<bofh>
23:59 <@soundnfury> you're either sad, or happy going backwards in time
<bofh>
23:59 <@egg> soundnfury: ... this sounds bleak
<bofh>
rofl, quote of the day
<egg>
this is transforming things into some ~weird other 3d space~ nonlinearly, and then leaves you with the task of doing an actual projection
<soundnfury>
!wpn bofh
* Qboid
gives bofh a threaded check which strongly resembles a theory
<egg>
mind you the composition is an actual projection, and a perspective one
<egg>
but the middle thing is a tad weird
<whitequark>
yes, that's the idea
<whitequark>
like I'm saying, it's for the export code
<egg>
(it may be useful, but it's mathematically weird)
<whitequark>
where you need to order things in draw order afterwards
<whitequark>
so you can't just discard z immediately
<egg>
yeah, I get it
<egg>
but it's not really a projection in and of itself
<egg>
hmm
<soundnfury>
egg: sometimes people have to write code that's not a direct translation of a mathematical paper
<soundnfury>
are... are you gonna be okay with that?
<egg>
wait, is the whole thing linear?
<egg>
no, obviously not.
<egg>
also I'm clearly not awake enough for linear algebra.
<egg>
or as it happens projective geometry
<egg>
whitequark: also I see you too have to deal with the joy of inconsistent handedness
<egg>
for *absolutely no good reason* half of KSP's physics is left handed and the other half is right handed
<whitequark>
lol
<egg>
and this is physics
<egg>
so you have bivectors
<egg>
(pseudovectors for the physicists in the room)
* soundnfury
gives egg a Levi-Civita tensor
<egg>
soundnfury: hmm
* soundnfury
gives egg a contravariant bibasis
<egg>
so...
<egg>
soundnfury: those are the... duals of bivectors?
<egg>
soundnfury: principia does not do GR
<soundnfury>
I don't know, I think so maybe?
<soundnfury>
also it might be called a cobasis rather than a bibasis
<egg>
and it does not distinguish linear forms from vectors (covectors from vectors)
<soundnfury>
egg: I am shocked!
<egg>
soundnfury: uh but that would just be the dual, not bivectors right?
* soundnfury
shrugs
<soundnfury>
as you may have noticed, I don't _actually_ know what I'm talking about
<egg>
anyway I am quite happy to have a canonical inner product and to not have to worry about the distinction between vector and covector :D
<bofh>
rofl
<egg>
whitequark: so, hopefully that answers your question, or rather you probably answered it yourself while I was puzzling myself with linear algebra :-p
<bofh>
"contravariant bibasis"
<bofh>
is that like actually a thing
<egg>
I don't think I've heard this term
<whitequark>
egg: it doesn't
<whitequark>
it doesn't even come close lol
<egg>
whitequark: ok?
<whitequark>
but
<egg>
whitequark: what's the question again :-p
<egg>
can you kill z to get the screen coords?
<whitequark>
i've realized i can't render that anyway
<egg>
yes
<whitequark>
nah, I know I can
<whitequark>
that's how I render everything else
<egg>
ok, but then what's the question
<whitequark>
the problem was that I didn't have texture cooridnates in a mesh
<whitequark>
actually, no, not quite
<soundnfury>
bofh: I don't know. But now I want it to be
<egg>
I, uh, I don't understand that question?
<whitequark>
I'm trying to emulate opengl with cairo and they use completely different transformations
<soundnfury>
whitequark: the images f(u), f(v) should give your basis vectors for the texture in the image, no?
<whitequark>
with opengl I have a 4x4 matrix that I multiply vertex coordinates with
<whitequark>
or call the ProjectPoint3, which is more or less the same thing
<whitequark>
and it takes care of putting the texture where it belongs
<whitequark>
with cairo, no such luxury, I have to express the transformation applied to the texture, somehow, with the 2x2 transformation matrix cairo uses
<whitequark>
since it only works with 2d surfaces
<egg>
whitequark: so that should be the 2x2 matrix that has which property? maps u to the image of u, and v to the image of v, under the camera projection? That's going to depend on where o is though, so is the matrix dependent on o?
<whitequark>
i think so
<egg>
but then isn't that just what soundnfury said?
<egg>
if f be the camera transformation, <soundnfury> whitequark: the images f(u), f(v) should give your basis vectors for the texture in the image, no?
<egg>
that is, the matrix that maps u to f(u) and v to f(v)
<whitequark>
um, i guess i hsould take them out of /ignore
<whitequark>
also yes that sounds right
<egg>
... ah, yes I vaguely recall some strange OS-related backlog from yesterday
<soundnfury>
egg: whitequark doesn't like me because I quoted esr
GreeningGalaxy has quit [Ping timeout: 206 seconds]
<egg>
you *are* rather annoying when the discussion pertains to operating systems at times
<soundnfury>
egg: I'm rather annoying _under most circumstances_
<egg>
a fair point
<soundnfury>
tbh, I'm surprised you still let me in here, given I'm not even in academia
<egg>
I'm not sure on what occasion you came in? but frankly look at who's in the channel, it's not really limited to academics
<egg>
even *I* am no longer an academia
<egg>
an academic rather
<SnoopJeDi>
I wear an academic's trappings, but...
<egg>
or well depending on your definition I never was, just an MSc student
<egg>
but now I have my MSc
<UmbralRaptor>
SnoopJeDi: Defacto engineer of some sort these days? >_>
<egg>
and I didn't go prof-hunting to get a Phd and my MSc prof didn't have the time to supervise a phd
<egg>
bofh, Stratege, TentacularGalaxy, UmbralRaptor and SnoopJeDi are
<SnoopJeDi>
even that is rather too generous, UmbralRaptor
<egg>
and... I think that's it?
<egg>
oh maybe Thomas is an undergrad now?
<UmbralRaptor>
egg: Uh, I think TentacularGalaxy is going to enter industry after her BSc?
<soundnfury>
egg: so what are you now?
<egg>
UmbralRaptor: yes but for now she's a student
<UmbralRaptor>
Fair enough.
<egg>
soundnfury: waiting for a hiring process
<whitequark>
TentacularGalaxy: did you grow tentacles recently :o
<soundnfury>
egg: I see, I hope that goes well
<egg>
Passed hiring committee, so there's that
Wetmelon has quit [Ping timeout: 206 seconds]
<egg>
talking to a YouTube person monday morning to discuss a couple of positions they have in fact
<egg>
well I say person, but this is YouTube, probably a cat
<UmbralRaptor>
You can apply to be a cat?
<egg>
whitequark: if you want someday I can talk to you about mathematical abstractions that are useful for geometry (and Principia, that implements them), these things can help clarify one's mind
<UmbralRaptor>
whitequark: From an offhand comment about how spiral galaxies really have tentacles instead of arms.
<soundnfury>
egg: fsvo "clarify", ymmv
<soundnfury>
:P
<whitequark>
egg: so I've read about them a few times
<egg>
We shall define *clarify* in section 5.
<whitequark>
but the underlying issue is uh
<whitequark>
what I *want* is, like, welding shit out of metal
<whitequark>
what I *ended up doing* is maintaining a parametric CAD
<whitequark>
I don't *want* to maintain a parametric CAD it is just the... easiest path to what I want, given constraints
<egg>
whitequark: sure, but it may be less boring if you know what's going on more abstractly
<whitequark>
egg: I literally pay a guy to do math for solvespace full-time so that I can deal with simple things
<SnoopJeDi>
Does anybody actually enjoy maintaining CAD?
<whitequark>
like keeping it working on three platforms at once
<whitequark>
and soon five
<SnoopJeDi>
I had to fight our typical drafter tooth and nail to get him to use parameters in Inventor instead of dimensioning everything ad-hoc
<whitequark>
(well, mostly licensing fees pay for it, but sometimes I have to...)
<whitequark>
SnoopJeDi: it wouldn't be so bad if it had like, tests,
<whitequark>
and if it wasn't written in the horrible C++-but-actually-C style
<SnoopJeDi>
yea the testing mentality would be pretty helpful
<whitequark>
the string class consisted of char s[64].
<egg>
...
<whitequark>
and one method, which was a wrapper around strcpy.
<SnoopJeDi>
oh you mean the software itself
<whitequark>
not even strncpy.
<SnoopJeDi>
oh god
<egg>
whitequark: ... you could learn about the math, do that, and hire someone to do the programming shit-shoveling :-p
<whitequark>
I wrangled it into submission and now it works correctly with Unicode, even on Windows, even regardless of the system codepage
<SnoopJeDi>
I know enough linear algebra to know that the innards of anything that can handle nontrivial constraints is probably a mad place
<egg>
SnoopJeDi: yes, but there are various kinds of madness
<whitequark>
egg: I don't actually know anyone competent who's anywhere within how much we can pay
<whitequark>
like *obviously* there are programmers at Google I could poach who can do this
<egg>
but then how do you afford a mathematician O_o
<SnoopJeDi>
egg, a friend of mine who works for a visualization group was explaining how their latest word cloud product is a projection down from a 500-dimensional space
<whitequark>
he's russian
<whitequark>
russia recently crashed its currency *hard*
<SnoopJeDi>
and *mumble mumble* something about rotations in the 500-D space
<whitequark>
well, fsvo recently
<whitequark>
... on the other hand we get the licensing fees in USD.
<egg>
hah
<whitequark>
also the reason we employ this specific guy is because he actually worked at a company that tried to commercialize solvespace
<egg>
whitequark: well, anyway, if you have questions I'm happy to answer them, though do bear in mind that my replies do involve some amount of trying to teach the underlying abstract nonsense :-p
<whitequark>
so he knows it decently well
<whitequark>
but his boss was an asshat so he left
<egg>
also I wanted to check that I had fixed a bug, brb
<whitequark>
egg: in general this is a useful offer but unfortunately i'm really bad at math
<whitequark>
I failed diffequations eight times back when I studied at an uni
<whitequark>
(out of eight)
* egg
tends to be shit at eggsams too though
<egg>
pretty sure I never managed to pass anything involving Lie algebras on the first try
<whitequark>
it has nothing to do with exams and everything to do with the actual level of understandin
<whitequark>
(also pretty sure I have some sort of attention disorder, but that's another story)
<SnoopJeDi>
I haven't gone on a Lockhart's Lament rant on an unsuspecting normie in a while
* SnoopJeDi
pencils that in for this week
<soundnfury>
SnoopJeDi: ooh, will you be selling tickets?
<SnoopJeDi>
I think I'll have to pay to get anyone outside my (frictionless) echo chamber to listen
<SnoopJeDi>
Last time I got my bile up about it I was talking to a teacher at a LEGO robotics event, I think
<egg>
yay that bug is gone
<egg>
whitequark: well anyway, I'll try nonetheless :-p (and here's a tidbit for today: you can distinguish points from vectors, with point - point = vector, point + vector = point; you can do that in typing, or at least when reasoning on some things, and it helps)
<egg>
(and that's called an affine space, but fancy terminology is just that)
<egg>
(a classic example is dates vs. durations)
<egg>
can't add points, can't add dates
<whitequark>
egg: yeah I called it a vector because that's how the class is called internally
<egg>
whitequark: it can help to make it a different type, though with a preexisting codebase it may be gargantuan
<egg>
(or pantagruelic?)
<whitequark>
egg: it's a lost cause, I think
<whitequark>
but also it never caused any issues so who cares?
<egg>
from what you say about your codebase, I think it is :-p
<soundnfury>
egg: brobdingnagian?
<whitequark>
egg: also in general CG seems pretty loose about its terminology
<whitequark>
in GL you only have vectors
<soundnfury>
(possibly s/ng/g/, I forget)
<whitequark>
a triangle is defined by three vectors, period
<egg>
whitequark: I would say about its typing, rather than terminology
<whitequark>
it's also a factor because I have to interop with CG libs and such
<whitequark>
same thing here
<whitequark>
it's distinguished neither here nor there
<egg>
much like physics engines don't strongly type quantities (which are of great relevance) or bivectors
<whitequark>
yes, with physics engines it's much worse
<egg>
[insert usual rant here]
<egg>
especially since dimensional analysis in particular is a *big* bug catcher
<soundnfury>
egg: I'm sure you could fix them by doing something like sparse
<soundnfury>
extend the type system when running your static checker, with annotations that get #defined out for the actual compiler
<egg>
meh, we just let the compiler do that in principia :-p
<soundnfury>
egg: sure, but the topic of conversation was _sucky_ physics engines :p
<egg>
well you can wrap them :D
<egg>
that's what we do
<egg>
we wrap KSP
<egg>
in a codebase that's slightly smaller than whitequark's :D
<egg>
(slightly)
<egg>
whitequark: also to those line counts I gave the other day one should add 2040 lines of proto specification :-p
<egg>
whitequark: so, have we answered your question with the linear map for that texture?
<whitequark>
egg: I still don't know how to go from basis vectors to a transformation matrix *but* I can probably just google that
<whitequark>
so yes, thanks
GreeningGalaxy has joined #kspacademia
<egg>
whitequark: so if u and v are the basis of the texture it's easy
<egg>
whitequark: the columns are what they get mapped to
<egg>
done
<whitequark>
... oh
<egg>
(assuming it gets applied on the left)
<egg>
(otherwise it's the lines)
<whitequark>
well that was easy
<whitequark>
I think I even understand why
<whitequark>
why it works, that is
<whitequark>
I was bad at linear algebra but less bad than at everything else
<egg>
yeah, u is (1, 0)^t, you extract a column
<egg>
\o/
<whitequark>
(now if you wanted to know something about physiology of invertebrata...)
<egg>
whitequark: you did biology? I do notice you seem to know about that
<whitequark>
I was studying at the department of bioengineering and bioinformatics
<whitequark>
that involved a bunch. human physiology, biochemistry, invertebrata, analytical chemistry, cutting dead animals open, of course bioinformatics
<whitequark>
I think my knowledge of biochemistry (and of how to read biochemistry papers) was the single most useful skill I carried out, followed by analytical chemistry
<whitequark>
egg: oh is that a millipede
<egg>
whitequark: yes, and good luck identifying the thing :-p
<whitequark>
haha no I dropped out before the invertebrate fieldwork
<egg>
the insect on that leaf might be doable
<whitequark>
I did have to draw and remember seventy cereals at some point
<whitequark>
couldn't remember a single one today if my life hinged on it
<whitequark>
I do have some of the drawings still...
<egg>
hah
<whitequark>
should probably scan 'em, some of them are pretty sweet
<egg>
(there are big gardens inside of Saigon, quite pleasant)
<egg>
twas unusually wet this year though
<GreeningGalaxy>
MPL?
<Qboid>
GreeningGalaxy: [MPL] => Mozilla Public License
<GreeningGalaxy>
so not Mars Polar Lander
<egg>
uhm, considering that this is a rather spacelike channel,
<egg>
!acr -update:MPL Mars Polar Lander
<Qboid>
egg: I updated the explanation for MPL
<bofh>
yeah, that's much better
<egg>
bofh: is zlaing the astronomer you were speaking of?
<bofh>
egg: ya
<bofh>
egg: they may be quiet for a bit, they lurk a lot in new irc channels
<egg>
:D
<egg>
bofh: UmbralRaptor is the other astronomer here
<bofh>
well that would explain the typeIII galaxies comment :P
<bofh>
egg: so who is who here
<bofh>
I know you and whitequark
* UmbralRaptor
pinged.
<bofh>
and SnoopJeDi
<bofh>
(sorta, from yesterday's HFML convo)
<GreeningGalaxy>
I don't know what made all these incredible people just showed up in this channel all of a sudden, but I'm so glad
<bofh>
<shrug> this has been the first IRC channel not involving IRL friends I've been to in years that I've actually *liked* the atmosphere in
<egg>
bofh: UmbralRaptor is @Newpa_Hasai on the twitters, GreeningGalaxy is TentacularGalaxy is @diodelass
<bofh>
but I guess part of that's b/c I've increasingly been hating programming lately
<bofh>
so that's most of freenod
<bofh>
freenode*
<egg>
Greening is uhm, an eleggtronics person afaict?
<GreeningGalaxy>
egg: spacelike? does it have Δs² < 0?
<egg>
obviously
<whitequark>
bofh: oh did I ever tell you about #solvespace on fn
<egg>
Stratege is a german CS student who is actually at ETHZ
<GreeningGalaxy>
I'm an electronics person, yeah. Although still a bit of a neophyte.
<whitequark>
GreeningGalaxy: do you like FPGAs
<bofh>
whitequark: I idle there
* egg
prepares psychologically to stare blankly at the eleggtronic magic
<GreeningGalaxy>
(I've well-surpassed the rest of my physics department, but probably very few upperclasspeople in an actual EE department)
<whitequark>
egg: this is how i feel about math
<GreeningGalaxy>
whitequark: juuuust getting into them. My professor just ordered some for research porpoises which I eggspect I will be learning how to use.
<whitequark>
like i understand the basic elements but the way they compose is beyond me
<egg>
I am eager to learn how this dark magic works though :-p
<whitequark>
GreeningGalaxy: wanna make a GPU
<whitequark>
from scratch
<whitequark>
on a FPGA with a fully FOSS toolchain
<GreeningGalaxy>
I got to see people operate the magic FPGA box at the APS when I went there with him, but didn't get to learn how it works on any really deep level
<GreeningGalaxy>
whitequark: that sounds... exciting?
<bofh>
GreeningGalaxy: research "porpoises"?
<egg>
bofh: Thomas is... Qboid's human? German? StollD on github, probably a student?
<whitequark>
GreeningGalaxy: niiiice
<egg>
bofh: recurrent pun here
<GreeningGalaxy>
bofh: blame UmbralRaptor and egg
<bofh>
egg: thanks, let me follow some folk on twitter now
<egg>
bofh: as well as eggpuns
<whitequark>
GreeningGalaxy: so I have some work done already, fairly little
<whitequark>
GreeningGalaxy: parts of the shader ISA
<whitequark>
and parts of the simulator
<whitequark>
and very small parts of the compiler
<UmbralRaptor>
bofh: For all intensive porpoises, cetacan puns are quite necessary.
<egg>
UmbralRaptor: cross-channel answer re. this being ANBO letters: I think we should formalize ANBO letters into a blog or somesuch at some point :-p
<GreeningGalaxy>
whitequark: can you, uh, give me an idea of how electronically complex FPGAs are? I don't get a sense of that. More or less so than an MCU?
<egg>
MCU?
<egg>
GreeningGalaxy?
<GreeningGalaxy>
!acr -add:MCU Microcontroller Unit
<Qboid>
GreeningGalaxy: I added the explanation for this acronym.
<GreeningGalaxy>
egg: a little chip what runs code.
<whitequark>
GreeningGalaxy: depends on which FPGA
<whitequark>
and which MCU
<whitequark>
I mean your Core i5 is an MCU
<egg>
so... isn't an FPGA that too?
<egg>
what's an FPGA
<GreeningGalaxy>
sure, I mean... right, how do I word this one
<whitequark>
LUTs are how you implement combinatoric functions
<whitequark>
flip-flops is how you get state
<egg>
whitequark: oh that thing that people ran genetic algorithms to configure and had weird circuits with disconnected bits that mattered?
<UmbralRaptor>
(er, unrelated graphs that egg suggested I post here)
<UmbralRaptor>
egg: capacitive madness?
<UmbralRaptor>
(or something like FETs?)
<whitequark>
special-purpose blocks are chunks of dense RAM (it's much cheaper than assembling it from individual flip-flops), multipliers, adders, etc
<whitequark>
this is basically it
<whitequark>
the more expensive the fpga is, the more complex the interconnect
<whitequark>
and the more complicated the selection of primitives
<GreeningGalaxy>
whitequark: I suppose a decent way of rewording my question is "does the simplest device which would be marketed as an FPGA have a greater or smaller number of individual transistor elements than the simplest device which would be marketed as a microcontroller?"
<whitequark>
GreeningGalaxy: neither
<whitequark>
there are extremely small MCUs and extremely small FPGAs as well
<whitequark>
for example, the GP4 series is called "FPGA" yet it has, like, dozens of flip-flops
<whitequark>
(not all of them directly programmable)
<bofh>
yeah I don't get how the GP4 is called an FPGA
<whitequark>
GP4 is on the extreme low end of course
<whitequark>
bofh: well I guess there's no real line between "CPLD" and "FPGA"
<bofh>
like for me I keep assuming that unless it has a certain number of programmable elements, it can't be called an FPGA
<whitequark>
there are definitely CPLDs larger than the smallest iCE40 member
<bofh>
otherwise it's just a toy
<whitequark>
which is like 300 slices
<whitequark>
on the other hand from a perspective of sb0, who works with kintex7...
<whitequark>
the entirety of iCE40 is just a toy
<egg>
whitequark: alright, and what's the advantage of this wrt to a nonreconfigurable microcontroller
<whitequark>
egg: parallel processing, chiefly
<whitequark>
you can respond to n things at exactly the same time indepnedently
<whitequark>
with a microcontroller once n>2 that becomes... quite hairy
<GreeningGalaxy>
could an FPGA be considered a more general case of a microcontroller? Could one configure a large FPGA to function similarly to a microcontroller?
<whitequark>
yes!
<GreeningGalaxy>
aha
<whitequark>
in fact, in ARTIQ, we configure our FPGA to contain two CPU cores
<whitequark>
and a bunch of custom peripherals
<whitequark>
so a large FPGA can be a microcontroller made just for you.
<GreeningGalaxy>
okay, that's cool
<whitequark>
prototyping digital logic is another primary purpose of FPGAs.
<whitequark>
every Intel CPU at some point ran in some Xilinx device
<egg>
whitequark: so, what's the advantage of using a reconfigurable chip rather than making an ad hoc nonreconfigurable chip for some application? economic considerations?
<whitequark>
egg: of course.
<GreeningGalaxy>
right, designing a new IC is a huge undertaking
<whitequark>
taping out an IC is tens of millions of $ and at least one year of work
<whitequark>
that is if you get it right the first time...
<whitequark>
... which you are only gonna do if you test it on an FPGA.
<whitequark>
(plus a lot of simulation)
<egg>
ah, makes sense
<whitequark>
once you have it running on an FPGA you can do an ASIC conversion cheaply
<whitequark>
basically you ask the fab to make masks that do the same sort of layout and routing that the FPGA has
<whitequark>
but you get an instant boost in fmax
<egg>
fmax?
<whitequark>
maxximum frequency
<egg>
so fpgas are slower than a corresponding ad hoc IC?
<whitequark>
yes
icefire has joined #kspacademia
<egg>
hmm
<GreeningGalaxy>
yeah, I think the reason the FPGA boxes at Argonne are developed in-house with a monumental budget is they need to be fast enough to be useful for an x-ray beam that arrives every 153 nanoseconds
<whitequark>
the cores we have in our kc705 are running at 200MHz
<whitequark>
which is laughable
<bofh>
GreeningGalaxy: wait, you're at Argonne?
<whitequark>
i think the usual speedup from an ASIC conversion is 3-4?
<GreeningGalaxy>
bofh: no, my prof and I got beamtime there a couple weeks ago
<bofh>
APS?
<Qboid>
bofh: [APS] => Advanced Photon Source
<egg>
whitequark: so, for computing, if you have something that's not highly parallel you're better off running it on your good ol' CPU, but for some sufficiently embarrassingly parallel application you could send it to one of those thingies as a coprocessor?
<whitequark>
egg: it depends on a lot of factors
<whitequark>
so for example we have an RTIO core that you can tell "toggle these 20 channels at exactly the same time, with nanosecond precision"
<whitequark>
you can't do this with a CPU alone
<egg>
RTIO?
<whitequark>
and there are no ready-made peripherals for this kinda thing
<whitequark>
real-time IO
<egg>
!acr -add:RTIO Real-Time IO
<Qboid>
egg: I added the explanation for this acronym.
<whitequark>
um, it's just something we made up for ARTIQ
<bofh>
well I assume APS since that's the main thing at Argonne you'd be asking for beamtime on
<GreeningGalaxy>
right, APS
<egg>
we have ANBO in that acronym directory
<bofh>
what was the experiment?
<whitequark>
!acr -add:ARTIQ Advanced Real-Time Infrastructure for Quantum physics
<Qboid>
whitequark: I added the explanation for this acronym.
<GreeningGalaxy>
sorry I assumed you were asking the bot lol
<egg>
RTIO will be fine :-p
<whitequark>
egg: there are also other reasons, like reconfigurability
<whitequark>
e.g. you could reuse the same FPGA to run different coprocessors, or to do field upgrades
<bofh>
GreeningGalaxy: so, like, I'm at UIUC and have visited Argonne before for laughs during an open house. I'm well aware with that the APS is :p
<whitequark>
most oscilloscopes come with an FPGA
<whitequark>
most phones come with a very small FPGA or even multiple
<bofh>
s/that/what/
<Qboid>
bofh meant to say: GreeningGalaxy: so, like, I'm at UIUC and have visited Argonne before for laughs during an open house. I'm well aware with what the APS is :p
<GreeningGalaxy>
ah, gotcha
<whitequark>
as a sort of "bundle of wire" thing that allows you to patch stuff together
<GreeningGalaxy>
we were zapping little crystals of GaAs with lasers and 1) watching the strain waves propagate through them by measuring the molecule spacing with the x-rays, 2) melting the crystal and watching it resolidify.
<egg>
whitequark: right
<bofh>
ooh, neat.
<egg>
so lots of tweakable things in embedded stuff
<bofh>
was there an actual purpose for 2? seems a bit strange
<GreeningGalaxy>
Previous experimenters reported that when you whack a piece of semi with a laser, it briefly enters a liquid phase before recrystalizing (even though the temperature is still much too low for it to truly melt). We were unable to reproduce this.
<bofh>
ohh, I see.
<whitequark>
egg: not necessarily embedded
<whitequark>
well, define embedded
<bofh>
by liquid do you mean like nematic order, or?
<whitequark>
cell towers have lots of FPGAs in them
<GreeningGalaxy>
er, that one's above my level, sorry.
<whitequark>
basically anywhere your volume doesn't warrant doing a full-on IC spin
<whitequark>
or where your competence doesnt allow you to get it right in the first place
<bofh>
also totally unrelated but I just learned that in supersymmetric QFT there is a particle called a 'wino'
<egg>
... supersymmetry people ...
<whitequark>
bofh: was that related to the W boson
<GreeningGalaxy>
My real interest is the machines themselves here, and I don't really know a lot about solid-state, x-ray, or other such sub-fields of physics. At least not so much as I do about electronics.
<bofh>
whitequark: yeah, the winos superpartners of the W bosons.
<bofh>
+are the*
<bofh>
(I'm sorry it's hard to type that with a straight face)
<egg>
whitequark: so do you have eggsamples of computing applications?
* egg
likes computing :-p
* egg
likes numerical analysis
<bofh>
GreeningGalaxy: ahh, cool. was just curious.
<bofh>
egg: numerical analysis is like the one subfield of programming that I consistently still like :P
<whitequark>
egg: lots of people trying to train NNs on FPGAs
<whitequark>
various accelerators, e.g. video encoding and such
<whitequark>
many digital cameras contain some sort of FPGA
<whitequark>
egg: ... shouldn't we have discussed that on #openfpga
<whitequark>
:P
<egg>
whitequark: eh
<egg>
whitequark: I did that with lots of eggpuns
<egg>
and Greening's there
<egg>
this channel is topic-free, so this is perfectly appropriate :D
<bofh>
dang, Pointlessness (Mathematics) does not exist
<egg>
GreeningGalaxy: I meant virtually sorry
<egg>
wikipedia.org/wiki/Virtually
<bofh>
I still like the pairs of pants
<whitequark>
GreeningGalaxy: anyway so the thing I got hung on with my GPU is memory access
<egg>
finite = virtually trivial
<GreeningGalaxy>
aha
<whitequark>
the idea is that the ALUs are SIMD
<whitequark>
this means that they all have to be on the same IP to proceed
<whitequark>
but the *memory loads* into them are sequential
<GreeningGalaxy>
whitequark: I saw your lab notebook thing about doing reflow soldering with a blowtorch, btw. that sounds like something I'm about to attempt in the seminear future.
<whitequark>
GreeningGalaxy: get a hotplate :p
<egg>
I'd stay to keep prodding you with stupid questions, but I have been typoing all day because lack of sleep and I'm sntarting to get reallly tyred so zzzzzzz
egg is now known as egg|zzz|egg
<egg|zzz|egg>
= 1
<bofh>
rofl
<GreeningGalaxy>
I have a soldering iron fine enough for SOICs, but nothing much smaller. Might be able to manage TSSOPs if I can find a TENMA soldering iron that works (we have very fine tips for those ones)
<bofh>
resolution of the identity
<whitequark>
GreeningGalaxy: get a "microwave" tip
<egg|zzz|egg>
the eggspectation of the observation of zzz on |egg> is 1
<bofh>
you all and your lack of essential tremor making it possible to do soldering :P
<GreeningGalaxy>
microwave?
<whitequark>
ie a slanted tip with a small cavity on the end
<whitequark>
for drag soldering
<whitequark>
perfect for SMT work
<egg|zzz|egg>
goodnight
<GreeningGalaxy>
ah, I think that's what our really fine-point ones are.
<whitequark>
hakko makes these, or you could take a slanted tip and a dremel
<GreeningGalaxy>
night egg|zzz|egg
<whitequark>
dosnt need to be finepint
<whitequark>
I have a 3mm one i think
<whitequark>
or 2mm
<whitequark>
anyway hotplates rock
<whitequark>
you absolutely want preheat for SMT even if it's literally just a frying pan
<whitequark>
even if you use a blowtorch
<GreeningGalaxy>
bofh: indeed, I am for now blessed with steady hands. I hope it lasts at least a decade or two.
<whitequark>
simplifies your job vastly
<GreeningGalaxy>
alright, I know we have hotplates somewhere
<whitequark>
"Essential tremor is the most common movement disorder; its cause is unknown."
<whitequark>
well that ... must be annoying
<whitequark>
actually my hands shake quite a bit too, just not enough to substantially impair soldering
<whitequark>
I can cope
<bofh>
whitequark: doesn't help that it gets way worse with beta2 adrenergic agonists. guess what has that as a side-effect that I take all the time? :P
e_14159 has quit [Ping timeout: 190 seconds]
<whitequark>
:/
<GreeningGalaxy>
same, my hands aren't remotely tremor-free, I can just dampen down the amplitude small enough to hit a pin on a SOIC without missing if I concentrate
<GreeningGalaxy>
same as whitequark I mean
<bofh>
so essential tremor has that fun property where it does the exact *opposite*
<whitequark>
ohhh
<bofh>
as in my hands normally aren't that unstable, but if I concentrate very strongly on precise movement the amplitude goes way the fuck up
<GreeningGalaxy>
Sometime I want to make some kind of soldering arm thing, kind of like a remote-surgery unit (except not remote) that averages the movements of its control inputs so you can still do fine soldering even if your hands shake a lot
<GreeningGalaxy>
stepper motors, microstepping, and a lot of gearing. As much parts metal as I can manage, but probably a few 3D printed as well.
<bofh>
yeah that sounds like a fairly cool use of feedback control theory
<whitequark>
so a damper
<GreeningGalaxy>
yeah, a damper.
<whitequark>
couldn't you do that with judicious use of mechanics alone
e_14159 has joined #kspacademia
<bofh>
actually huh I wonder if that explains in some ways what is happening with essential tremor: some neural circuit's feedback term is set too high
<whitequark>
yeah that was my thought
* whitequark
drops sleeping
* whitequark
zzz
<GreeningGalaxy>
oh, probably. with a bunch of well-balanced masses and springs, you mean?
<GreeningGalaxy>
night whitequark o/
<egg|zzz|egg>
goodnight whitequark
<UmbralRaptor>
'night whitequark
<bofh>
'night whitequark
<GreeningGalaxy>
I suppose it would also be fun to do teleoperated soldering too, if for no other reason than gee-whiz value
<kmath_>
<romeerizwan> Beautiful moon tonight! Nikon 300 f4 Pf hand held (d500). https://t.co/BDTVMKQo0l
<GreeningGalaxy>
whoa
<icefire>
it better be a good photo when you're shooting it with gear that costs 4 grand
icefire has quit [Read error: Connection reset by peer]
<bofh>
icefire: so I've seen pretty good moon photos taken with old shoddy cellphone photos
<bofh>
old shoddy cellphone cameras*
<bofh>
like, surprisingly good ones :P
<GreeningGalaxy>
of the moon?
<bofh>
yeah
<bofh>
sorry, just got over a slight headache. they had these amaretto cookie thingies here for really cheap and I are like 4 packets
<GreeningGalaxy>
Most phone cameras are pretty wide-angle so that surprises me
<bofh>
then I saw on the ingredients "20% apricot kernel extract" so judging by the headache I suspect they insufficiently decyanidated the damn things
<GreeningGalaxy>
nak
<bofh>
thanks, assholes
<bofh>
this happened to a friend of mine with ~*~ Organic Bitter Almonds ~*~ too
<bofh>
sometimes, you *don't* want what nature gives you
<GreeningGalaxy>
organic bitter almonds, oh god
<bofh>
Organic UNPROCESSED Bitter Almonds
<bofh>
which it did not clearly say on the label
<bofh>
(I'mn not even sure selling that was legal but lol USA "health food" stores)
<GreeningGalaxy>
jesus, doesn't it only take 50 or so of those to kill an adult?
<bofh>
around 50, yeah
<GreeningGalaxy>
lol whoops I just got into another fight about Arch Linux on freenode
<bofh>
my condolences, why would you do that to yourself?
<GreeningGalaxy>
my position is "arch is terrible but manjaro is pretty good" which manages to annoy both main camps.
<bofh>
excellent :D
<GreeningGalaxy>
I discovered that I had a page for `man apt' but no such command `apt' on my system, and someone was like "wow get a sane distro then"
<GreeningGalaxy>
apparently manjaro does not qualify as sane.
<GreeningGalaxy>
I knew I should've gone with womanjaro!
<bofh>
clearly :P
<bofh>
I like how whatever I run tends to entropy into I Can't Believe It's Not LFS after a few years
* soundnfury
sits in the corner muttering 'debian' over and over
<bofh>
(basically, the package manager breaks horribly and it becomes easier for me to just download the .deb's and ar xv, then tar -axvf then than it does to fix it)
<GreeningGalaxy>
ack
<GreeningGalaxy>
I haven't seriously broken anything on this system yet. sudo touch /dev/wood
<GreeningGalaxy>
have you ever used the Enlightenment desktop?
<GreeningGalaxy>
I've tried numerous times but it's always found a way to be fatally broken
GreeningGalaxy has quit [Ping timeout: 201 seconds]
<SnoopJeDi>
Someone in freenode's #python really loves NixOS and has been promoting it a lot. It seems...interesting.
TentacularGalaxy has quit [Remote host closed the connection]
egg|laptop|egg has quit [Ping timeout: 180 seconds]
<egg|zzz|egg>
ooooooookay wtf
<egg|zzz|egg>
replaying a journal with the same binaries, we get a result that differs by a couple of ULPs from what was recorded in the journal
<egg|zzz|egg>
how
<egg|zzz|egg>
why
<Thomas>
Magic
<egg|zzz|egg>
e_14159: ok, so for the 9.8 g_0 thing it's me using the dry masses I think
* egg|zzz|egg
stabs egg
<egg|zzz|egg>
now all my parts are called "Part" though, so that's an issue :-p
<egg|zzz|egg>
(more of ~KSP API~: Part has a mass, a physicsMass, a resourceMass, a rb.mass, and a prefabMass)
<egg|zzz|egg>
though that doesn't answer the much weirder FP off-by-one-ULP journaling question
<egg|zzz|egg>
does *KSP* touch the floating-point environment?
GreeningGalaxy has joined #kspacademia
zlaing has left #kspacademia [#kspacademia]
<egg|zzz|egg>
!wpn GreeningGalaxy
* Qboid
gives GreeningGalaxy an intensive trapezohedron
<GreeningGalaxy>
!wpn |egg⟩
* Qboid
gives |egg⟩ a cyclic Abelian spectrometer with a field extension attachment
<egg|zzz|egg>
:D
<GreeningGalaxy>
what are the eigenvalues of the operator zzz, anyway?
<e_14159>
egg|zzz|egg: Is there documentation which mass is which?
<GreeningGalaxy>
wait, I suppose the eigenvalues would be everything egg says while zzz
<egg|zzz|egg>
e_14159: the documentation is silent on the subject
<e_14159>
Because of course it is.
<GreeningGalaxy>
ive been watching videos about how to build a trap for mice/rats, for some reason
<egg|zzz|egg>
a cat?
<GreeningGalaxy>
ideally alive
<GreeningGalaxy>
also if the rats around here become a problem, I doubt most housecats are strong enough to fight those.
<GreeningGalaxy>
clearly I just need a Norwegian Forest
<GreeningGalaxy>
and of course if I do catch mice live, I have to figure out what I'm going to do with them. I obviously can't just release them on the sidewalk out front or they'll be back in minutes (or in someone else's house)
<GreeningGalaxy>
might be better to just get a cat and let it eat them
<egg|zzz|egg>
hmm, the problem with that 6-week release cycle idea is that it's insufficiently silly; every full moon would be more appropriate, but too frequent...
<egg|zzz|egg>
*looks at periods of moons of the solar system*
<egg|zzz|egg>
blarg
<bofh>
oh man chemical-free coconut oil
<e_14159>
bofh: Chemistry-free coconut oil?
<egg|zzz|egg>
bofh: btw I forgot about two of our actual people in academia here, e_14159 and Iskierka
<Qboid>
egg: I added the explanation for this acronym.
<soundnfury>
!acr -add:QBD Quantum BogoDynamics
<Qboid>
soundnfury: I added the explanation for this acronym.
<egg>
pizzaoverhead: I guess after Cardano we'd want to look at utilities for rendez-vous
<egg>
because right now it's just about impossible
GreeningGalaxy has joined #kspacademia
<GreeningGalaxy>
I have one of those professors whose primary goal, instead of "teach the material," is "bust as many students' balls as possible"
<GreeningGalaxy>
he doesn't provide help, he provides punishment. If you don't know something, that doesn't mean you should work on learning it; to him that means you should feel very bad and not pass the course.
<egg>
whitequark: I noticed that you can use fullwidth slashes in identifiers in C++ (because they are bad at standards)... I have vaguely pondered using that as an alternative to _over_
<egg>
but it seems morally wrong, so I think I won't :-p
<soundnfury>
who are you, and what have you done with egg?
<soundnfury>
:P
<egg>
I'm egg, and therefore I care about the documents referenced by the note in section 1 of ISO/IEC 10646:2003 :-p
<egg>
(C++ doesn't, because contrary to WG 9, WG 21 doesn't have someone who came up with that way of referring to Unicode)
<whitequark>
egg: did you know that if you put a slash into a file name in dolphin (kde's file manager) it transforms it into a fullwidth slash?
<whitequark>
that's actually super handy because it lets me just copy and paste paper names from pdfs
* egg
slaps pizzaoverhead with whitequark's baryonic trigraphs
<egg>
whitequark: fraction slash is not allowed in identifiers
<pizzaoverhead>
:D
<egg>
see, C++'s handling of ISO/IEC 10646 is neither Unicode-aware nor in good taste
<egg>
I would be ok, I would even like, some SYMBOL characters in identifiers
<egg>
e.g. to use INCREMENT rather than GREEK CAPITAL LETTER DELTA
Thomas is now known as Thomas|AWAY
<egg>
or N-ARY SUMMATION rather than GREEK CAPITAL LETTER SIGMA
<egg>
but I cannot
<egg>
and otoh, I'm pretty sure FULLWIDTH SOLIDUS in an identifier is a bad idea
<bofh>
LOL
<bofh>
Is there any reason to ever use "Increment" over "Captial Delta"? Never seen the former used and some fonts even lack it as a codepoint despite having the latter in it.
<bofh>
(Easy fix with ttx since it just involves adding a CMap entry, but ugh)
<bofh>
> 18:30 <@GreeningGalaxy> I have one of those professors whose primary goal, instead of "teach the material," is "bust as many students' balls as possible"
<bofh>
I *really* hate these sorts of profs.
<egg>
bofh: well, when I use XeLaTeX it emits that
<egg>
bofh: as for reason to use, the same reason to use capital alpha over capital a :-p
<egg>
if what you need is an operator and not the letter delta, use that
egg is now known as egg|nomz|egg
<bofh>
well ultimately it all winds up being the same GID anythow
<bofh>
anyhow*
<egg|phone|egg>
Not always
<egg|phone|egg>
Some fonts distinguish
<egg|phone|egg>
Also nomz
<bofh>
between A and capital Alpha?
<bofh>
how?
<GreeningGalaxy>
bofh: he's a right bastard. And also my faculty advisor. >_<
<GreeningGalaxy>
my research prof is my de facto advisor for most things, but anything that involves actual clerical work has to go through him.
<bofh>
my condolences
<GreeningGalaxy>
we had several class periods throughout the quarter where almost everyone in the room was like "You're going too fast, we just barely started on this worksheet and now you're saying we should be done" and he consistently responded by explaining how we'd never pass the physics GRE if we didn't get faster at doing things.
<GreeningGalaxy>
1) not all of us in that class are even taking the physics GRE. Probably not even half. 2) fuck the physics GRE.
<bofh>
the physics GRE is a pile of ass and ass
<bofh>
holy fuck
<bofh>
waste of time
<GreeningGalaxy>
it's always his favorite thing to talk about. He even threatens the freshmen with it.
<bofh>
(as was the math subject GRE, but that one was at least marginally *almost* fun?)
<bofh>
so like it's not even required for a lot of places!
<Qboid>
soundnfury: I already know an explanation for GRE! (Update it with !acr -update:GRE Generic Routing Encapsulation)
<soundnfury>
GRE?
<Qboid>
soundnfury: [GRE] => Gamma Ray Explorer
* soundnfury
rolls eyes
<GreeningGalaxy>
right, like, I'm going into EE, and everyone tells me that the only subject GRE I should need is the math one.
<GreeningGalaxy>
anyway I made sure to give the guy the worst evaluation I've ever given this quarter. Hopefully it results in some nonzero change, but he's very tenured and probably up next to be the department chair so lol shrug
icefire has quit [Quit: Leaving]
<bofh>
rofl, I've never had a prof eval be of any ue
<bofh>
use*
<bofh>
okay, lie, I've had very positive ones I left for a prof in 3 courses I had him in help to give him a distinguished teaching award
<bofh>
but I've never had a negative one be of any use, regardless of if it was polite and detailed their shortcomings with possible ways to improve from there
<bofh>
or if it was basically me channelling Samuel-ʟ-Jackson
<soundnfury>
bofh: you've had it with these monday-to-friday lectures from this maths-faculty prof?
<egg|nomz|egg>
!wpn UmbralRaptor
* Qboid
gives UmbralRaptor a Euclidean checksum
<egg|nomz|egg>
!wpn GreeningGalaxy
* Qboid
gives GreeningGalaxy a [DATA EXPUNGED] beta dimetrodon which vaguely resembles a photodiode
<egg|nomz|egg>
!wpn bofh
* Qboid
gives bofh a state-of-the-art thyristor
<egg|nomz|egg>
!wpn whitequark
* Qboid
gives whitequark a sodium Schläfli automorphism
<egg|nomz|egg>
ooh Schläfli
* egg|nomz|egg
isn't really familiar with his notation, is more used to Coxeter diagrams
<egg|nomz|egg>
UmbralRaptor: so after discussing that we've elected to release every full moon following Cardano
<egg|nomz|egg>
that way we'll get through mathematicians faster
<kmath_>
<whitequark> If you condense many couples on the same vessel do you get a poly(relationship)
<egg|nomz|egg>
whitequark: I... I don't think I get it?
<egg|nomz|egg>
but I'm a bit slow right now
egg|nomz|egg is now known as egg
<whitequark>
egg: so poly(x) is a chemical notation for a polymer, e.g. "poly(methyl methacrylate)" for what's colloquially known as acrylic
<egg>
ah
<egg>
right, wasn't thinking about chemistry
<whitequark>
and polycondensation, along with polyaddition, is a way many polymers form
<whitequark>
... and having many couples on the same (seafaring) vessel is known as a cruise
<egg>
aaah right
<whitequark>
approximately half of what motivates me to learn german is the ability to make german/russian/english cross-puns
<egg>
whitequark: the one time I was on a cruise was in the Galapagos, where the point was to stare at strange animals rather than to be an element of a set of tuples
<whitequark>
I've never been on one
<egg>
whitequark: I might have some photographs of fun animals from that
<whitequark>
mmm
<soundnfury>
"element of a set of tuples"? So, you're being a bit more inclusive than "node of a bipartite graph", huh?
<whitequark>
lol
<whitequark>
soundnfury: wait I don't think "bipartite graph" is... right... here
<egg>
yeah, you mean matching
<egg>
or...
<egg>
or maybe you don't? who knows
<egg>
whitequark: uh, so I'm looking at those photos, apparently I have some nsfw tortoise imagery
<whitequark>
egg: um
<whitequark>
define nsfw
<egg>
well probably nsfw if you were a tortoise
<egg>
(i.e. if I'm parsing that photograph well, those tortoises are shagging)
<soundnfury>
whitequark: bipartite graph comes from the assumption that all the passengers are hetero but not all of them are faithful
<egg>
ah, right.
<soundnfury>
(functors?)
<egg>
were they all faithful you'd get a matching
<whitequark>
boring
<egg>
ah yes of course faithful means something in math
<pizzaoverhead>
egg: Quote of the day.
<egg>
which has nothing to do with anything
<egg>
pizzaoverhead: which one?
<soundnfury>
I could have phrased it better, "not all of them are faithful, and none of them are homomorphisms" perhaps
<egg>
whitequark: also there are boobies, of the blue and red footed variety, as well as nazca
<pizzaoverhead>
"if I'm parsing that photograph well, those tortoises are shagging" amuses me greatly.
<whitequark>
hot tortoise-on-tortoise action
<egg>
red-footed boobies are really funny looking, with their prehensile webbed feet
<egg>
looks like they have rubber sheets folded around the tree branch :-p
<egg>
some albatrosses too
<whitequark>
>prehensile webbed feet
<pizzaoverhead>
egg: So there are boobies in your NSFW animal photo collection?
<pizzaoverhead>
:D
<egg>
pizzaoverhead: the boobies are quite sfw, there are nesting pairs (of Nazca boobies and of albatrosses), but they're not doing much
<egg>
BLARGLBALREB still this floating point off by an ULP issue on replay
<egg>
even with the same binary
* UmbralRaptor
has heard ice cream trucks on rainy Februaries in Kansas.
<egg>
stabbity stabbity stab
<UmbralRaptor>
Unity flaw?
* egg
ponders votive offerings to Kahan
<UmbralRaptor>
Do you have a shrine to Dark Lord Kahan yet?
<bofh>
I do
<egg>
whitequark: btw, something that came up in discussion around these part a while ago: what kind of a magnetic field would you need to have for ferrite core to undergo bit flips
<egg>
(what mad celestial bodies can you not fly around with ferrite core :-p)
<whitequark>
egg: I'm not sure
<whitequark>
look at the hysteresis diagram?
<pizzaoverhead>
Has anyone modelled a magnetic field in KSP? I've been considering doing a magnetorquer mod for a while.
<egg>
has somebody modelled bit flips in KOS :-p
<whitequark>
just do numeric QFT simulation for every part
<egg>
write your KOS autopilot so it goes into safe mode
<whitequark>
given how dreadful the KSP UI is...
<egg>
I've looked at KOS once, saw the language, and went to do other things never to look at it again :-p
<whitequark>
I wonder if I should just use it as the level editor and do the rest from CLI
<whitequark>
yeah, the language is bad.
<whitequark>
fortunately other languages can target its bytecode... maybe?
<whitequark>
the bytecode is also bad but slightly less bad
<egg>
much like you, I'd probably want to write a compiler front-end; contrary to you, I'm not competent to do that
<egg>
also I'd probably want to write an Ada compiler for it, because Ada & aerospace