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.
<egg|zzz|egg>
bofh: today in genetic algorithm things: taking a power of the log of a probability
<UmbralRaptor>
egg|zzz|egg: Are you *sure* that you aren't doing stat mech?
<bofh>
UmbralRaptor: the two are honestly very closely related...
<B787_300>
What would taking the power of the log of probability give you? Or rather why do you have your probability in log form and are trying to get it back to a normal form
<bofh>
Yeah, now that you mention it I am slightly confused and think you want to do operations in a slightly different order here.
<UmbralRaptor>
bofh: admittedly, ergodicity is probably indirectly relevant.
<B787_300>
!define ergodicity
<bofh>
B787_300: time average equals ensemble average
<UmbralRaptor>
B787_300: a small chunk of a thermodynamic system ~equilibrium will go through every state of the whole system, given enough time. And time spent in a given state = amount of large system in that state.
<UmbralRaptor>
(bofh's version is more concise)
<egg|zzz|egg>
B787_300: bofh: I have my probability in log form because it would underflow otherwise, and the power is just because I need to steepen the fitness function so that the genetic algorithm converges
<egg|zzz|egg>
(the exponential would be way too steep, underflowingly so even)
* UmbralRaptor
wonders if people who modeled the system before included code.
<egg|zzz|egg>
some of them
<B787_300>
Underflow? As in be too small to accurately simukate
<egg|zzz|egg>
underflow as in it's too smol for IEEE 754 binary64
<egg|zzz|egg>
whitequark: okay so script smol l is probably one of those legacy-for-interchange-shitty characters, see recommendation 3 above
<egg|zzz|egg>
but it seems people tend to use it instead of 1D4C1
<egg|zzz|egg>
stabbity
<whitequark>
aw
<whitequark>
wtf
<whitequark>
where did that ℓ come from
<whitequark>
I dind't type it for sure
<egg|zzz|egg>
whitequark: no you didn't type it, I was just looking at the semilatus rectum article on wikipedia
<egg|zzz|egg>
whitequark: and they use ℓ instead of 𝓁 which I think isn't appropriate (at least according to the Irish defect report)
<whitequark>
egg|zzz|egg: oh
<whitequark>
you typed it and now it's stuck in my terminal emulator
<egg|zzz|egg>
oh lɒl
<egg|zzz|egg>
whitequark: somehow I am good at breaking the systems of other people with UCS silliness
<egg|zzz|egg>
whitequark: I permanently broke someone's IRC client at some point with the egyptian hieroglyph for star followed by the sumero-akkadian cuneiform sign for star
<egg|zzz|egg>
(somehow other combinations of astral characters didn't do that, it was specifically this egypto-akkadian astral curse that broke their client)
tawny has quit [Ping timeout: 194 seconds]
<whitequark>
lo
<whitequark>
ℓoℓ
<egg|zzz|egg>
:D
<whitequark>
котя is back
APlayer has joined #kspacademia
<egg|zzz|egg>
whitequark: yay! :D
<egg|zzz|egg>
whitequark: котяpics?
<whitequark>
i'm laaazy
* egg|zzz|egg
pets whitequark
<egg|zzz|egg>
!wpn whitequark
* Qboid
gives whitequark a steel-cut functional
egg|phone|egg has quit [Ping timeout: 182 seconds]
<egg|zzz|egg>
bofh: wait the product of the p-values for each transit is probably a horribly silly thing, right?
* egg|zzz|egg
stares at χ-squared
* egg|zzz|egg
is confused
* egg|zzz|egg
stabs statistics
APlayer has quit [Read error: Connection reset by peer]
APlayer has joined #kspacademia
<egg|zzz|egg>
bofh: um χ² = 150 000ish with n = 282 >_>
<SilverFox>
hey guys, yall know that saying "function over form", so what do you get when you do function / form?
<bofh>
yes, the product will be silly, yes.
<egg|zzz|egg>
bofh: yeah but then again the χ² is silly because I am terrible at optimization so
* egg|zzz|egg
stabs things
egg|phone|egg has joined #kspacademia
APlayer has quit [Ping timeout: 182 seconds]
<SnoopJeDi>
bofh, that's a wonderful image
APlayer has joined #kspacademia
tawny has joined #kspacademia
tawny has quit [Remote host closed the connection]
<Qboid>
(1,2): error CS0103: The name `e' does not exist in the current context
<bofh>
well other than the fact that there's no justification that f(x) is analytic
<bofh>
!c M_E
<Qboid>
(1,2): error CS0103: The name `M_E' does not exist in the current context
<SilverFox>
it's porn math
<SilverFox>
math written on boards during porn to make it seem mathy
<bofh>
(if f(x) is analytic and that integral is taken over a compact subset of the radius of convergence of the power series then that integral/sum interchange is valid, mind you).
<bofh>
ROFL THAT'S OUT OF PORN???
<SilverFox>
yeah
<SilverFox>
that's a porno
<SilverFox>
that's why I said "Have some porn math"
<egg|zzz|egg>
bofh: phl just noted that it works slightly better if the mutation rate goes down with time
<egg|zzz|egg>
bofh: is our genetic algorithm going to have an annealing schedule,
<SnoopJeDi>
bofh, remarkably accurate, considering the context!
<SnoopJeDi>
guess somebody had a math text on hand
<bofh>
egg|zzz|egg: I mean,
<egg|zzz|egg>
bofh: see the radiation source that induces the mutations decays,
tawny has quit [Remote host closed the connection]
tawny has joined #kspacademia
<egg|zzz|egg>
!wpn коте, whitequark, and the котяchrome kitten
* Qboid
gives коте, whitequark, and the котяchrome kitten a bustard
* egg|zzz|egg
assumes the котяchrome kitten will eat the bustard
<bofh>
18:19:47 < SilverFox> g=10 gets me
<bofh>
g=10 is close enough for some government work.
<SilverFox>
disgusting
<SilverFox>
what's g at top of mount everest?
<SilverFox>
is it higher or lower than sea level g?
<SilverFox>
at first thought I'd think higher, but idk
<bofh>
Lower, consider G*m1*m2/r^2, m1 ~ 1, m2 the Earth's mass, and r the Earth radius vs the Earth radius + the height of Mt.Everest.
<UmbralRaptor>
I'd be hesitant, given the non-spherical nature of earth. Got a gravity map on hand?
<egg|zzz|egg>
!wpn bofh
* Qboid
gives bofh a polonium code
<egg|zzz|egg>
!wpn UmbralRaptor
* Qboid
gives UmbralRaptor an IEEE 1394 traversal
<egg|zzz|egg>
!wpn whitequark
* Qboid
gives whitequark a surjective grammar
<UmbralRaptor>
!wpn egg
* Qboid
gives egg a hacked field extension
<egg|zzz|egg>
whitequark: why are you still a black square on github
APlayer has quit [Ping timeout: 182 seconds]
<egg|zzz|egg>
whitequark: also how is the grey kitten
<egg|zzz|egg>
bofh: re. the integral, tbh we don't have much context so perhaps they discuss whether f is sufficiently regular in the dialogue
<UmbralRaptor>
egg|zzz|egg: pulling/pushing a cat is likely to annoy it, but a black square will be fine.
<egg|zzz|egg>
hmm
* egg|zzz|egg
just started typing synapsid in the address bar instead of mastodon
<egg|zzz|egg>
!wpn UmbralRaptor
* Qboid
gives UmbralRaptor a Dirac ネコバス
<UmbralRaptor>
A ネコバス capable of traversing the sea?
<egg|zzz|egg>
I mean that doesn't seem unreasonable
<SnoopJeDi>
猫ノーチラス号
<UmbralRaptor>
Something like that.
<UmbralRaptor>
!wpn -add:adj skaarj
<Qboid>
UmbralRaptor: Adjective added!
<UmbralRaptor>
!wpn -add:adj nali
<Qboid>
UmbralRaptor: Adjective added!
awang has quit [Ping timeout: 202 seconds]
<bofh>
egg|zzz|egg: touché.
<egg|zzz|egg>
meanwhile I am bad at optimization
<egg|zzz|egg>
bofh: what's the p value for χ² = 1511078 n = 282 >_>
<bofh>
AAAAAAAAAAAAAAAAAAAA
<egg|zzz|egg>
:-p
<egg|zzz|egg>
bofh: okay but now I managed to get to χ² < 130 000
<egg|zzz|egg>
bofh: which is still so stupidly high that I don't know how to compute the P-value, but it's an order of magnitude smaller! :-p
<bofh>
knock it down another two orders of magnitude and then we'll talk :P
<egg|zzz|egg>
bofh: but I don't know how to do that :-p
<UmbralRaptor>
Something something χ²/ν.
<UmbralRaptor>
But still…
<egg|zzz|egg>
UmbralRaptor: 282 is smol
<egg|zzz|egg>
bofh: yeah with an other order of magnitude throwing 2000 sig. dec. intermediate precision at it means I can compute the log of the P-value
<egg|zzz|egg>
it's -5826
<egg|zzz|egg>
bofh: and with two orders like you ask, 100 sig. dec. is enough! and you get log P = -298
<bofh>
rofl
<egg|zzz|egg>
bofh: which probably still means that you can confidently say that the system isn't the one you observe :-p
<egg|zzz|egg>
bofh: argh my mutation rate cools too fast
<egg|zzz|egg>
(I'm not sure what kind of a monstrous sausage of an optimization algorithm I've made here)
<egg|zzz|egg>
!u 𒆪𒋾𒀀𒂍𒊹𒊏
<Qboid>
U+121AA CUNEIFORM SIGN KU (𒆪)
<Qboid>
U+122FE CUNEIFORM SIGN TI (𒋾)
<Qboid>
U+12000 CUNEIFORM SIGN A (𒀀)
<Qboid>
U+1208D CUNEIFORM SIGN E2 (𒂍)
<Qboid>
U+122B9 CUNEIFORM SIGN SHAR2 (𒊹)
<Qboid>
U+1228F CUNEIFORM SIGN RA (𒊏)
<bofh>
uh is SHAR2 supposed to just be a circle?
<bofh>
oh, that's Tiglath-Pileser I
<bofh>
hurr
<egg|zzz|egg>
bofh: well that's tiglath-pileser, any of them :-p