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.
awang has quit [Ping timeout: 383 seconds]
UmbralRaptop has joined #kspacademia
<whitequark>
hm, awang quit
<whitequark>
I might actually have been solving the same problem as them
UmbralRaptor has quit [Ping timeout: 198 seconds]
awang has joined #kspacademia
UmbralRaptop has quit [Quit: Bye]
UmbralRaptop has joined #kspacademia
* UmbralRaptop
🔪 people
<UmbralRaptop>
Why would you send me an email at 4 am about something I should register for and attend at 11 am?
<awang>
whitequark: I was probably on my way home
<awang>
What problem are you looking at?
<whitequark>
I have pointer loads that I know always result in valid pointers
<whitequark>
but LLVM doesn't always hoist these out of loops
<bofh>
UmbralRaptop: so that sounds like something I might even have done before in the past (in my defense, the deadline for registration was 11AM the next day and it was closer to 4:30AM).
<UmbralRaptop>
bofh: On the upside, it looks like a thing that recurrs about once a week?
<kmath>
<UCBProf> Enough people are asking that I thought I'd explain in one thread my take on the recent EDGES paper on a possible d… https://t.co/HS00jMYGp6
<kmath>
<aerincreer> Boys please learn how to take selfies so I stop embarrassing myself when I show my friends your pictures. Now I got… https://t.co/j5gyLnmqbd
<kmath>
<xabean> Uh. Guys? Intel is trying to put a snake in your UEFI. ⏎ ⏎ "Session 5 Intel : Implementing Micro Python in UEFI"… https://t.co/9YbpKAqFIT
<bofh>
UmbralRaptop: I usually manage "stereotypical mad scientist" somehow, probably due to hilariously unkempt hair.
<APlayer>
Wait, is that a stereotype?
<UmbralRaptop>
APlayer: watch Back to the Future for a good example
<egg|zzz|egg>
bofh: is there a distinct mad mathematician look? or mad numericist?
<egg|zzz|egg>
binary128 everywhere! *thunder*
<APlayer>
It's fun, because there are perhaps three students at our school in our year that could qualify as mad scientists, and all three have unkempt hair
<awang>
@wpn -add:adj curiously recurring
icefire has joined #kspacademia
<awang>
!wpn -add:adj curiously recurring
<Qboid>
awang: Adjective added!
<egg|zzz|egg>
!wpn bofh
* Qboid
gives bofh a bovine quadruple Brand New Photoneutronic Engine
<egg|zzz|egg>
!wpn UmbralRaptop
* Qboid
gives UmbralRaptop a yttrium endomorphism
<UmbralRaptop>
!wpn egg|zzz|egg
* Qboid
gives egg|zzz|egg an optically thick acid-sensing snakehead
<bofh>
just what I always wanted :P
<SnoopJeDi>
acid-sensing sounds like a really good D&D power
<SnoopJeDi>
pull up in the cart like "get in loser we're going green dragon hunting"
<UmbralRaptop>
:D
<SnoopJeDi>
tangentially, my DM is way too generous with us, we've gotten away with THREE dragon kills in our campaign
<UmbralRaptop>
Also, alerts you to slimes and gelatinous cubes?
<SnoopJeDi>
ooh yea good point
<awang>
So
<awang>
I can either go with dimensionless units that don't really provide much in the way of type safety
<awang>
Or I can measure pixel distances in terms of moles
<awang>
What should I pick?
<awang>
Or C) Write my own units library
<awang>
I'm pretty sure I'm not supposed to be measuring numbers of pixels in moles
<awang>
But pixels aren't really a length, since they can't be converted to/from other lengths without a scale factor
<SnoopJeDi>
right, they're dimensionless by design
<SnoopJeDi>
Is type safety a huge-huge priority?
<awang>
1200 x 900 moles of pixels doesn't make sense
<awang>
No, not a huge-huge priority
<SnoopJeDi>
I would probably just stick to (A) then
<awang>
I'm more curious as to how this could be accomplished
<awang>
Hmmm
<awang>
Alright, let's see what happens
<SnoopJeDi>
um, assuming lexicographic ordering for the un-labeled approahces
<SnoopJeDi>
the approaches, too
egg|phone|egg has joined #kspacademia
<egg|phone|egg>
Meow?
<SnoopJeDi>
so I got a "please forward to any interested colleagues" email from my collaborator about a computational accelerator job with them...not sure if being directly solicited or not 🤔
<kmath>
<NuclearAnthro> 😻 проект плуто LIVES!😻 ⏎ ⏎ I mean, this is awful that we’re going this route but Project Pluto is a barrel of apocalyp… https://t.co/1mjT7CoyyL
<SnoopJeDi>
I'm shocked the submarinenuke is real
<SnoopJeDi>
well, surprised. shocked is too strong a word for post 2016-2017
<UmbralRaptop>
This is a weapon that Charles Stross used in multiple stories because it's got way more capability than you need/really can use in the real world.
<UmbralRaptop>
Project Pluto was predicted to have what >100,000 km range?
<SnoopJeDi>
John Walker gives it as "global range and the ability to stay airborne for months"
<kmath>
<stephentyrone> @volatile_void `powr`, `pown` and `rootn` are what IEEE 754 actually wants. Including `pow` was a case of someone t… https://t.co/W6PLzAhtxh
<bofh>
egg|zzz|egg: I still don't have a good answer for that, honestly. I think the behaviour of sqrt is just odd.
<egg|zzz|egg>
you could try asking the cat again
<bofh>
!kd -add Atlas 5 • An Atlas V-541 rocket will launch the GOES-S next-generation geostationary weather satellite for NOAA/NASA.|2018-03-01T22:02Z
<Qboid>
bofh: Invalid syntax. The correct syntax would be name|description|time.
<bofh>
!kd -add Atlas 5 • GOES-S|An Atlas V-541 rocket will launch the GOES-S next-generation geostationary weather satellite for NOAA/NASA.|2018-03-01T22:02Z
<e_14159>
"As a note on the scale of the setup, the computational budget to reproduce those experiments is approximately 60K P100 GPU hours"
<SnoopJeDi>
I turned over intentionally-wasteful proof-of-work re: cryptocurrency with my gf last night...I still can't quite get my head right over it
<SnoopJeDi>
e_14159, "Our experimental results suggest that future GAN research should be based on more systematic and objective evaluation procedures" LOL GOOD LUCK
<egg|zzz|egg>
bofh: so what are the restrictions on aliasing in FORTRAN? is it just that arguments can't alias, or are there other things that can't alias
<e_14159>
SnoopJeDi: I especially like the uncertainties in Figure 4 and 5.
<e_14159>
Also, s/GAN/machine learning
<SnoopJeDi>
agreed
<SnoopJeDi>
although you could reasonably s/GAN// and the joke still works
<SnoopJeDi>
Figure 4 makes me think the authors would do well in astronomy
<kmath>
<catherineols> An interesting punchline: claimed improvements in previous work were due to implementation mistakes. So, the improv… https://t.co/u5a5I86SjP
<e_14159>
Reminds me of a master's thesis I supervised on GAIL. If reinforcement learning is difficult to train and GANs are difficult to train, then generative adversarial imitation learning has to be hell.
<Qboid>
0d 0h 10m 0s left to event #25: Atlas 5 • GOES-S [at 2018-03-01 22:02:00]. Say '!kountdown 25' for details
<SnoopJeDi>
bofh, lmao
<e_14159>
bofh: Oh, I hadn't seen that paper yet. Bookmarked!
<awang>
egg|zzz|egg: RP2Point?
<awang>
I may or may not have caught template feaver
<awang>
Too many templates :(
<awang>
Too generic
<awang>
Pretty sure I'm doing this horribly wrong, too
<Qboid>
0d 0h 2m 30s left to event #25: Atlas 5 • GOES-S [at 2018-03-01 22:02:00]. Say '!kountdown 25' for details
<Qboid>
0d 0h 0m 0s left to event #25: Atlas 5 • GOES-S [at 2018-03-01 22:02:00]. Say '!kountdown 25' for details
<egg|zzz|egg>
awang: A point on the projective plane
egg|phone|egg has quit [Ping timeout: 383 seconds]
<awang>
egg|zzz|egg: I see
<awang>
Also, what's your stance on traits?
<awang>
Good/bad/maybe both?
<egg|zzz|egg>
what do you mean
<awang>
Uh
<awang>
So I have this class
<awang>
Which may be backed by some other geometry library
<awang>
And I'm trying to figure out how to get the rest of the code to not care what library is used
<awang>
So I have free functions for what would normally be member functions and dispatch to the traits class for that type, which has the appropriate implementation
<awang>
Wait
<awang>
Is this CRTP?
<awang>
Dangit
<awang>
Is it normal to have bits and pieces of implementation details as template parameters?
<awang>
It feels dirty
<bofh>
Currently have one window open to the Centaur stage launch and another open to DSN Now!
<kmath>
<✔jamesaydelott> You'll love this. ⏎ GOES-East weather satellite caught the rocket exhaust plume from the launch of its sister weath… https://t.co/SuR2ndVni0
<bofh>
:D *
<SnoopJeDi>
💙💙💙
<SnoopJeDi>
oh man, Tabetha Boyajian (of Tabby's star fame) is giving a seminar here on Monday!
<bofh>
Nice!
<SnoopJeDi>
oh my god, I just realized why I am having so much trouble building this software on a fresh VM, when it worked on the old one
<SnoopJeDi>
their binary release includes copies of like *EVERYTHING* in the dependency tree, all the way down to the particular MPI compilers used to build it blah blah
<SnoopJeDi>
and I think my old rc was including some of that in its path, making the hyperspecific versions (or whatever strange magick I'm missing atm) available for a fresh buidl
<SnoopJeDi>
angery.jpg
<bofh>
scientific_computer_programming.jpg
<bofh>
I can't even pretend to be surprised. :/
<egg|zzz|egg>
meow
<SnoopJeDi>
It's not a *completely* insane way to distribute for HPC I guess, but I'm just so salty to have discovered the "huh I can't build from source anymore despite meeting all the deps they list" and that it's such an inane reason
<bofh>
Heh.
<bofh>
I'm not sure if I agree that it's not completely insane. *Common*? Sure, I've done it before even. But I still think it's completely insane.
<egg|zzz|egg>
!wpn bofh
* Qboid
gives bofh an Apollonian Qboid
<bofh>
Damnit, I wanted a Dionysian Qboid. :P
<SnoopJeDi>
I'm trying to be generous to science bofh, it's a character flaw :P
<bofh>
touché
<bofh>
:P
<SnoopJeDi>
I don't, for example, think it's bad per se to have /share/apps/very_specific_package and source an env from there
<bofh>
Okay, I'll agree there.
<SnoopJeDi>
that said, you should um *probably* try to limit the specificity of the env you build for that deployment
<SnoopJeDi>
shipping ALL OF OPENMPI and sourcing them into the user's $PATH, probably wrecking their compilers and libraries, is maybe not quite the best
<SnoopJeDi>
(although I'm pretty sure what broke was libhdf5...somehow? that's where my build from src seems to fail)
<whitequark>
!wpn egg|zzz|egg
* Qboid
gives egg|zzz|egg a Giger operator/list hybrid
<SnoopJeDi>
doesn't super help that there are 57 varieties of libhdf5
<SnoopJeDi>
(and the devs don't specify which they depend on)
<egg|zzz|egg>
!wpn whitequark
* Qboid
gives whitequark a hafnium reflector
<egg|zzz|egg>
whitequark: any котяnews?
* bofh
waits (im)patiently for AOS of GOES-17
<egg|zzz|egg>
Извескотя
<whitequark>
!wpn egg|zzz|egg
* Qboid
gives egg|zzz|egg a constexpr graph
ferram4_ has joined #kspacademia
ferram4 has quit [Read error: -0x1: UNKNOWN ERROR CODE (0001)]
<bofh>
AOS of G-17 S-band downlink on DSS34/DSS36 just now, looks like things are going according to plan. :)
<bofh>
I swear I treat orbital insertion as a spectator sport or somesuch.
<SnoopJeDi>
Watching you geek out on the links is like a meta-sport for me