AM
psh, but that's boring
also in the original context they had $(D-2)(D-3)y=e^{5x}$, so knowing how to 'invert' operators is handy for that

@Semiclassical Yeah, but it works.

you should get $y=(D-2)^{-1}(D-3)^{-1}e^{5x}=3^{-1}2^{-1}e^{5x}=\frac16 e^{5x}$
(using the notion that "$D$ acts like multiplication by 5" on $e^{5x}$)

Actually, $\frac16e^{5x}+C_1e^{2x}+C_2e^{3x}$

Glad to see you back @robjohn.

particular solution
operator methods generally aren't helpful for the homogeneous equation

3:38 AM
@TedShifrin Thanks. I need to get back from the real world for a while ;-)

Hope you and the pup are feeling better.

She is eating a bit better. Keeping our fingers crossed. I am doing better. Still not really interested in food. We both are in the same boat as far as that goes.

i guess there's no reason to do expansion here tho. i'm really appealing to $De^{5x}=5e^{5x}\implies f(D)e^{5x}=f(5)e^{5x}$ and hoping that nothing explodes

I was taught to do partial fractions with higher order, Semiclassic, back in 1970

However, watching the election results made the stay in the hospital decent.

3:41 AM
I think you are sympathetically ill with the pup, @robjohn.

@TedShifrin if i was doing the Fourier transform method I'd almost certainly do the same
though i guess one might be able to do something-something residue theorem

Freshman year I didn’t know Fourier. Just formal inverse of $D$ Is integral.

right

@TedShifrin Doc thinks I had acute gastroenteritis with some vaso-vagal action. She has pancreatitis, which is treated symptomatically.

What caused yours?

3:44 AM
@TedShifrin I thought it might have been some food poisoning.

Ah. You been cooking again? ;)

The food I had Monday tasted a bit off, but that could be because of a stomach flu, too.

I thought that was more recent.

I'm still not real interested in food, but it's easier to get me to eat when I don't feel like it than it is to get Dolly to eat.

3:45 AM
there was a more recent one that was not as bad. i think i've poisoned myself three times since the pandemic started.

makes a note never to eat with leslie

i think it's because we rely more on takeout food or ingredients prepared by someone else. it is hard to find good produce in stores so we don't even bother a lot of the time. so we are just rolling the dice more with a lot of overworked people's food safety measures.

@TedShifrin Worrying about her certainly did not help, and I am sure that I was not feeling well because I have been so worried about her.
not sleeping also hurt

I totally get it, robjohn.

With pancreatitis, you don't want too much fat, as that stresses the pancreas. But you need to keep them eating and drinking. So if a bit of cheese gets her eating, I give her a bit of cheese.

3:50 AM
Or a wonton.

Haven't had a wonton to try
Just need to keep her alive until the pancreas gets better.
it's dicey

good luck. we had a cat once with some kind of trouble with similar symptoms of not wanting to eat/drink yet needing to. it may have been the liver or kidneys and not the pancreas. i fed him water with an eyedropper for several weeks and he got better. lots of long nights.

You are good people. Munchkin’s bratty behavior notwithstanding.

@leslietownes Thanks, we are going to order some protein stuff to help rebuild lost muscle mass.

she was throwing her pajamas on the cat today. i asked her to stop. she said "when i turn into a cat, livvy's going to throw my pajamas on ME" and then collapsed laughing. i asked when she was going to turn into a cat. she said "at 6:45." (this is one of the few times she knows, because it's dinner time)
but 6:45 came and went with no transformation.

4:04 AM
Maybe she needs to read Kafka’s Metamorphosis.

that would have been a lighter read if he'd woken up to realize he'd turned into a cat
he got up, ate, chased stuff on the floor, then fell right back asleep. the end

Well, yes, less tortured.

Left son at UCSC. Wow. I wish I went to UG there. Apparently I am some sort of pariah.

how so
yeah UC > SUNY

it's a beautiful campus. SC itself used to be OK, a bit like berkeley without some of the harder edges. although i imagine it is gentrified to high heaven now, like everything else along the coast.

4:10 AM
To my son :-). My offspring do not like me talking to people. Somehow I ended up chatting at length with the RAs (which I thought meant research assistant).

i hope you introduced your son to every girl in the dorm. nothing like a father's assistance to break the ice. first impressions matter a lot.

I would never have imagined you chatted up a stranger copper

So many aspects, beautiful campus, lots of pretty girls, ocean within view, real grass playing firlds
etc etc

I'd never seen UCSC before

@JoeShmo Me? :-) I love talking to strangers.

4:11 AM
WOW
(I know)

He has his own room and more IT equipment than I had before age 32.
His own laser printer, microwave, all sorts of stuff. Maybe I am just showing my background :-)

why does he need a printer
thats what the library is for

I dunno. I would have viewed it as an opportunity to meet people :-)
I met lots of folks today. All very helpful.
Probably just didn't understand my accent with the mask on.

heh
alright I'm going to sleep

Had to fill out a ton of preparatory paperwork, none of which was looked at of course.

4:14 AM
toodles

Good night Joe!
I'm Joe too :-).

are ya?

Yup. I think its on my profile even :-)

release from liability this. agree to arbitrate in kangaroo court that. agree to payments on life insurance policy on my entire family with UCSC as the beneficiary. the standard terms and conditions.

oh lots of purple haired people I see in the photographs of the place
you should have dyed your hair pink to fit in

4:17 AM
california has some law where they can't force you into an agreement to arbitrate as a condition of treatment in a hospital. i didn't know about this, but the intake paperwork for my daughter's leg had it in what i'm sure was the legal minimum size it could be printed.

toc arbitration is sinful.

i love not agreeing to stuff, it makes me feel big.

surely having some standards and standard of care is in the consumer's interest.
i hate being bullied by stuff.
early on when you went to visit a company they would have you sign an NDA.

i wouldn't play games with a hospital contract (all i had to do was tick a box saying i don't agree) but with less important form contracts i sometimes cross out terms i don't like as i initial the various pages. nobody ever looks at that stuff. not sure it would hold up in court but it would be a headache for somebody someday.

i'm like no f* way. why would i?

4:22 AM
people are less NDA-happy than they used to be. they still seem to be a huge thing on the east coast.

i have mellowed considerably. my current contract is under illinois law which is not as favourable as CA law
i am good with it in certain cirsumstances, but not as a fishing trip
that is bullying the unwary
did you get the xrays for your daughter's break?

@copper.hat wow! a printer too!

@Koro Yup. A lot different to my days :-)
The first (sort of) printer I say was a teletext machine. I am sure I have rambled in about one here at some stage :-)
I loved the fanfold paper. I used to collect scrap to use as scratch paper.

a few times i've been exposed to situations where a relatively unsophisticated party, e.g. inventor or entrepreneur, visits a big company, signs some NDA that might be two-way, gives a talk about some technological direction, the big company says we'll think about it, then rips it off, and if the little guy complains they get threatened with violating the NDA by supposedly stealing the idea from the company instead of the other way around.

@copper.hat When I was at UCLA undergrad, I set up my PET in my dorm room. I was labeled a geek at that point.

4:26 AM
Back then they didn't care that million dollar/pound contracts were printed on the back.

@copper.hat my wife has a blurry photo of a screen they showed her at the appointment. i'm no doctor, but i could tell it didn't look good.

@copper.hat None of my classmates at college ever needed to own one, there were various stationery shops around to get photocopies/prints etc. Everybody utilized the service.

@robjohn A PET would have been awesome then. We had one in the house which was very unusual at the time.
@leslietownes Next time ask the dr for prints on a cd.dvd.

@copper.hat I hadn't gotten my dual 400K floppy drive yet

4:27 AM
I used the cassette for data

we may have gotten them. i was not allowed into facility. only one parent could go in although the paperwork said we could both go. i had a fun time reading email on the ground floor.

When I was back a few weeks ago we came across some old tapes. The cassette player had disappeared unfortunately.
@leslietownes they always pay more attention when you ask for media. in your case you could mention your profession to help.

as a kid i drew pictures exclusively on fanfold paper from some old machine at my dad's office that was on one of the news wires, printed on one side only. we had gobs of the stuff. never bought paper.

It will usually be in some format like dicom, but you can find a viewer easily and taking pics on your screen is always better.

if my parents kept any of my art you could flip it over and see what the news was in 1985, or whatever.

4:31 AM
i love that dual use.

@leslietownes update on the project:
Various arrow styles now available for lines and the arrowhead. Not the tail yet. That's next

@SmokenSieEinBitteChebaHitBits are you taking a programming class?

but can i play Doom in this? :)

@copper.hat I'm a professional, just have no work
As you can see by my app, it's quite advanced

@SmokenSieEinBitteChebaHitBits sorry to hear it. good luck looking.

4:32 AM
For this area (no one other than maybe WildCats and q.uiver.app has done CD stuff
For a great editor go to Quiver
For all proof-related stuff, come to us
I'm making a website for launching it too

thankfully nobody has yet figured out that i am a crap programmer

it's open source, but I think I'm going to open up a data marketplace

modern stuff is different. all github and the like

Yes, well you can't close source a PyQt5 (Python library to C++ Qt Framework)
application without paying $2000 which I don't have So the next best bet is to sell my data Since making it using the tool itself (that's what it's for - for making educational diagrams) and logically apply rules to diagrams You can specify with it pretty much any type system / formal system you want As long as it can be built up from rules about text + diagrams So for instance a package called "group theory" with montly updates might be sold to the researcher for$50 / year
I'm looking for help if anyone's willing to help code it
It's weird, but coding speed nowadays is slow as hell (at least my pace)
It's not surprising since apps are built piece-by-piece like one enlargening rubik's cube
And you're always rewriting because you can't have forseen an issue and so on

i work in what is called the eda industry, basically tools to design chips.
good enough is.

4:38 AM
Nice, i've used Altium Designer befre
for board

here is a geometry problem that's been on my mind this evening

fpga stuff?

No, never got to that stuff

what a long name

Pretty sure there were some in the layouts I was PCB layouting

4:39 AM
ah, altium, pcb stuff

I'm making something similar to altium but it's for math itself, not electronics
It's like a drafting tool for math basically

I am thinking of naming my new project RSTDS
random software that does stuff

suppose we start with the following four unit vectors in the plane: $(1,0,0), (\cos \theta,\sin \theta,0), (\cos 2\theta,\sin 2\theta,0), (\cos 3\theta,\sin 3\theta,0)$

It should also output in TickCD or at least to Quiver
What about STDSIRTU, software that does stuff intelligently at random to u
because software messes with my head. A simple bug could take days

for simplicity, i'll specialize to two cases in particular $\theta=30^\circ, 150^\circ$. Note that we have $3\theta=90^\circ$ (mod 360 degrees) in both cases.

4:43 AM

not right now, later
l a t e r

Please star it and watch it as a github member

now, I'm reasonable confident of the following: given an arbitrary angle $\theta'$ between $30^\circ$ and $150^\circ$, i can find vectors which are at angle $\theta'$ with eath other and with $(1,0,0),(0,1,0)$

I use github for open and bitbucket for closed source (both freely)

what i haven't found is a nice way to construct those vectors

4:45 AM
@Semiclassical sine/cos

what?

Use the sine-cosine rotation matrix formulae

that's pretty well useless.

Why's that

because finding vectors which are $\theta'$ away from $(1,0,0)$ and $(0,1,0)$ is easy
it's finding a pair which also have that angle relative to each other that's the problem
also, for rotation matrix you need to know your axis of rotation

4:47 AM
@Semiclassical you say "in the plane" above

and i very much do not here

what do you mean - those are 3D vectors

that's where the vectors start
but the moment i change to another angle $\theta$, they won't be in the plane

Oh, so they are 0 on z-axis
so the XY-plane

that's where they start, yes
that's not where they'll stay

4:48 AM
Okay, so given these 4 unit vectors then what

let me try again. suppose i start with the vectors $e_1=(1,0,0)$ and $e_2=(0,1,0)$

So $\hat{x}, \hat{y}$ for short
Or $e_i$, i see!

i want to find vectors $u_1,u_2$ such that $e_1\cdot u_1=u_1\cdot u_2=e_2\cdot u_2=\cos\theta'$
the limiting cases should be $\theta'=30^\circ,150^\circ$, since in those cases i can choose $u_1,u_2$ in the plane
in general, though, they won't be

What do you mean by limiting cases?

well, there's no way to pick $\theta'=0$
because you'd need $u_1=u_2=e_1=e_2$
which is manifestly false

4:51 AM
How do you see that?

by knowing that $e_1$ and $e_2$ are different vectors?

Are they not parallel?

what
(1,0,0) and (0,1,0)?

Oh nvm
I see now

right. so you can't make $\theta'$ arbitrarily small. you also can't make it too big
the range is set by the cases where $u_1,u_2$ lie in the XY plane

4:54 AM
why are those the min and max areas?
I mean why is it that the min / max will occur there?
of theta

for the lower end, it's because the shortest path between two vectors is a straight line
going from (1,0,0) to (0,1,0) is a 90 degree shift
so in three equal steps the shortest possible is three 30 degree shifts

So you're asking for unit vectors $u_i$?

Clearly as you said you just divide the total angle range by 3
What's wrong with that?

that works to set the lower bound, yes
but you can also get there in steps of 40 degrees, if you don't take a straight path

4:57 AM
I see I see

the problem is to figure out what paths work, depending on the angle

So you're asking for what is the set of all such pairs $u_1, u_2$?
I mean all such sequences $u_i, i =1..n$?

as a function of theta, yes. there's some redundancy to that---if you find one pair, you can rotate the sphere to get another one---but one pair should suffice

Are you always using elementary $e_i$ or do those basis vectors also change, and do they always form an orthonormal set?

if they formed an orthonormal set, they wouldn't be at angle theta to each other
oh
no, the endpoints always stay the same
so always (1,0,0) and (0,1,0) for this example.

5:03 AM
What form of solution are you looking for?

parametrization of the middle vectors with angle
or some way to construct them

If $n = 1$ and $\theta < 90/(n+1)$ then out of luck, so we can assume that you only want a formula for $n \lt 90/\theta - 1$, right? That means there's a finite number of $n$ cases.
for each theta

angles are always messy

n=2 is the only case i'm interested in

So that's 3 angle sections then in my notation that would be n = 3
?

5:08 AM
sure. but only two vectors whose positions can change
angle from e1 to v1 = angle from v1 to v2 = angle from v2 to e2 = theta

There is rotation involved
You take the vector $e_1$. Then Rotating $u_1$ about $e_1$ gives a circle drawn on the surface of the unit sphere

rotation is always fun
i used to tell my dates.

yeah. and you do the same for u2 about e2

That represents all possible values that $u_1$ can be
@Semiclassical make use of symmetry
do the same thing for $u_2$ and $e_2$ yes you got it
Not sure how to handle the solution yet though, working on it

so fixing t, one circle is (cos(t), sin(t) cos(s), sin(t) sin(s)) and the other is (sin(t) cos(u), cos(t), sin(t) sin(u)), and you want to see if there are s and u for which the dot product of these is cos(t)
knowing that for some extreme values of t there won't be. do you want to identify the range of t for which there is a solution?

5:16 AM
It will invovolve solving a polynomial in $t$ I think
$t,s$

ugly formulas but given t a computer could handle this numerically, or provide convincing evidence that there is no solution

yeah. right now numerically seems the only recourse

I think maybe convert it over to exponential form before concluding that

i mean, the condition is that the dot products between leslie's vectors be cos(t)

Yes, so there should be some cancellation

5:18 AM
so cos(t) = cos(t)sin(t)(cos(u)+cos(s))+sin(t)^2 sin(s)^2

or combining atleast
Can t = 90 here?

yes, that's one case

So then that gives you 0 = (sin t sin s)^2

looking at some numerics, it may be best to focus on the transition from theta = 30 deg to 90 deg in fact

Thanks for starring my github repo whoever did that - it should also be starred on the github site and "watched" for updates

5:22 AM
in the case of theta=90, one can pick (1,0,0), (0,-1,0), (-1,0,0), (0,1,0)
hmm. that feels wrong
maybe not tho

how would people know where your github is?

@leslietownes you know i am an indolent

i'm captious.

love that word. have yet to use it in a sentence with someone

5:31 AM
Sweet, thanks to the two starrers and watchers on github!
@copper.hat I will make it an awesome tools. I'm fixing about 3 bugs a day now since some bugs are harder than others. I'm almost into the nitty gritty - the logic / math code which will be done by graph matching using the networkx lib
It will slowly but surely be ported to C++ / Qt for the purpose of going mobile and speed of C++ 2D graphics scenes
It's nice to be able to prototype in Python though since it is releaseable once completed

@SmokenSieEinBitteChebaHitBits good software is what is does well, do not waste time on bugs unless they matter.

cpp good

@copper.hat not sure what you said: is what does well?
These bugs definitely matter: for instance the mouse grabbing and sticking to a control point indefinitely until you place another node, not even hitting escape will save you!
That sort of thing

looking at the surface i get for (u,s,t)
i begin to appreciate exactly how annoying this problem is

I mean people often focus on bugs as if real software does not have bugs. All software has bugs. So, focus on making it do something really good and don't worry about irrelevant edges.

5:36 AM
I know, same goes for features
you use the simple default behavior if it's useable and it's not critical to to the main thesis

i work with very few customers, so 'quality' in the traditional sense is a luxury.

I shouldn't rewrite the context-menu code feature then before beginning the main logic engine
The code feature would be making it easy to add/remove context menus for any type of object in the scene and it would also work on any selection of objects and display the right subset of the union of all the items' context menus
Quite a complext feature that only the maintainer will appreciate, so I'll skip the selected items context menu for now

i think i get the approach now, though. basically the construction one needs to use for angles 30 < theta <90 is different than the one for 90 < theta < 150
i don't quite get why it doesn't work otherwise but it doesn't work

@SmokenSieEinBitteChebaHitBits I'm still working for a living, so perhaps you should not take my advice :-)

You've convinced me
You've altered the course of the Abstract Spacecraft development

5:43 AM
oops :-)

seems like people who chat here are redditors
only emoticons, no emojis

What is the smallest formal system known about that has all the symbols that appear in a standard definition of group?

@Euler2 ascii is just fine :-)

Lol, using unicode, I simply copy/ pasted in tables that came into the source code as nicely formatted blocks
from wikipedia
I copied all math-related symbols that are currently only used for auto-indexing
Then there is a link to shapecatcher.com for drawing a unicode letter and giving the user copyable text
A unicode widget is out of the question - too many buttons make loading it slow
A cache would be nice but the user can make their own copy / paste cache simply by opening another tab or window in the app

good night folks!

5:57 AM
Night!

02:00 - 06:0006:00 - 00:00