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.
hattivat has quit [Quit: WeeChat 1.8]
e_14159 has quit [Ping timeout: 201 seconds]
e_14159 has joined #kspacademia
icefire has quit [Read error: Connection reset by peer]
Majiir is now known as Snoozee
ferram4 has quit [Ping timeout: 204 seconds]
egg|cell|egg has joined #kspacademia
egg|phone|egg has quit [Ping timeout: 383 seconds]
Thomas|AWAY is now known as Thomas
<Iskierka>
egg: write a book in glass such that it must be held up to a light source for the text to appear in focussed shadows
<egg|work|egg>
:D
<Iskierka>
or better, chapter 1 does that and describes schlieren, then the rest is written in minute density differences that require a schlieren setup
<SnoopJeDi>
I think it's the completely arbitrary facet layout
<UmbralRaptor>
What are the variables on each axis, even?
<hattivat>
small resolution combined with the vertical stripes caues eye strain
<hattivat>
s/caues/causes
<Qboid>
hattivat meant to say: small resolution combined with the vertical stripes causes eye strain
<UmbralRaptor>
They list categories, but the axes…?
<SnoopJeDi>
I don't mind so much the smallness or the stripes, but looking at it along the eye-strain axis, I now see that the labels are...really haphazardly placed.
<hattivat>
I think they are 'global'
<SnoopJeDi>
UmbralRaptor, yea facet plots *usually* list the shared axis along the outer edges
<hattivat>
ie. one label for all plots
<SnoopJeDi>
I think it might just be too much in one figure
<hattivat>
I think it would be easier to read if different plots used different colours
<SnoopJeDi>
i.e. ethnicity/income could be their own 2x2 facets
<SnoopJeDi>
it definitely does an adequate job getting the main point across, but I've had plot quality on the brain
<SnoopJeDi>
Tried to drum up interest in a Tufte reading group, only one person said they'd particiapte :(
<hattivat>
and the stripes make no sense, seem to just be an artifact of integer values
<hattivat>
what I mean is that the visual difference between striped and non-striped plots takes attention away from actually meaningful differences
<SnoopJeDi>
yea that's a binning thing with the Quantcast data I believe
<SnoopJeDi>
I'm not entirely sure if the contrast there is a good or bad thing: it does highlight that the data are spotty in places.
<hattivat>
so true values for college-educated people all happen to be integers, while for grad school they are floats?
<hattivat>
makes no sense to me
<SnoopJeDi>
No they're striated in every case, it's just a lot less obvious for more dense sets
<SnoopJeDi>
or that's what I can pick out: it *is* hard at that resolution
<SnoopJeDi>
that's not a plotting issue though, that's the data
<hattivat>
true
<SnoopJeDi>
oh, I think might be a filter they apply rather than something from their training data
<SnoopJeDi>
I obviously haven't read it in great detail ?
<hattivat>
I pointed this particular pair because the difference is so glaring (college is so perfectly striped while grad school is so dense that that difference seemingly has far more statistical significance than any other difference between these sets
<hattivat>
)
<SnoopJeDi>
I got mildly nerd-sniped by a demographics question and went looking to see how good unsupervised classifications can pick out user ethnicity on Twitter (apparently pretty well?)
<hattivat>
and I see no obvious explanation other than some error in data collection/categorization, since I find it highly unlikely that a dataset would have significantly fewer college graduates than *both* grad school graduates and non-graudates
<SnoopJeDi>
But they represent each user by a subset of their friends, which is then scrubbed to exclude things like private profiles etc., so maybe what that's showing is that college students are less promiscuous on Twitter
<SnoopJeDi>
there's a definitely trend in the age row
<SnoopJeDi>
yes, a definitely trend...
<hattivat>
yeah, it seems to have accuracy overall
<hattivat>
with 35-44 year-olds somehow being the most stealthy
<SnoopJeDi>
I'm sure it's somewhere in the body (perhaps implicitly...), definitely a flaw not to point it out in the caption
<SnoopJeDi>
And would be a lot more identifiable if that was broken in several figures, I think
<SnoopJeDi>
hattivat, ugh I've just realized the vertical scales are all different
<hattivat>
horizontal ones as well
<SnoopJeDi>
AAAAAAAAAAAΑΑΑAAAΑΑΑΑΑAAAA
<hattivat>
and some of them fail to label the starting point
<hattivat>
a pretty fatal mistakes as far as plots are concerned
<egg>
hattivat: ping?
<hattivat>
s/mistakes/mistake
<Qboid>
hattivat meant to say: a pretty fatal mistake as far as plots are concerned
<egg>
(talking about floating point pings me)
<SnoopJeDi>
egg, and root beer floats
<hattivat>
you have alerts set up for cs terms? ;D
<hattivat>
also, I wanted to say that I'm in awe of your power of tri-location
<SnoopJeDi>
!wpn -add:adj triangulated
<Qboid>
SnoopJeDi: Adjective added!
<hattivat>
!wpn -add:wpn teledildonics
<Qboid>
hattivat: Weapon added!
<hattivat>
yes, that's an actual word, don't google it at work
<SnoopJeDi>
...huh. Word just came down that DOE has funded the Phase II we put in. This *should* be joyous news...
<egg>
frabjous day?
<SnoopJeDi>
kinda?
<SnoopJeDi>
My advisor and I had completely written off the possibility of the award and worked over the last 3 months to change my thesis topic to this cyclotron business
<SnoopJeDi>
so at this exact moment it looks like I'm doubly committed :|
<SnoopJeDi>
embarrassment of riches I guess
<egg>
!wpn SnoopJeDi
* Qboid
gives SnoopJeDi a toasted submersion with a statistic attachment
<SnoopJeDi>
And it's a fun excuse to grate 8 oz of Gruyère all at once
<hattivat>
yeah, stereotypically French level of health-friendliness
<hattivat>
still looks delicious though, I think I'll try it next week, thanks for sharing
<SnoopJeDi>
oh jeez don't get me started on the h-word
<hattivat>
as a mayonnaise fiend I do have to think about it, going for full indulgence is guaranteed to end badly with my tastes ;d
ferram4 has quit [Read error: Connection reset by peer]
ferram4 has joined #kspacademia
SnoopJeDi has quit [Ping timeout: 180 seconds]
SnoopJeDi has joined #kspacademia
* UmbralRaptor
blinks at the scrollback.
<SnoopJeDi>
oh so that did go through, um my bad
<UmbralRaptor>
Probably not.
<SnoopJeDi>
did I spew a bunch of base64?
<hattivat>
nope
<SnoopJeDi>
oh phew
<UmbralRaptor>
^
<UmbralRaptor>
More of I assumed that everyone knew what "teledildonics" meant, since I first ran into the term in Wired, uh, close to 20 years ago. O_o
SnoopJeDi has quit [Read error: Connection reset by peer]
SnoopJeDi has joined #kspacademia
<hattivat>
I didn't until a few years ago
<SnoopJeDi>
ffs, I did it again.
<UmbralRaptor>
SnoopJeDi: it just looks like you got peered.
<SnoopJeDi>
UmbralRaptor, 5'11" (or 180cm if you like)
<UmbralRaptor>
ah
<SnoopJeDi>
current BMI is ~ 23, %BF to be discovered sometime in the next month or so
<SnoopJeDi>
hattivat, it's been a lingering to-do to combine that with my food diary data and perhaps compare the two
<SnoopJeDi>
truth be told, that notebook was an excuse to learn Pandas :P
<UmbralRaptor>
hah
<hattivat>
that's a very decent excuse as far as excuses go
<SnoopJeDi>
I wonder about the periodic components since ~ Aug 2016
<Iskierka>
payday?
<hattivat>
holidays?
<SnoopJeDi>
hollandaise?
<hattivat>
(personally I'd certainly see a spike at Easter if I gathered such data)
<UmbralRaptor>
Iskierka: usually "starving grad student" isn't literal.
<SnoopJeDi>
oh I see, Iskierka. I have a sneaking suspicion it's just "ahhh but it's the weekend!" plus Friday happy hours catching up with me.
<Iskierka>
doesn't look to me like weekend periods, more like month, so I figured a reduction in intake as budget gets a bit on the tighter side
<SnoopJeDi>
alcohol *seems* to cause a droop (even if my total intake is above daily energy expenditure), which I assume is dehydration
<SnoopJeDi>
It's quite possible, I'm certainly looser with eating out when my checking account isn't screaming for mercy
<Iskierka>
maybe your body doesn't process alcohol much energetically? (since it is more complex than energy in/out, based on what the body actually chooses to take in and put out)
<SnoopJeDi>
Nah the caloric content of alcohol is very well-established, but I just assume the lost water content dominates in the short-term
<SnoopJeDi>
Lots of interesting questions to ask though
<hattivat>
caloric content and actual assimilation of nutrients are two separate things
<Iskierka>
Caloric content is established, but sometimes the body just won't take it in. For most people there are limits on how much will be accepted. (for most because the 700-kg people clearly must be exceptions)
<SnoopJeDi>
And suuuuper helpful to have something to point to when talking to a newcomer to the weight control community.
<SnoopJeDi>
Do either of you have a ref you can point me to for vetting what the ± on such a thing is?
<hattivat>
a good example is nuts and seeds, they have a high caloric content in theory, but it is quite established that human body is unable to take full advantage of it
<hattivat>
I'm not a nutritionist, I just learned about this from a distant friend who is
<SnoopJeDi>
I've only ever bothered to look at the ± on TDEE, which is not large enough to matter for our community's purposes
<hattivat>
IIRC establishing the "caloric content" you see on labels is a weird and rather inexact science
<hattivat>
I think it is done through literally burning the food and measuring the heat
<SnoopJeDi>
I'd be fully willing to believe it's bomb-calorimetric rather than based on bioavailability
<hattivat>
or at least that's one of the methods
<Iskierka>
inexact for the ones that want to claim realistically low values like nuts despite high actual. For ones with relatively high availability bomb-calorimeter is definitely the method
<SnoopJeDi>
Anecdotally, logging the sticker-label value has worked for me, to within the limits of logging precision anyway
<SnoopJeDi>
Surely someone out there has done a DLW study on alcohol
<Iskierka>
While energy uptake is harder to definitively observe (at least would take a while to google), that the body can vary its output quite considerably is fairly obvious from known starvation modes and how little people with anorexia can endure on
<SnoopJeDi>
Ehhhh
<Iskierka>
for small variations for long-term weightloss, it probably works, but the people who say "just cut out 500 calories for a few months" are inaccurate. And you won't lose weight as fast as the energetic prediction made there
<SnoopJeDi>
Starvation mode is generally an indicator that hands are being waved unless one is speaking specifically about the Minnesota experiment or holocaust victims
<SnoopJeDi>
IME
<Iskierka>
There is a definite starvation mode observed where greater fat formation and preservation happens
<SnoopJeDi>
I try hard not to get my head stuck in the metabolic adaptation literature: it's too unsettled to sort up from down
<SnoopJeDi>
(and off-topic to the part I care about at any rate)
<SnoopJeDi>
most of the stuff that biochemists/nutrition scientists love gets washed out by the relatively massive error associated with a food diary anyway, heh
<hattivat>
my limited understanding is that if your body becomes convinced that there is a famine going on, it burns muscle first
<hattivat>
and preserves fat for later
<SnoopJeDi>
It's Complicated™
<hattivat>
which makes evolutionary sense, I guess
<hattivat>
since the more muscle you have, the more energy you burn
<hattivat>
so getting rid of muscle first makes fat last for longer
<SnoopJeDi>
It takes a not-insubstantial time to get into glycogen depletion, so it's largely a non-issue for anyone living in a modern society
<SnoopJeDi>
Muscle catabolism is definitely inevitable on an energetic deficit, though
<hattivat>
that certainly seemed to be my experience, even with frequent exercise and a relatively high protein intake to stave this off
<SnoopJeDi>
It's a really tricky dance for me: as a scientist, I loooove the details. As a person trying to control their weight, it's crucial not to lose the sense of pragmatism
<SnoopJeDi>
and for the sort of new-comers I do most of this reading for, it's dire. The average person will do impressive mental gymnastics to avoid confronting the simplicity of "To lose weight, eat less. Adjust as you go."
<hattivat>
haha, yeah
<SnoopJeDi>
I should really fix up that bibliography for that reason, it's a mess
<hattivat>
actually, not even eat less, exercise more can work too, depending on current diet
<SnoopJeDi>
Emphasis on "can", heh. You're right, of course, but it's a thorny subject at best for newcomers
<Iskierka>
exercise is generally harder to achieve as it is required to be very intensive for a reasonably long time to actually amount to that much expenditure
<Iskierka>
barring food addiction eating less is much less demanding
<SnoopJeDi>
It depends exactly what you mean, but at a bare minimum, keeping a food diary is reaaaally easy nowadays
<hattivat>
perosnally I find it much easier to exercise more than to go around being hungry
<hattivat>
but that's personal preference, obviously
<hattivat>
that being hungry part is also what irks me about most diets, even the more-scientific kinds
<SnoopJeDi>
I pretty much eschew any formalized system that isn't "use a food diary, set goals, assess regularly"
<egg>
wow lots of backlog
<hattivat>
my eureka moment when I was struggling to lose weight was when I specifically targeted eating foods with a high feeling full to actual-calories ratio
<SnoopJeDi>
I did warn hattivat not to get me started egg (and then got myself started)
* egg
tends to skip lunches lately
<egg>
mostly because too lazy to go down to the cafeteria
<hattivat>
I recall lentils being particularly glorious on that front
<SnoopJeDi>
yea, legumes and pulses are awesome for satiety
<egg>
but espresso counts as food right
<SnoopJeDi>
(and eminently practical for a graduate student >_>)
<hattivat>
if you are French, I think it does
<egg>
<hattivat> yeah, stereotypically French level of health-friendliness <<< :D :D :D
* egg
gives hattivat a butter
hattivat is now known as hattivat|omnomnom
<egg>
:D
<SnoopJeDi>
I have a jar of ghee that has gone unloved for far too long
<SnoopJeDi>
Perhaps a nice chicken biryani is in order
<egg>
bofh: I'm getting sidetracked down an infinite sequence of error analysis rabbit holes
<Iskierka>
With eating less being less demanding, I mean more like that most low-intensity exercise that people can actually keep up for a decent time is like 300 cal/hour (aerobics, jogging, sensible cycling speed), and an hour out of your day doing those is a lot
<UmbralRaptor>
egg: caffeine is a vitamin, right?
<Iskierka>
and with high intensity stuff you then need to shower afterwards which is more time out of your day, so you're either not doing them every day or taking out a lot of time, assuming you can do high-intensity
<egg>
UmbralRaptor: also walking to the tram and back is a sport right
* Iskierka
has not personally yet decided which would be preferable, however. Just has the commute providing sensible-intensity exercise
<egg>
typing, too
<UmbralRaptor>
Egg: depends on how aggressively you're dodging drivers?
<egg>
uh
<egg>
this is switzerland
<egg>
when there isn't a light (which I think is the case on all the crossings between here and the office, amusingly), drivers stop
<SnoopJeDi>
Iskierka, you're singing my tune, heh. Granted, I do a lot more cardiovascular stuff these days because my margins are smaller, but we push pretty hard that "eat less" should be prioritized over "move more"
icefire has joined #kspacademia
<SnoopJeDi>
I'm thankful that there are resources like the National Weight Control Registry that turn an investigative eye on these sorts of questions, though.
<egg>
!wpn Iskierka
* Qboid
gives Iskierka a Лидов–古在 expression
<egg>
!wpn UmbralRaptor
* Qboid
gives UmbralRaptor a Lyman raptor-like torus
<egg>
!wpn hattivat|omnomnom
* Qboid
gives hattivat|omnomnom an Agile decomposition
* UmbralRaptor
shines a 1216 Å light at egg.
<egg>
UmbralRaptor: I'm wondering how to compute the time since apoapsis from cartesian coordinates with good error bounds
* UmbralRaptor
may be updating a mac. Then hopefully updating the observation simulation codes.
<UmbralRaptor>
hrm
<hattivat|omnomnom>
Iskierka: it's probably not an option for people living in big cities, or sparsely populated ones like they have in the US, but for me burning 300 kcal more a day was trivial - I just walked to work instead of taking a tram/driving
<egg>
UmbralRaptor: can't find any error analysis of this stuff, and it's full of horribly-conditioned madness
<hattivat|omnomnom>
half an hour each way, the same journey would be 15 minutes by car or tram once we account for the inevitable time spent wiring for the tram / searching for a parking spot
<egg>
UmbralRaptor: As this is a tart rather than a cheesecake, you should be disturbed.
<hattivat|omnomnom>
so just half an hour of additional time expenditure, not much more than I'd spend keeping a food journal
* UmbralRaptor
assumes madness involved with finding true anomaly.
hattivat|omnomnom is now known as hattivat
<egg>
UmbralRaptor: well if one starts from the cartesian coords the true anomaly should be trivial
* UmbralRaptor
suspects that hattivat is grossly underestimating parking space availability.
<UmbralRaptor>
egg: doh
<hattivat>
at my previous workplace? hotly contested
<egg>
with error as good as it gets, it's just computing an angle from cartesian coords (well if the orbit is circular it's hell but anyway)
<hattivat>
since most employees seemed to prefer the suburban lifestyle
<Iskierka>
hattivat, that 300 cal is certainly easy enough. Then how would you spend another 300?
<hattivat>
I wouldn't, I would just be patient
<hattivat>
healthier that way too, if my limited understanding of the topic is correct
<Iskierka>
but if you were still gaining weight so you couldn't?
<hattivat>
well then, obviously my diet would be the problem
<Iskierka>
In which case it's then eating less, as more exercise is impractical
<hattivat>
I did mention the caveat above that it depends on one's diet
<hattivat>
sure
<hattivat>
my personal method was to make sure that I don't gain weight with a completely sedentary lifestyle first, and then add exercise on top of this
<SnoopJeDi>
Yep, that's the attitude my community aligns with
<Iskierka>
though it does make an interesting question of how much I burn cycling. Like 15 minutes through the city, but at a decent pace, but with stop/start, but the city centre is on a small hill that I cross ... Easiest bit would probably be the ride home after the train. All uphill so no energetic advantage to cycling (disadvantage due to bike weight), it's a half hour walk on a gentle but noticeable uphill
<Iskierka>
(15 minutes each way through the city that is)
<hattivat>
but since I never had the luxury of not caring what I eat (strongly allergic as a child, first-gen drugs tended to have increased hunger as a side-effect), that first step was fairly easy for me
<Iskierka>
!wa 60m * g * 120 kg
<Qboid>
Iskierka: 60 g m (gram meters)×120 kg (kilograms): 7.2 kg^2m (kilogram squared meters)
<Iskierka>
!wa 60m * 9.81 m/s^2 * 120 kg
<Qboid>
Iskierka: 60 meters×9.81 m/s^2 (meters per second squared)×120 kg (kilograms): 70630 kg m^2/s^2 (kilogram meters squared per second squared)
<Iskierka>
!wa 60m * 9.81 m/s^2 * 120 kg in calories
<Qboid>
Iskierka: convert 60 meters×9.81 m/s^2 (meters per second squared)×120 kg (kilograms) to thermochemical calories: 16880 cal_th (thermochemical calories)
<Iskierka>
so the uphill bit isn't actually adding that much beyond preventing free-rolling, at least if google earth terrain geometry is accurate
<UmbralRaptor>
egg: also madness in going from angles to time to, traverse the angles?
<egg>
UmbralRaptor: yeah that's my question: how do you compute time with low errors
* UmbralRaptor
doesn't know
<egg>
UmbralRaptor: I think if you do it from the eccentricity and the true anomaly it's intrinsically shit, can it be less shitty (better-conditioned) if you compute it from other things
<Iskierka>
which it definitely isn't that accurate as there's definitely flat places that aren't. Should probably use an OS map's contours
<egg>
and anyway computing the mean anomaly from the true is numerically tricky even if you ignore the fact that the resulting function is ill-conditioned at high eccentricities...
<egg>
!seen bofh
<Qboid>
egg: I last saw bofh on [23.05.2017 20:13:50] in #kspacademia saying: "UmbralRaptor: that is amazing, going to read that posthaste."
<kmath>
YouTube - Green Army Fraction - Agrarian Übermensch
<UmbralRaptor>
We like our jokes to be scientifically accurate. ^_^
<bofh>
egg: ?
<egg|phone|egg>
Bofh: numerics
<bofh>
wait, why woupd you want to compute mean anomaly from true/eccentric, like, ever? I can't think of a situation where you hsve one if the latter but NOT the normer.
<bofh>
would*
<bofh>
former*
<bofh>
have*
<bofh>
yaey, typos
<Ellied>
!wpn bofh
* Qboid
gives bofh an antidisambiguated vernacular heptagon with a mine attachment
TonyC1 has quit [Read error: Connection reset by peer]
<Iskierka>
and the feedback is, literally, the % of people who answered a question correctly
<Iskierka>
the whole question 3, not even the parts
<soundnfury>
Iskierka: it's an entirely reasonable question... for a first-year _mathematics_ paper. Not sure what they're doing asking it of compscis o_O
<soundnfury>
!wpn Iskierka
* Qboid
gives Iskierka a Phoenician bomb
<UmbralRaptor>
c) comes from b) extended to reals?
<Iskierka>
... is it actually connected to the part above? It seems so out of the blue and improbable
<Iskierka>
and looking at the form of fibonacci on wiki, which I didn't realise actually had a solution ... it doesn't appear to even be the form they're using here?
<Iskierka>
I'm just gonna hope that computation isn't going to ask for recursion proofs and that'll stay with maths who use saner ones
<UmbralRaptor>
c) has an n and an n+1, so it looks like one could do a substitution?
<Iskierka>
I'm not sure how you show it to match by induction though
<UmbralRaptor>
er, I seem to be looking at it wrong,