UmbralRaptor changed the topic of #kspacademia to: https://gist.github.com/pdn4kd/164b9b85435d87afbec0c3a7e69d3e6d | Dogs are cats. Spiders are cat interferometers. | Космизм сегодня! | Document well, for tomorrow you may get mauled by a ネコバス. | <UmbralRaptor> … one of the other grad students just compared me to nomal O_o | <ferram4> I shall beat my problems to death with an engineer.
<bofh>
UmbralRaptor: also let's add a visual albedo of 0.72, b/c what
<UmbralRaptor>
!wa Pluto albedo
<Qboid>
UmbralRaptor: Pluto | albedo: 0.3
<UmbralRaptor>
O_o
<UmbralRaptor>
What indeed.
<Iskierka>
... I thought pluto had a bizarrely high albedo?
<bofh>
is that visual or Bond?
<bofh>
!wa Triton albedo
<Qboid>
bofh: Triton | albedo: 0.756
<Iskierka>
or has it turned out its moons do instead and we added it up?
<Iskierka>
!wa charon albedo
<Qboid>
Iskierka: Charon | albedo: 0.372
<bofh>
!wa Charon albedo
<Qboid>
bofh: Charon | albedo: 0.372
<bofh>
oops
<bofh>
like that's a bit high
<Iskierka>
and the rest shouldn't be large enough to matter much. odd
<Iskierka>
!wa moon albedo
<Qboid>
Iskierka: Moon | albedo: 0.12
<Iskierka>
I suppose it's bright compared to that
<bofh>
still, 0.76 is absurd
<bofh>
WAIT WHAT
<bofh>
!wa Enceladus albedo
<Qboid>
bofh: Enceladus | albedo: 1
<UmbralRaptor>
Pluto has a high albedo for a KBO.
<UmbralRaptor>
IIRC
<Iskierka>
that's gotta be incorrect data
<bofh>
for what, Enceladus? NASA factsheets gives the same one
<Iskierka>
"The fresh, clean ice that dominates its surface gives Enceladus the most reflective surface of any body in the Solar System, with a visual geometric albedo of 1.38[7] and bolometric Bond albedo of 0.81±0.04."
<Iskierka>
dead on 1 is nonsense. there should at least be imperfections
<bofh>
okay, true. I think they rounded it way too much
<Iskierka>
or possibly took the visual geometric and clamped it, when that's not one that does have to be limited to 1
<Iskierka>
wiki also says pluto is variable .49-.66, which is brighter than wolfram thinks
<bofh>
well let's see what the web interface sez
<bofh>
trying to specify 'visual albedo', 'visual geometric albedo' or 'Bond albedo' all give me "using closest known expression 'albedo'"
* Qboid
gives whitequark a bismuth geobukseon-like ring
Guest has joined #kspacademia
Profound has quit [Ping timeout: 383 seconds]
Moistmelon has quit [Ping timeout: 183 seconds]
Profound has joined #kspacademia
Guest has quit [Ping timeout: 195 seconds]
Guest has joined #kspacademia
Profound has quit [Ping timeout: 186 seconds]
<egg|zzz|egg>
bofh: why is sqrt(-0) not NaN
<egg|zzz|egg>
bofh: on the one hand I guess if im(complex sqrt(z)) = 0 then the real sqrt might be expected to return z, on the other hand Kahan's much ado about nothing's sign bit seems to like distinguithing +0i and -0i from no imaginary part at all (and same for real parts)
<egg|zzz|egg>
kahanwishes: the expression CMPLX (0,8)
<egg|zzz|egg>
should be treated as is, whereas CMPLX(+O,S) and CMPLX(-O,8)
<egg|zzz|egg>
should be treated as intentional attempts by the programmer to
<egg|zzz|egg>
introduce an appropriately signed zero into the calculation.
<kmath>
<bofh453> <egg> every increasing ω-sequence of screams converges to a limit in the long scream <bofh> I think you just explained US politics twitter.
<UmbralRaptor>
hah. AAAAAAAAAAAAAAAA!
<UmbralRaptor>
"Methyl iodide is a reagent used for O-methylation, like dimethyl sulfate, but is less hazardous…"
<UmbralRaptor>
uhm
<egg|zzz|egg>
!wpn UmbralRaptor
* Qboid
gives UmbralRaptor a coffee
* UmbralRaptor
drinks the coffee.
<egg|zzz|egg>
!wpn bofh
* Qboid
gives bofh an ultraviolet death
<bofh>
!wpn egg
* Qboid
gives egg a proactinium mass-driver
<bofh>
UmbralRaptor: that's literally like saying "sucralose is a chemical used for artificial sweetening, like lead acetate, but is less hazardous"
<UmbralRaptor>
The context makes it sound like reduced hazard is a downside.
<egg|zzz|egg>
bofh: sqrt(-0) is weird
<egg|zzz|egg>
bofh: also much ado about nothing's sign bit gives sqrt(-β+/-i0)=+0+/-i sqrt(β) for all β >= 0, so how is sqrt(-0)=-0 Ꙩ_ꙩ
<bofh>
egg|zzz|egg: analytically continue sqrt(z) along the principal branch, then consider what you need to make sqrt(z), z->0- in order to get continuity there.
<kmath>
<pentaquark7364> Protip: do not be the person that denotes the set of quaternions by ℚ. Nobody likes this person. Yes, I have actually seen this done.
<kmath>
<1aprildaniels> So, now that I'm banned from the military, can I stop paying taxes? I'm not a full citizen anymore, I'd like a tax break in exchange.
<egg|zzz|egg>
bofh: hmmm
* UmbralRaptor
ponders.
<egg|zzz|egg>
bofh: also, aaaaa... at that notation for the quaternions
<egg|zzz|egg>
that's what the letter H is for!
<bofh>
EXACTLY
* UmbralRaptor
H was for henergy?
<bofh>
double-struck H
<UmbralRaptor>
ah
<bofh>
ℍ, even.
<egg|zzz|egg>
bofh: ok, if you want continuity at 0 you need it to be 0, but that's a purely imaginary 0, why give its real part a - sign
<egg|zzz|egg>
UmbralRaptor: same H. tho
<egg|zzz|egg>
for the definitions of ℍ and ?, see the works of H.
<egg|zzz|egg>
bofh: especially since (at least Kahan's) complex sqrt gives it a positive real part
<bofh>
egg|zzz|egg: so I *believe* after thinking about it now the reason is to be able to tell whether you underflowed from a small +ve or a small -ve number, but not sure why that behaviour specifically instead of mapping the latter to pi*i
<egg|zzz|egg>
the cat is hardly being informative here tbh
<bofh>
ah, hadn't read it yet. should re-poke him.
<bofh>
but it did make the question make sense to me now
<bofh>
since I realized you're talking about sqrt(x), whereas mentally I was working with sqrt(z) >_>
<egg|zzz|egg>
bofh: yeah, Kahan's blurb is quite eggsplicit about the fact that pure reals and pure imaginaries are not to be treated like arbitrary complexes :-p
<bofh>
come to think of it, I do think it's a symmetry/limits argument after call, consider sqrt(x +/- 0i)
<egg|zzz|egg>
bofh: though his CMPLX(0,β) not being CMPLX(+0,β) nor CMPLX(-0,β) sounds a bit wishful :-p
<egg|zzz|egg>
overload operator+, operator-? :D
<egg|zzz|egg>
(combined with user-defined literals I guess)
<bofh>
LOL. seems like way too much work IMO.
<egg|zzz|egg>
bofh: iirc Ada has an imaginary type, so you can do things with pure imaginaries
<egg|zzz|egg>
I mean, that makes sense in some places
<egg|zzz|egg>
bofh: much ado about nothing's sign bit also suggests that infinities be supplied with an angle, to avoid the silliness of the 9 infinities :D
<bofh>
so I'm predominantly actually used to Fortran's complex number implementation, which Kahan disparages since it's literally just a struct _complex_s { double x, y; } __attribute__((packed)) complex_t;
<kmath>
YouTube - This Mushroom Starts Killing You Before You Even Realize It | Deep Look
<egg|zzz|egg>
bofh: I think it has pretty much won anyway, that's what C went with too, right
<egg|zzz|egg>
or is it implementation-defined
<bofh>
Huh, TIL the North American destroying angel is Amanita Bisporigera, not Amanita Virosa
<egg|zzz|egg>
Each complex type has the same object representation and alignment requirements as an array of two elements of the corresponding real type (float for float complex, double for double complex, long double for long double complex). The first element of the array holds the real part, and the second element of the array holds the imaginary component.
<egg|zzz|egg>
says cppreference
<bofh>
oh so C99's *is* the same
<egg|zzz|egg>
so it's the usual cartesian thing
<egg|zzz|egg>
bofh: I mean directed infinities are cute but being able to add is cool too :-p
<bofh>
go figure. I always roll my own in C, due to lack of support in the past (and nowadays also to avoid calls to _mulsc3 which is idiotic)
<bofh>
touché
<egg|zzz|egg>
that's basically the alternative, right? polar + trig every time you want to add numbers
<bofh>
yyyyyyyyyyyyyyyyyyyep
<egg|zzz|egg>
bofh: otoh if we had that we wouldn't be stuck with the weird-arse behaviour of 2-parameter arc tg on signed 0s that only makes sense because of arg's algebraic properties
<egg|zzz|egg>
(which is of course explained in much ado about nothing's sign bit :-p)
<egg|zzz|egg>
bofh: phl actually bought a book in 1989 to get a copy of much ado about nothing's sign bit
<egg|zzz|egg>
bofh: uh, what's with sqrt(x +/- 0i)?
<bofh>
egg|zzz|egg: consider limits of x near 0 and what you want the symmetry in C to look like when restricted to R
<egg|zzz|egg>
bofh: uh, I don't get it
<egg|zzz|egg>
no matter what you don't get a negative real part?
<bofh>
egg|zzz|egg: so again, none of this makes sense in ℂ because the principal square root *in ℂ* always has a positive real part.
<bofh>
but real sqrt does a different quadrant mapping it seems
icefire has joined #kspacademia
<egg|zzz|egg>
bofh: right, but then if you restrict to reals why does sqrt(-0) even make sense
<bofh>
I'm now honestly half-suspecting sqrt(-0) being -0 is literally just to save a branch in a special case :P since I'm actually not sure why they chose this convention.
<egg|zzz|egg>
bofh: hmmm, I guess you still want to define it because sqrt(-(subtraction that usually is negative)) would be unpleasant if it NaNed
<egg|zzz|egg>
bofh: and then -0 so you can inspect whether it's that weirdness or the normal thing?
<egg|zzz|egg>
bofh: but then that make's atlas's answer entirely inaccurate as well as uninformative, which seems odd
<egg|zzz|egg>
I mean, that's 4-year-ago-atlas, but still
<kmath>
<bofh453> @stephentyrone Math question this time (& sorry for so many questions!), but basically which complex results are si… https://t.co/RaismCk51w
<egg|zzz|egg>
bofh: maybe if we talk about numerics and post catpics enough Atlas will spontaneously appear in this channel :-p
<egg|zzz|egg>
bofh: note the lowercase r, it's an adjective and there's no such thing as a proper adjective
<egg|zzz|egg>
Cartan uses that term
<egg|zzz|egg>
"La théorie de M. Einstein, réduite à ce qu'elle a d'essentiel, ramenait à la Géométrie la théorie physique de la gravitation dans le vide."
<bofh>
rofl. what would a proper adjective even be like? :P
<egg|zzz|egg>
bofh: well Riemannian or French or whatnot take capitals in English, right?
<bofh>
yeah, I think in English any adjective derived from a proper noun tends to take a capital in English.
<egg|zzz|egg>
bofh: whereas riemannienne, française, etc. don't
<bofh>
huh. interesting.
<egg|zzz|egg>
bofh: even if they're made into nouns!
<egg|zzz|egg>
e.g. l'anglais for either the language or "the guy who is English"
<egg|zzz|egg>
bofh: yeah so it makes some sense to return a 0 for sqrt(-0), (even though the sensible thing from the point of view of underflow might be NaN), since e.g. sqrt(-log(x)) for x<=1 probably shouldn't blow up in your face; but -0 I really can't find a rationale aside from "here's a free bit to tell you something got weird"
<egg|zzz|egg>
bofh: and tbh sqrt(-0) = NaN wouldn't be that awful, 1/-log(x) or exp(-log(x)) are going to screw up at x=1 anyway
<bofh>
No, I disagree, NaN would be daft there. There are only two reasonable choices and they are +0 and -0
<egg|zzz|egg>
bofh: hm, not sure what makes -0 reasonable :-p
<egg|zzz|egg>
bofh: Ꙩ_ꙩ there's a pow on [−∞, +∞] × [−∞, +∞] and a powr on [0, +∞] × [−∞, +∞] in the ieee recommended functions
<bofh>
so mostly I want my fucking ipow function to always exist without having to roll it myself
<bofh>
on [−∞, +∞] x ℤ
<egg|zzz|egg>
bofh: you mean pown?
<egg|zzz|egg>
the standard also provides that
<egg|zzz|egg>
bofh: but pow vs. powr (with different exception behaviour etc.) are weird
<egg|zzz|egg>
bofh: it also provides rootn
<egg|zzz|egg>
bofh: yeah I guess the standard has nothing that NaNs based on the sign of 0 so that would be obnoxious
<bofh>
is there any reason for rootn? I mean n being arbitrary means you can't nicely implement Newton's Method
<bofh>
because you'd need to symbolically differentiate the expression for it
<bofh>
powr seems strange and nonsensical IMO
<egg|zzz|egg>
bofh: so like, just because something is in the standard doesn't mean it's not daft or at least weird, (ahem sNaN), which is why I'm asking about -0
<bofh>
it pops up often but there's no easier way to implement it than pow(x,1/n)
<bofh>
so why not just use that?
<egg|zzz|egg>
bofh: hmm, bad conditioning of pow for small x maybe?
<bofh>
erm, I thought pow is mandated to not be badly conditioned for small x, hence why it's so fucking slow
<egg|zzz|egg>
bofh: um, I do not mean ill-conditioned as a term of abused
<egg|zzz|egg>
s/.$//
<Qboid>
egg|zzz|egg meant to say: bofh: um, I do not mean ill-conditioned as a term of abuse
Moistmelon has joined #kspacademia
<egg|zzz|egg>
bofh: I mean it as in, the condition number of x^y in y is y log x
<egg|zzz|egg>
pow is mandated by the standard to be correctly rounded (in practice you're lucky if it's faithfully rounded), but that doesn't make it well-conditioned
<egg|zzz|egg>
bofh: so if you have a well-rounded pow it's hard to make a well-rounded nth root for sizeable n
<egg|zzz|egg>
bofh: there's also the condition in x, which is y
<bofh>
yes but my point in 18:27 is that the condition number of rootn will be 1/n * log(x)
<egg|zzz|egg>
bofh: yes but at that point you won't be taking any rounding errors from the division 1/n
<bofh>
because there is no way of implementing rootn other than the same thing you'd use to implement pow(x, y) for *general* y.
<egg|zzz|egg>
bofh: so, like, implementing it is another question, but if you just do pow(x, 1/n) with a correctly-rounded pow, you haven't done what the standard says
<egg|zzz|egg>
bofh: also why can't you newton it for arbitrary n?
<bofh>
because your expression for Newton for each n differs
<egg|zzz|egg>
how so
<bofh>
you have \frac{1}{n} * (x^{-n} - a)*(x^{n+1}) -> \frac{1}{n} * (x - a*x^{n-1}) okay I suppose you can do that with a loop
<bofh>
I was for some reason thinking of Halley's method where you have both second derivatives and a subtraction on the denominator and so the expression rearrangement becomes nontrivial
<egg|zzz|egg>
bofh: well a pown, but aside from that it looks nice
<bofh>
I suppose for second-order Householder methods you're okay, go figure.
<egg|zzz|egg>
bofh: but the question of how to implement rootn seems different from the question of why you would want a rootn (vs. just using pow), and the only answer I can find is the rounding error in 1/n getting amplified by pow for small or large x
<egg|zzz|egg>
bofh: the standard also recommends sinPi and cosPi, that's fairly understandable
<bofh>
PLEASE GIVE ME A FUCKING SINPI AND COSPI, I currently have to implement those myself.
<bofh>
Thankfully it's just a case of lifting k_sinf.c, k_cosf.c and a dramatically simpler argument reduction routine compared to what you need for sin(x)/cos(x).
<bofh>
like those two are easily the things I miss from libm the most by far
<egg|zzz|egg>
bofh: well the standard recommends those! (except it recommends them correctly-rounded, which seems hard)
<egg|zzz|egg>
There is a nontrivial elementary embedding of L(Vλ+1 ) into itself with the critical point below λ :D
egg|zzz|egg is now known as egg|nomz|egg
<bofh>
like I have an implementation, it's trivial, minimax polynomial on [0,1], Chebyshev series on (1,8), asymptotic minimax polynomial in 1/x on [8,\infty)
<bofh>
just if you're going to give me J0 (useful) and Y0 (useless), give me I0 as well ffs
<bofh>
(I0 is an even function so we just need to consider [0,\infty))
<bofh>
sorry, minimax polynomial on (0,1], the value 1 for the input 0 :P
<bofh>
+**
<SnoopJeDi>
I am so confused by this storm's classification vs time. Weather is neat.
<bofh>
I have a copy, in fact I think I grabbed it from exactly that site.
<egg|nomz|egg>
bofh: home directories of academics, one of the best ways of getting papers...
<bofh>
yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyep
<egg|nomz|egg>
UmbralRaptor: anything fun in the sky lately?
<UmbralRaptor>
A moon?
<egg|nomz|egg>
UmbralRaptor: less than a degree of altitude
<SnoopJeDi>
egg|nomz|egg, LOL
<SnoopJeDi>
also woo UMBC \o/
* SnoopJeDi
has Retriever fever
<UmbralRaptor>
Hrm
<UmbralRaptor>
M15? Coathanger? M27?
<egg|nomz|egg>
!wpn bofh, SnoopJeDi, UmbralRaptor
* Qboid
gives bofh, SnoopJeDi, UmbralRaptor a europium suitcase
<bofh>
ooh, always wanted one of those :D
<SnoopJeDi>
I think 3 academics sharing one suitcase will make things a bit...
<SnoopJeDi>
...packed.
<UmbralRaptor>
hexagonally close?
<SnoopJeDi>
HCP nation
<egg|nomz|egg>
bofh: okay I'm confused, is this 1. a reference to whitequark's tweet pattern about molecules that have 2 different applications 2. a pre-existing meme 3. convergent evolution https://twitter.com/Zerochan/status/901174987121319936
<kmath>
<Zerochan> I'm at the MAGwest I'm at the CrunchyCon I'm at the combination MAGwest and CrunchyCon
<SnoopJeDi>
MAG* has been playing lingual games for a few years AIUI, but I'm not super plugged into it, only went to MAGFest once (but what a once it was!)
<UmbralRaptor>
1 is a subset of 2, I think?
<UmbralRaptor>
Also, bets on CrunchyCon being the next DashCon?
<egg|nomz|egg>
UmbralRaptor: what are they
<SnoopJeDi>
a-apparently "crunchy conservative" is a thing and it makes this difficult to google
<SnoopJeDi>
Thanks. happily not for us, but a county not far off. We had some rotation at 8 CDT but I don't think anything visual came of it
<egg|nomz|egg>
!wpn Fiora
* Qboid
gives Fiora a pink automorphism
<SnoopJeDi>
The one that touched down here last year was about a mile from the lab (where we were at the time)
<SnoopJeDi>
People ask me about living down here and tornados and I'm usually like "ah we don't get 'em, that's more a thing for Dallas," but there have been 2 + whatever comes of this since I've lived here :|
<UmbralRaptor>
They just want to be your friend.
<bofh>
^
<SnoopJeDi>
I have at least confirmed to my personal satisfaction the null hypothesis in the case of dating a meteorologist and attracting severe weather.
<kmath>
<FioraAeterna> crunchyroll expo just committed the cardinal sin of cons their security is confiscating all drinks. even those bought here. IN SUMMER.
<bofh>
So a bunch of friends in NYC have a theory that winter storms tend to follow me since every time I've been there it's been like a day before the biggest winter storm they've seen that year
<bofh>
or like, the only appreciably cool weekend in the entire summer, that got absurdly hot again immediately after I left
<bofh>
I am 100% okay with this.
<egg|nomz|egg>
Fiora: Ꙩ_ꙩ
<SnoopJeDi>
ew, why would they do this
<SnoopJeDi>
I mean I know why but still don't be an idiot
<SnoopJeDi>
bofh, when I moved here in 2011, the very bad drought broke two days later, and shortly after that was the VA earthquake. I enjoyed some idle streetlamp phenomenon thoughts
<kmath>
<jgerity> It's like the Jargon File but half the entries are stupid I love it https://t.co/jDghgz5U9c
<SnoopJeDi>
The author and I swapped some eclipse tweets courtesy of Alex Parker's thread on how scientists were wholly unprepared for the emotional shock of totality, and I stumbled into it
<SnoopJeDi>
had not ever heard of this Devil's Dictionary but it looks very much like something I need
<kmath>
<DrEugeniaCheng> Step 1: Read newspaper article about new scientific research. Step 2: Download research paper. Step 3: Find no resemblance between the two.
* UmbralRaptor
?
<egg|nomz|egg>
bofh: I avoid doing 1. tbh
<egg|nomz|egg>
bofh: well unless you count Nature as a newspaper
<SnoopJeDi>
is it anything but?
<SnoopJeDi>
(hah, take *that*, Nature!)
<bofh>
LOL
<UmbralRaptor>
It's frustrating -- specialized sites like Universe Today, Planetary Report, etc. get things right.
<UmbralRaptor>
Also Astro Bites, but that's probably cheating.
<egg|nomz|egg>
UmbralRaptor: also Nature :-p
<UmbralRaptor>
egg|nomz|egg: I'm not sure telling me to pay $35 or get a subscription for $199 counts as being about the research.
<egg|nomz|egg>
UmbralRaptor: well if you count it as a newspaper you pay for it might fit the criteria for step 1 :-p
<egg|nomz|egg>
(I usually read it on the toilet, phl has as subscription)
<SnoopJeDi>
A newspaper that charges contributing authors, heh.
<UmbralRaptor>
At least it hasn't pivoted to video.
<kmath>
<RohacsTibor> @sq_microbio Have this posted in my lab, not sure about source, but very useful at every level. More practical than… https://t.co/JXd2s055kM
<SnoopJeDi>
As far as I can tell, most folks in STEM have very little interest in branching out to begin with
<SnoopJeDi>
Even free beer cannot get the local grads to send me one thing they read that week that they thought was worth sharing
<bofh>
SnoopJeDi: fucking astounding
<SnoopJeDi>
one of the worst cohort experiences I had in ugrad was coming out of a class (quantum?) talking excitedly with 2 or 3 people about something or other and one of them just kinda put their hand up and was like "whoa man, we're out of class, cool it" or something like that
<SnoopJeDi>
physics student does not want to discuss physics (or misc) outside of physics classroom? brain.exe has stopped working
<UmbralRaptor>
…
<UmbralRaptor>
wat
<bofh>
SnoopJeDi: https://arxiv.org/abs/1708.04624 like here's a cute result I randomly read yesterday, courtesy @CondensedPapers
<UmbralRaptor>
Silly thing that happened at a bar at EPRV: we tried to talk about other things, buy ended up circling around to astro.
<SnoopJeDi>
creatures of habit are we, UmbralRaptor :)
<bofh>
indeed :)
<SnoopJeDi>
bofh, can you elevator-talk those? The abstracts are obnoxious and I'm a bit frazzled. (and lousy with anything CM on my best day!)
<UmbralRaptor>
bofh: I, uh, recognize some of those words? =S
<UmbralRaptor>
But it sounds like there are surfaces where superconductivity gets silly, sorta like crystal structures do?
<bofh>
fundamentally, yeah.
<bofh>
SnoopJeDi: sure, sec
<SnoopJeDi>
also wow the entry for WEATHER: "The setting up official weather bureaus and their maintenance in mendacity prove that even governments are accessible to suasion by the rude forefathers of the jungle."
<UmbralRaptor>
!wpn -add:wpn collapsar
<Qboid>
UmbralRaptor: Weapon added!
* egg|nomz|egg
hears insect noises above, looks up, insect is being eaten by some sort of pholcid