egg|nomz|egg 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> egg|nomz|egg: generally if your eyes are dewing over, that's not the weather. | <ferram4> I shall beat my problems to death with an engineer.
<whitequark>
!wpn egg
* Qboid
gives egg a theta compass
<egg|zzz|egg>
!wpn whitequark
* Qboid
gives whitequark a bzip'd minion
* egg|zzz|egg
stares at Pol and Bop
<iximeow>
!wpn the bzip'd minion
* Qboid
gives the bzip'd minion a magnetohydrodynamic beaver
<egg|zzz|egg>
whitequark: so remember that TODO about local error in the retrobop analysis
<egg|zzz|egg>
I should have looked at it earlier >_> #1741
<egg|zzz|egg>
(bad local error is tolerable with single-step methods, it's catastrophic with multistep methods since the integrator is stateful and becomes garbled)
e_14159 has quit [Ping timeout: 383 seconds]
e_14159- is now known as e_14159
<bofh>
egg|zzz|egg: oh so it *is* integrator state getting messed up?
<egg|zzz|egg>
bofh: yeah, it's fairly clear from local error plots
<egg|zzz|egg>
bofh: I finally got around to forking another (single-step) integration every 24 h from the main integration and comparing it over 24 h
<egg|zzz|egg>
same integrator and time step, but here there is no state corruption
<egg|zzz|egg>
the local error remains constant and rather dull
<bofh>
well yes, our solar system is generally better-behaved than that of the Kerbals', since any massive instability would've led to either collisions, getting Roche'd, or an escape trajectory billions of millenia ago. :P
<bofh>
Also I'm impressed RSS does Ouranian satellites correctly. I was trying to get an ephemeris for those working a bit ago and that was a surprising pain in the arse.
<bofh>
topmost blue line is Phobos, I presume?
<egg|zzz|egg>
ah the moons of Herschel are fun, yes
<egg|zzz|egg>
iirc the timestep is limited by the accuracy on Miranda (in global error at least)
<egg|zzz|egg>
(this plot is local error so might show different things)
<egg|zzz|egg>
not sure, could be Ariel?
<bofh>
Miranda is a fucking pain, tho I'm not sure how much of that is actual chaotic dynamics and how much of that is our mean anomaly / semi-major axis data for Miranda is crap b/c we only really have data from Voyager 2 for it, which is for a particularly narrow time slice.
<egg|zzz|egg>
bofh: well, that's not the issue here, I'm not comparing with observations so model error matters not
<egg|zzz|egg>
it's just numerical annoyance that is plotted
<bofh>
Well, yes, and the numerical error there seems pretty consistent b/c the orbit is fairly nice.
<bofh>
Especially when compared to Pol & retroBop :P
<egg|zzz|egg>
yeah, local error is constant as it should
<egg|zzz|egg>
bofh: the thing is, with a single-step method a badly-behaved retrobop will have an occasional blip of high local error and go back to sanity, whereas with a multistep method the state will keep that inconsistency and have consistently high local error, which is shit
<egg|zzz|egg>
(and leads to entertaining orbit plots)
<egg|zzz|egg>
bofh: seems the top few are Ariel, Enceladus, Dione, and Miranda
<bofh>
Enceladus is weird since that one's orbit is nice and very simple iirc.
<bofh>
is it possible to detect high local error in a multi-step method in a manner that is faster than comparing against a higher-precision single-step method at every step? so you could "detect" error spikes and temporarily switch to a single-step method or reset integrator state or w/e.
tawny has joined #kspacademia
tawny- has joined #kspacademia
TonyC1 has joined #kspacademia
B787_300 has quit [Ping timeout: 186 seconds]
B787_300 has joined #kspacademia
tawny has quit [Ping timeout: 186 seconds]
TonyC has quit [Ping timeout: 186 seconds]
icefire has quit [Quit: gone]
<Ellied>
bofh: ....I have no memory of the cosmic toad thing. When did I say that?
<Ellied>
oh, I found it in my logs
<bofh>
Ellied: no idea, check topicquotes
<Ellied>
wow, I really don't remember saying "cosmic toad" but there it is
<Ellied>
last July apparently
<Ellied>
context is I was testing the linearity of the oscillating light sensor I built and it got really weird as it started to saturate
<Ellied>
it didn't just start to droop, it kinda formed a jagged path through several different relatively-linear regions, each with a different slope, with sharp knees between them. It looked random but it was very consistent upon going back over it multiple times.
<Ellied>
lol I'd forgotten some of these quotes from myself
<Ellied>
<@Ellied> oh good it's just general relativity, I thought I was about to see a paper about timekeeping standards.
<bofh>
ROFL
<bofh>
now I want to read a paper about timekeeping standards.
<iximeow>
it's a very short paper
<iximeow>
"use UTC"
<iximeow>
actually, kind of wonder how many books have been written on the history of time keeping
<UmbralRaptor>
No, no. Use TAI and adjust Earth's spin to fix the errors.
Profound has joined #kspacademia
<bofh>
Nah, just "use TAI" full stop.
<bofh>
but, like, I want to read a paper about how one could, say, implement a Strontium ion-trap atomic reference at home using spare parts you have lying around.
<bofh>
like if you have Rb-85 handy making a Rubidium lamp is just a glassblowing exercise, & then all ypu really need is a stable VCO, a heater & a flyback upconverter to get a 6834.682611MHz freq reference.
<bofh>
Rb-87*
<rqou>
bofh: re your RT: if i were to ever operate a game of chance, i think i would make "large bets associated with Cambridge MA" an immediate red flag :P
<UmbralRaptor>
That sounds reasonable.
<egg|zzz|egg>
bofh: no, use TT(BIPM), not TAI
<egg|zzz|egg>
!wpn whitequark
* Qboid
gives whitequark a sub-calibre quaternion
* iximeow
wpn computer
<iximeow>
er
<iximeow>
!wpn computer
* Qboid
gives computer a treasonous trie
<iximeow>
ah, that was fun. tracking down a bug, i realize the code works correctly, but makes no sense.
<iximeow>
consult the paper i was referencing, to then find that the code that makes no sense doesn't appear in it at all and i appear to have just Made It Up
Technicalfool has quit [Read error: Connection reset by peer]
APlayer has joined #kspacademia
* soundnfury
likes the idea of a treasonous trie
icefire has joined #kspacademia
APlayer has quit [Read error: Connection reset by peer]
* Qboid
gives UmbralRaptor a precessing alveolar equivalence
<UmbralRaptor>
!wpn egg
* Qboid
gives egg an insecure geodesic
<egg|zzz|egg>
!wpn retrobop
* Qboid
gives retrobop a barreled theatre
<egg|zzz|egg>
!wpn -add:wpn retrobop
<Qboid>
egg|zzz|egg: Weapon added!
<egg|zzz|egg>
bofh: so today the argv encoding issue pops up again
<egg|zzz|egg>
bofh: I have flags that I parse as quantities
<egg|zzz|egg>
bofh: very amusingly, if I pass μs on the command line (typing a mu), it gets sent to main (in CP1252) as B5 73, so there's a fancy fuzzy encoding thing going on here
<egg|zzz|egg>
(but obviousy principia then assume it's UTF-8 and it's downhill from there)
<UmbralRaptor>
!u u+b573
<Qboid>
U+B573 HANGUL SYLLABLE DDYALB (땳)
<egg|zzz|egg>
UmbralRaptor: okay that would be differently entertaining
<egg|zzz|egg>
although
<egg|zzz|egg>
!u U+73B5
<Qboid>
U+73B5 CJK IDEOGRAPH-73B5 (玵)
<bofh>
egg|zzz|egg: what the hell
<egg|zzz|egg>
bofh: I mean, that seems perfectly reasonable tbh
<egg|zzz|egg>
Unicode -> CP1252 will be lossy no matter what, try to preserve as much as you can
<egg|zzz|egg>
bofh: haha long pointer
<bofh>
I assume that's some stupid version of FAR pointer?
<egg|zzz|egg>
bofh: ?
<egg|zzz|egg>
"Microsoft Visual Studio: Insufficient memory to continue the execution of the program"
<egg|zzz|egg>
Ꙩ_ꙩ
<egg|zzz|egg>
bofh: https://msdn.microsoft.com/en-us/library/windows/desktop/ff381404(v=vs.85).aspx "Historically, P stands for "pointer" and LP stands for "long pointer". Long pointers (also called far pointers) are a holdover from 16-bit Windows, when they were needed to address memory ranges outside the current segment. The LP prefix was preserved to make it easier to port 16-bit code to 32-bit Windows."
<UmbralRaptor>
"FAR pointer" sounds like a combination of memory allocation and partial differential equation insanity.
<egg|zzz|egg>
bofh: yeah, same as UmbralRaptor, I associate FAR (with that capitalization) with ferram4's mod
<egg|zzz|egg>
bofh: but "long pointer" seems far more memeable
<egg|zzz|egg>
wtf, there are no results for "haha long pointer" Ꙩ_ꙩ
<egg|zzz|egg>
oh well, there will be the logs of kspacademia soon
<bofh>
rofl
<bofh>
egg|zzz|egg: I associate FAR (with that capitalization) with reading bad old MSDOS code ported poorly to modern systems via #define FAR
<bofh>
I mean it *is* insane
icefire has quit [Read error: Connection reset by peer]
<kmath>
<✔mbeisen> it is very easy for me to mask my identity when reviewing anonymously - I simply don’t swear
<bofh>
UmbralRaptop: so when I was doing my master's I actually got to do review for my supervisor a few times, removing the swearing before sending him my comments was like a full fifth of the time I spent :p
<UmbralRaptop>
Hah
tawny- has joined #kspacademia
<bofh>
UmbralRaptop: I swear he gave me the papers he explicitly thought were trash and didn't want to review himself
<bofh>
which I don't mind, it was good review practise and it meant I wouldn't spent days agonizing over if me recommending reject was too harsh.
<egg|zzz|egg>
bofh: clearly you need to find a physics conference in Vienna urgently
<bofh>
I mean 5 December is awhile away :P
<egg|zzz|egg>
!wpn UmbralRaptop
* Qboid
gives UmbralRaptop a triangulated 🔪
<egg|zzz|egg>
!wpn bofh
* Qboid
gives bofh a greebled pun
UmbralRaptop is now known as ElectrodynamicRaptor
<egg|zzz|egg>
bofh: I wonder whether I can estimate the ляпунов exponent by that integration-forking trick instead of going down the path of the variational equation (forking with perturbed initial conditions, and using an integrator that I know isn't in a demented state)
<bofh>
egg|zzz|egg: I mean that's sort of what I was asking about yesterday, if that integration-forking trick is sufficiently fast as to be worth bothering with.
<bofh>
also goddamnit reading the wikipedia page for sachertorte is giving me chocolate cravings *again* >_>
<egg|zzz|egg>
bofh: entirely unrelated question, is black lagoon good, besides being a good way to nerd-snipe people satellite identification questions
* egg|zzz|egg
has somehow never actually watched any anime and wants to try
<egg|zzz|egg>
cc whitequark
<bofh>
egg|zzz|egg: watch Monogatari
<bofh>
actually no, my recommendation for a starting point for anime is the same as always: Ghost in the Shell: Stand Alone Complex.
<bofh>
like, for serious. it's the best series I've ever seen *by far* and incredibly fascinating.
<bofh>
(as for black lagoon, dunno, never seen it. whitequark'd be able to answer that question better).
<ElectrodynamicRaptor>
Black Lagoon is silly.
tawny- is now known as tawny
<tawny>
black lagoon's pretty good, it's got a lot of stuff you might not like to it tho
<bofh>
ElectrodynamicRaptor: how's The Jacksoning?
<ElectrodynamicRaptor>
I'm with 4 other grad students, and we're all struggling.
<bofh>
Ouch. At least that's fairly typical...
<ElectrodynamicRaptor>
Also, we seem to have problems that work from the "examples hinder understanding" school of thought >_<
<ElectrodynamicRaptor>
egg|zzz|egg: watch "My Neighbor Totoro". Because ネコバス.
<egg|zzz|egg>
ElectrodynamicRaptor: can't find it easily tho
<ElectrodynamicRaptor>
Odd.
<bofh>
egg|zzz|egg: rutracker.org
<egg|zzz|egg>
!wpn ElectrodynamicRaptor
* Qboid
gives ElectrodynamicRaptor a penguin
* ElectrodynamicRaptor
takes the laplacian of the penguin.
<bofh>
!wpn egg|zzz|egg
* Qboid
gives egg|zzz|egg a dynamic conflict
<ElectrodynamicRaptor>
git conflict -force
<egg|zzz|egg>
!seen whitequark
<Qboid>
egg|zzz|egg: I last saw whitequark on [04.03.2018 01:27:34] in #kspacademia saying: "!wpn egg"
<kmath>
<RStarDinoPirate> I am angry that my careers officer never told me this was an option and now I feel like I've wasted my life https://t.co/M7JkUp5Pdk
<kmath>
<stephentyrone> "The reals are defined as the set of all rational and irrational numbers on the number line." ⏎ ⏎ Thanks, programmer d… https://t.co/64s3qjLWJl
<egg|zzz|egg>
!wpn -add:wpn crab
<Qboid>
egg|zzz|egg: Weapon added!
<UmbralRaptop>
!wpn egg|zzz|egg
* Qboid
gives egg|zzz|egg a rhodium lithium glaive beam
<SilverFox>
do we have any GPU enthusiasts here, in terms of architecture-loving or getting into the hardware/software performance of designs?
<SilverFox>
also, more coding centered, lets say you have a 2D list/array thing of things, with the first dimension potentially being a string or an int, but not both, depending on how we do things. My question is, is it better to have the first dimension be an int that we can use with other lists in the future to help index things, or should we keep it a string which is more direct and would take less lists in general? Which is better coding practice?
<UmbralRaptop>
Question 0:how big is your array
<SilverFox>
uhhhhhhhhhhh
<SilverFox>
how big is big?
<UmbralRaptop>
Entries, bytes, whatever
<SilverFox>
right now it's 100+ items, but it could be very much larger, but I'm also talking about things in general, rather than a specific example
<SilverFox>
also the string entries in the first dimension would not be large either, like basically the size of a person's name
<UmbralRaptop>
It sounds like you could comfortably load the table in TOPCAT, Excel, etc?
<SilverFox>
probably
<SilverFox>
it's not some huge multi-thousand entry several GB array
<UmbralRaptop>
If performance isn't an issue, I'd go with data types that fit what is being stored.
<SilverFox>
but im just focusing on the concepts in coding, rather than basing things off of specific examples I have in my projects
<egg|zzz|egg>
!wpn UmbralRaptop
* Qboid
gives UmbralRaptop a down revolving compass
<egg|zzz|egg>
!wpn bofh
* Qboid
gives bofh a photothyristor
<SilverFox>
thyristor? is that the potentiometer/rheostat thing?
<egg|zzz|egg>
bofh: A survey of shadowing methods for numerical solutions of ordinary differential equations seems interesting
<bofh>
SilverFox: no, a thyristor is just an SCR (PNPN junction). I think you're thinking of a thyratron?
<bofh>
egg|zzz|egg: ooh, neat. :3
<egg|zzz|egg>
bofh: so unfortunately looking for "backward error" gives you lots of defect-based backward errors, not so much shadowing (by that terminology), which isn't great if you're doing gravitation
<SilverFox>
I've had such a history with only viewing journals and papers related to medicine that seeing one about math and numbers creates such a disconnect for me
<egg|zzz|egg>
bofh: but looking for "backward error" "lyapunov" brought me to this, which is neat
* egg|zzz|egg
likes survey papers
* egg|zzz|egg
pets a survey paper
<SilverFox>
what is a survey paper?
<SilverFox>
surely it's not an actual survey
<bofh>
Survey papers are great, saves you the time of actually scouring the literature yourself for a good overview of the field.
<UmbralRaptop>
egg|zzz|egg: Ah, but is a paper solely about eg: LSST a survey paper?
<egg|zzz|egg>
bofh: I mean, not like I don't do the former too
<bofh>
Same, but it's nice when I don't *have* to.
<egg|zzz|egg>
see my background pic on twitter
<UmbralRaptop>
SilverFox: A paper that surveys the literature/state of knowledge of a given field.
<SilverFox>
so you just survey a bunch of professionals?
<egg|zzz|egg>
bofh: but survey papers also make that easier, by giving pointers
<bofh>
Indeed!
<egg|zzz|egg>
thurvey papers.
<SilverFox>
okay that's better than what I anticipated