The h Bar

user351417

16:00

@Loong So... Were you bored? =)

user351417

I figured it was an interesting thing to see in a sandbox, so I put it here.

user351417

In other news... physics.meta.stackexchange.com/a/11262

user351417

I wonder who Emilio Pisanty was talking about when he said he was expecting some ~10k rep user(s) to nominate themselves.

user351417

Or did he say he was hoping they would? Whatever.

ACuriousMind

@Chair you could ask him :P

But it was "hoping", not "expecting", indeed

user351417

16:07

He sounded like he was deliberately avoiding a name, but I guess implied meanings are different over the internet, and I have no idea if I'm missing some background.

ACuriousMind

No, I agree he was likely not being explicit to avoid appearing to put pressure on them

user351417

Eh, he'll probably see AFT's answer (or he'll notice that we don't have any nominations yet), and he'll say it if he feels it's appropriate.

Sir Cumference

0

A: 2019 Community Moderator Election

So... the elephant in the room? The nomination phase is almost over, and we have no candidates. Shouldn't we be concerned? What happens if nominations close and no one decides to run for moderation? Can we, the rest of users, nominate someone else and hope they'll agree?

Uhh...

@JohnRennie the hero we need

John Rennie

@SirCumference :-)

Being a mod would actually take away some of the freedoms I enjoy now.

For example I can vote to close fairly freely knowing that four other people have to agree with me for anything to happen. As a mod I would be very hesitant to VTC except in the most egregious cases.

Sir Cumference

I mean you're already most of the way there. All the privileges unlocked, room owner powers, well known in the community, etc. Honestly the only difference is you'd be an official mod :P

Loong

16:17

@SirCumference That's why, on Chemistry, we checked first before we started the election:

20

Q: 2019 Potential Moderator Election Community Interest Check

Quick Update We welcome the reception this post has gained, and appreciate all of you who are considering to run. We will therefore hold an election. As soon as new details are available, they will be featured in a separate Meta post. Again, thank you all for your participation. This communit...

JMac

it's way more relaxing to participate in sites instead of moderate them

Sir Cumference

@JohnRennie Fair enough, seems odd that there isn't a VTC option for mods

JMac

@JohnRennie I would be terrified of automatic mod powers. Is it you who I'm thinking of who is usually really reluctant about closing with your automatic close badge?

John Rennie

@SirCumference the problem is that for mods a VTC immediately closes the question.

Sir Cumference

@JohnRennie Yeah I meant there's no real option to start a "vote"

John Rennie

16:20

@JMac the dupehammer, yes. I will often vote to close as Off topic/Other and type a Possible duplicate of message to avoid immediately hammering the question.

JMac

@JohnRennie Right, it's only for dupes, I forgot how that worked

Sir Cumference

I mean that said you can probably recognize when a question needs closing as well as any other mod can, given you've been here for longer than most people on the site

John Rennie

@SirCumference The thing is that I believe strongly in the site as a resource for budding young physicists. My going in point is that you should be able to use info from any answer of mine in an exam and get full marks.

JMac

... I wonder if it would be a terrible thing for a mod to use an alternate account for regular flagging. That seems like it would be a really legitimate use of secondary accounts

John Rennie

But there are people, naming no names but we all know who I mean, who are quite happy to type answers that would get you zero marks in an exam.

ACuriousMind

16:23

@JMac You get used to it. Frankly, the binding VTC is one of the least destructive mod powers if used wrongly.

John Rennie

And that makes me really, really angry. To the point where I would find it really challenging to deal with that sort of person if I had mod powers.

JMac

@ACuriousMind I could understand that. I also totally get why it makes people reluctant to take the position

Regular review queues sometimes have me really unsure

knzhou

@JohnRennie That's another important reason I don't want to be a candidate.

Sir Cumference

@JohnRennie Yeah, I know what ya mean

knzhou

I'm happy if somebody else takes up the messy business of dealing with my flags.

ACuriousMind

16:27

@JMac Sure. And on the other hand, moderators must be willing to use the tools they're given, and they will get attacked for it. (aka "You aren't a mod until a user you just took action against has cast profane aspersions on your entire ancestry in response")

Sir Cumference

Well with regard to this current election it's pretty weird. Weren't there a bunch of candidates last time?

knzhou

Jeez, I get profane aspersions just as a regular user, I must be doing it all wrong!

bolbteppa

I could be the Trump (or Bernie!) of this election and pardon the Ocelot's and Ron's etc, but no time/energy and it's a waste, ahh...

John Rennie

@SirCumference my recollection of the last election was that a lot of candidates appeared in the last 48 or even 24 hours.

JMac

@ACuriousMind Yeah, exactly. I think it's pretty obvious why some people might want to avoid that stress :P

Sir Cumference

16:28

Let's hope then. If only one person nominated themselves would they get the position by default?

ACuriousMind

Yes

vzn

← defn not a fan of the mod VTC "hammer" either but am amused at the idea that it would be the main sticking point against volunteering as a mod o_O

Sir Cumference

@ACuriousMind Well thankfully this hasn't gotten the attention of a certain crank then.

John Rennie

@SirCumference he can't stand.

knzhou

If it came to that I'd jump on the grenade and self-nominate...

John Rennie

16:31

Being banned in the last year disqualifies you.

Sir Cumference

Oh huh, TIL

JMac

I'm pretty sure there would be ways in place to manually prevent something like that, no? Like for example a user with a candidate score of 2?

ACuriousMind

If someone thinks defaulting could be disastrous, let me remind you that users that have been suspended during the last year cannot nominate (unless there are extenuating circumstances discussed in advance with the SE team).

Sir Cumference

At any rate a 1 option moderator isn't really the best idea imo. It literally hands power to someone without anyone agreeing on how they'd affect the community

JMac

Wouldn't there be some method of intervening besides just having the nominated person rescind the nomination?

ACuriousMind

16:33

@JMac The main method of intervention is other users nominating ;)

Sir Cumference

Is there a reason we'd need to hand out power to the first (if only) person who asks?

JMac

@ACuriousMind Okay, so the panic button nomination strategy would have been the actual solution there I guess. I thought that worst case scenario, I'm sure a few people would have popped up had that situation happened... but still seems strange that nothing could be done

ACuriousMind

I described SE elections in the last election already as an elaborate game of chicken: Given the choice between 1. another highly qualified user becomes moderator and 2. oneself becomes moderator, almost everyone prefers 1. So most qualified users won't nominate until they are convinced that no one they consider more qualified than themselves will nominate

4

JMac

Yeah that makes a lot of sense with how this process is structured

It's kinda similar to online auction sniping in a way

Emilio Pisanty

@ACuriousMind last time was quite different, though

I think the conditions that made that game of chicken so obvious are no longer in place

JMac

16:38

Was it like background site drama last election or something?

ACuriousMind

@EmilioPisanty The last election had a lot of nominees (and non-nominees) who had discussed their strategy for the chicken game already in advance in this chat ;)

Sir Cumference

@EmilioPisanty why not apply for mod? You're probably more active on the site than anyone else who'd run

JMac

Was it a lot of "I won't run unless so-and-so isn't running"?

Emilio Pisanty

@SirCumference which is why I think I'd make a bad candidate

I'd like to see a fresh generation coming in

@JMac more of an "I'm running because so-and-so is running and that'd be awful", as I recall

2

JMac

Fun

Saksham Rastogi

16:41

What do you guys think about the only nomination as of now, by TechDroid?

JMac

Spite-running is always a good call

ACuriousMind

@JMac Not explicitly, but for instance I partly nominated so early because I already knew that many people I'd really liked to have seen running wouldn't run at all. There were also several, ahem, low-quality nominations (that were later removed) that I think sort-of forced early nominations from other users

Sir Cumference

@EmilioPisanty Welp if we don't find 2 mods in time perhaps it'd be something to consider?

ACuriousMind

@SakshamRastogi They retracted their nomination.

knzhou

@SakshamRastogi There are zero candidates right now.

vzn

16:41

2016 3rd election pgs physics.stackexchange.com/election/3 physics.meta.stackexchange.com/questions/9256/…

JMac

TechDroid had good intentions, they were just too inexperienced to even understand the scope of what they were proposing

Sir Cumference

Maybe it's just me but none of the most active people on the site seem to be interested :/

Saksham Rastogi

@ACuriousMind Oh. Why? (Sorry I'm relatively new to this)

Sir Cumference

Oh @Jim said he was interested last year iirc

ACuriousMind

@SakshamRastogi Probably because of the feedback to their questionnaire answers, but that comment is all of the public statements we've got on that matter.

Saksham Rastogi

16:44

@ACuriousMind Hmm. I read those earlier.

Emilio Pisanty

@SirCumference and, on the other hand, I really don't think my current job conditions allow for it.

enumaris

16:56

ugh

John Rennie

Re this question: am I correct in thinking that only 2D (or 1D) matrices multiply is the way we normally multiply matrices. For higher dimensions we can take a tensor product but that's a rather different thing.

ACuriousMind

@JohnRennie Let's be clear about terminology here: An n-dimensional matrix is a 2D array of n^2 numbers, right?

Because I can't quite tell whether the question is inquiring about that or whether it uses "n-dimensional matrix" to mean a tensor of rank n.

John Rennie

I think the way the OP has posted the question they mean a n dimensional array to be an array with n indices.

I guess it's worth asking them to clarify in a comment.

ACuriousMind

Yeah, for tensors of rank higher than 2 there is more than one possible way to "multiply" them - it's called contraction over different tuples of indices, then.

enumaris

so I didn't think literally everyone would be working from home today...I'm the only one on my team here lol

John Rennie

17:07

@ACuriousMind so how would I write multiplication of $A_{ijk}$ and $B_{lmn}$ using Einstein notation?

(any of the possible multiplications)

ACuriousMind

@JohnRennie Well, my answer is that it's not clear what you mean by "multiplication"

But you could certainly call $A_{ijn}B^{nlm}$ "a multiplication"

Unless by "multiplication" you want something that gives another tensor of the same rank. In that case you can only multiply tensors of even rank.

John Rennie

@ACuriousMind Which would presumably give you a four index matrix $C_{ij}^{lm}$

ACuriousMind

@JohnRennie yeah (although I really cringe at calling it a "matrix". Do we really have to overload the term? :P)

John Rennie

I guess that's the problem. The work matrix strictly means a two index object. Anything with a larger number of indices isn't a matrix so asking what matrix multiplication means in that context is meaningless.

@JohnRennie what I mean is if 3-D, then a cuboidal. — Sanika Khadkikar 5 mins ago

enumaris

does the lack of spacing in the indices trigger anyone?

John Rennie

17:12

Aha!

enumaris

$C_{ij}^{~~lm}$

John Rennie

Critics everywhere! $C_{ij}{}^{lm}$

enumaris

XD

John Rennie

Damn, ninjad again! :-)

enumaris

you guys familiar with the Curry-Howard Correspondence?

ACuriousMind

17:16

vaguely, but yes

knzhou

Hmm, I tried to avoid "wasting" rep on a post that was taking off by setting a big bounty.

But now I'm still rep capped, further upvotes aren't doing anything.

enumaris

familiar enough to give me a 2 sentence overview?

knzhou

Is the rep cap on net rep, or just on rep via upvotes?

enumaris

that I can actually understand? :P

ACuriousMind

@knzhou It's on net rep via votes.

knzhou

17:17

Damn.

Oh well, the bounty was deserved anyway!

Emilio Pisanty

John Rennie

@knzhou it is possible to game the rep cap, if only in small ways.

Emilio Pisanty

0

A: Community Promotion Ads — 2019

2

let's get that ad up and running!

ACuriousMind

@enumaris Let's try: The H-C correspondence is the idea that asking whether a formal type is inhabited (i.e. there exists at least one object of the type) is a problem of exactly the same kind as asking whether a logical proposition is true.

enumaris

hmmm...here a "type" is like...the "type" of a variable in computing?

e.g. int, str, char, etc?

ACuriousMind

17:26

@enumaris Yes, but not all programming languages are suitable examples

enumaris

seems pretty abstract

ACuriousMind

They need to have "algebraic types", which at the very least means they need the notion of a product type (a tuple) and a sum type (an either-or type, like a union in C, or an enum in Rust)

The product type and its correspondence is probably the easiest to understand because the tuple is a pretty common structure.

Examples are really the key to demystifying the correspondence. Let's take the logical statement "A and B". Then it is true if and only if both A and B are true.

enumaris

I'm familiar with tuples, but not so much with a "sum type"

ACuriousMind

The corresponding statement about types is that for two types A' and B', the product type that describes tuples of one A' and one B' has at least one inhabitant if and only if there is at least one object of type A' and one object of type B', because otherwise you can't construct a tuple

@enumaris It's a pretty handy thing. Given two types S and T, it's a type that says "I'm either an S or a T"

enumaris

is there a version of this in python

ACuriousMind

17:32

(and by handy I mean that it's actually useful in the real-world languages where it exists)

@enumaris No, Python doesn't have a sum type

enumaris

alrighty

ACuriousMind

So, if you understand how the product type corresponds to the logical and, then you've basically already understood the idea of the correspondence ;)

enumaris

hmmm

ACuriousMind

To discuss the full correspondence, in particular what it does for "for all" and "there exists" quantfiers, we'd need to discuss some stranger types which you usually see as "generic types" in programming languages, if they exist (they occur e.g. in C++ templates, parametrized Rust types). I don't think Python has them either, because Python hides its type system from the user and I don't think it has use for generics in this sense, but I'm sure @PM2Ring will set me straight if I'm wrong

enumaris

my C++ knowledge is pretty poor

since it's only theoretical

I've not used it much

ACuriousMind

17:42

(To be clear: Very formally neither C++ nor Rust actually has these types, since they are made "ungeneric" (monomorphized) at compile time. but I don't know any other non-functional languages that have these types, and functional languages like Haskell are somewhat dissatisfying examples because they are explicitly built to have a type system for which the H-C correspondence works perfectly.

That is, these languages are in a way better examples, but they don't serve someone very well who doesn't yet understand the more abstract notions of types

PM 2Ring

Python isn't helpful for modeling this kind of stuff. In most languages, a variable is essentially a box for holding a value of a specific type. So you cannot assign a value to a variable if
the value doesn't have the type that the box was defined with. Although most such languages will perform safe conversions automatically, eg if you assign a 8 bit integer value to a 32 bit integer box. In contrast, in Python, a variable is just a nametag, with no intrinsic type. A Python object may have 1 or more nametags, or None, and its type is an intrinsic property of the object itself, not its namet…

enumaris

Don't think I've used a "functional programming" language

jeea

PLease look at this question someone :

6 hours ago, by jeea

In image b, the flux trapped in ring must be quantised so its value will be different from that inside ring in image 'a'

ACuriousMind

@enumaris Sorry, I must retract my statement just there. The generic types I was thinking of are not enough for true dependent types, and Haskell doesn't have them either.

enumaris

huh?

what's a true dependent type?

PM 2Ring

17:48

@enumaris Haskell will blow your mind. :) But for practicality, I advise you to learn C solidly first. I tried learning Haskell a few years ago, and although it was interesting, I found some of its concepts hard to grok. But I might give it another go...

enumaris

I don't have like...a project...to do in C so it's hard to learn

I'm very hands on when it comes to learning programming

I've read like a whole set of tutorials on C++ , but it's not like I could just go use it...

ACuriousMind

@enumaris The types that correspond to the logical quantifiers are so-called dependent types. I briefly fooled myself into thinking that generic types would be a close enough analogy, but they're not.

enumaris

python is just so convenient...it can handle everything I need to do...so I don't end up learning anything else lol

ACuriousMind

@enumaris Was what I said in any way helpful? :P Any specific questions?

enumaris

I think it's helpful at a high level. Can't say I have a great understanding right now, but that's probably just cus I don't have the background necessary to really understand it well

Thanks :D

PM 2Ring

18:00

I can relate to that. OTOH, I learned quite a few languages for a few decades before I learned Python. It's good to learn a few languages of different language families, just to get exposure to the various different ways that those families do stuff. Even if you end up doing most of your coding in Python, the awareness of those other languages will enhance the way you approach a coding problem.

enumaris

yeah, I'm sure learning C would be useful

but I just don't have the self-motivation to do it atm lol

PM 2Ring

Coding in C gives you more direct control over the machine, and the feeling of power from being so close to the metal is kinda nice. :) And the fact that code written in C runs 100 times faster than pure Python's rather nice too. Of course, Python often leverages the power of C, eg a lot of list operations can do most of their work at C speed. And for serious number crunching we can use Numpy, which does most of its stuff at C speed.

enumaris

yeah...I just use numpy :P

I guess if I can think up a concrete project that I would like to actually tackle with C

then I can learn C

Jim

18:17

@SirCumference I was convinced into running. Kind of glad I didn't win. And while I'm usually always around these days, I'm not attentive enough to make a good mod

bolbteppa

►

'XENON1T observes the rarest decay process ever happening in a detector!'

vzn

← lol moved away from C in early 2000s to scripting/ ruby and was delighted at time, amused at idea of opposite direction ~2decades later o_O :P

enumaris

I'm mostly interested in C cus CUDA is C

and massively parallel programming is interesting to me

vzn

C is a great language "of its time/ era". winner of a Turing award 1983 amturing.acm.org/award_winners/ritchie_1506389.cfm

bolbteppa

18:34

Video is about this: Dark-matter detector observes exotic nuclear decay

> 'a, An iodine-124 atom can decay with a half-life of 4.2 days to an atom of tellurium-124, through a process called electron capture. The nucleus of the iodine-124 atom captures an electron from the electron shells that surround it. A proton (circled) in the nucleus is converted into a neutron, and a neutrino is emitted. b, A xenon-124 atom cannot decay by electron capture, because of the law of energy conservation.

> However, it can decay with an extremely long half-life to a tellurium-124 atom, through a process known as two-neutrino double electron capture. The xenon-124 nucleus captures two electrons from the surrounding electron shells, which results in the conversion of two protons (circled) into neutrons, and the emission of two neutrinos. The XENON Collaboration2 has measured the half-life of this process to be 1.8 × 1022 years — about one trillion times the age of the Universe.'

The opening paragraph is incredible (bold font is mine)

> 'For half a century, our view of the world has been based on the standard model of particle physics. However, this view has been challenged by theories1 that can overcome some of the limitations of the standard model. These theories allow neutrinos to be Majorana particles (that is, they are indistinguishable from their own antiparticles) and predict the existence of weakly interacting massive particles (WIMPs) as the constituents of invisible ‘dark matter’ in the Universe.

> Majorana neutrinos mediate a type of nuclear decay called neutrinoless double-β decay, an example of which is neutrinoless double electron capture. A crucial step towards observing this decay is to detect its standard-model equivalent: two-neutrino double electron capture. In a paper in Nature, the XENON Collaboration2 reports the first direct observation of this process in xenon-124 nuclei, using a detector that was built to detect WIMPs.'

Novalium Company

18:51

Is it possible for consciousness to be just a genetic trait? Also why aren't dolphins smarter? Does the brain size matter for intelligence and consciousness...? Is it possible to raise a dolphin like a child

ACuriousMind

@NovaliumCompany If it was brain size alone that mattered, our interactions with whales would be a lot different :P

Novalium Company

Ending up with us being their slaves 😆

ACuriousMind

Don't project. Not all intelligences need to be as morally bankrupt as humans :P

Novalium Company

Hehe

Has anyone tried to plug up the data of a brain of EEG into some sorta machine learning algorithm and see what pops out?

ACuriousMind

@NovaliumCompany You can't really just throw some data into a learning algorithm and say "learn this".

Novalium Company

19:07

I understand it's complicated but hasn't anyone tried to do anything with brains and ml?

If we can't interpret our brains, maybe AI could?

SDFG

@NovaliumCompany Oooh. That would be fun

ACuriousMind

@NovaliumCompany I certainly don't have a full overview of what has been done in ML, but I have difficulty seeing what the problem here is at which machine learning is supposed to be better than traditional algorithms.

ML is not a cure for every problem.

3

PM 2Ring

Oh-oh. "A PMH is the simplest confirmation of -- what we've mistaken to be a separate force known as -- gravity generated by the magnetic lock of a hollow planet." From physics.stackexchange.com/a/476189/123208 The original version of that answer was relatively innocuous, but the sneaky SOB edited it 2 hours afterwards.

SDFG

@PM2Ring Sorry, but I'm new here. What's "SOB"?

PM 2Ring

@NovaliumCompany EEG data is a pretty crude indicator of what a brain is doing. fMRI data would be more interesting.

@SDFG It's an insult.

Novalium Company

19:13

In the case of neurotech and neuroimaging, I think the next step is developing a better interface between the brain and the computer? Anyone agree?

ACuriousMind

@PM2Ring fMRI data also is only valuable if analyzed with the correct statistical tools, otherwise you get salmons giving emotional responses from beyond the veil of death

SDFG

@PM2Ring Oh-kay, I get it now XD. I earlier thought it was some acronym used here on SE :P

ACuriousMind

Correct statistics, in turn, is something a traditional algorithm already excels at, so it's not clear what ML is supposed to improve

SDFG

@NovaliumCompany Well, partly. Frankly, this whole idea of "connecting" brain and computers seems a bit scary.

ACuriousMind

@NovaliumCompany Next step to what end? To end up in a real-life performance of The Matrix?

Novalium Company

19:21

All of these mathematics, physics and so on are just looking through the key hole and interpreting what we see, we call it through and now when we have the chance to open the door, people are scared and want to run away.

AI is coming. We either prepare, or go extinct. Both ways, it's still technically evolution.

Morality is relative, so that's not a valid argument.

It may be a scary future, I agree, but my curiosity is stronger than my fear.
And if I've learned something about fear, it's that some of the time if protects me from real dangers, but most of the time it just blocks me from seeing and doing more.

PM 2Ring

19:46

@NovaliumCompany Maybe it is, but nobody really knows how to create an AI. But some interesting stuff has been learned, and some useful stuff has been made in the process of working on AI. Eg, natural language processing has made a lot of progress. But we still don't have perfect machine translation because a good translator needs to understand what they're translating, unless the material is fairly simple.

But I'm concerned that someone might develop powerful AI before we've discovered how to make friendly AI. I'm no disciple of Eliezer Yudkowsky, but I agree with him that it would be a calamity to develop a true AI that wasn't friendly.

enumaris

20:26

@NovaliumCompany there's actually a whole branch of AI research that's based on trying to imitate the cognition of the human brain. It generally stands in contrast to the standard data-driven ML methods...but bringing the two together is an active area of research.

Surely someone is researching this :)

ethical AI is also an active area of research

e.g. OpenAI has a whole team devoted to AI ethics

enumaris

20:41

ahhhh

danielunderwood

22:16

hmmm

enumaris

welp, we are now officially a generic "AI" innovation center. No more healthcare domain...whomp whomp...

my research plan was written with the healthcare application in mind tho...so now I'm not sure what to research lol

and this is also why I'm not a big fan of documentation....produced 17 pages of documentation...research plans on how to go forward building the application...and then the project is canceled and now those research plans are worth dick

ACuriousMind

@enumaris write an AI that tells you what to research

enumaris

lol

rewriting that plan is gonna be annoying af

vzn

22:33

@enumaris (sorry to hear of the disruption, but...) the plot thickens! maybe mentioned this before, curiosity is a hot/ rising topic in ML/ AI research... can help you out with lots of juicy leads if you go in that direction :)

knzhou

@JohnRennie by unleashing a lot of downvotes and not paying the -1 rep price? I'm not sure what else...

enumaris

@vzn not the current direction unfortunately

Alright well...I guess I will write one more research plan...

if they make me write another one after that I'm gonna spew blood and die

danielunderwood

23:23

just find the thing to research that requires the least modification of your existing plan!

Transcript for