Logs: liberachat/#haskell
| 2021-07-29 18:29:35 | <lechner> | tautological mess is a contradiction in terms. with a tautology, you have absolute certainty! |
| 2021-07-29 18:29:44 | <motle> | its the fundamental construction of a partially specified stylesheet |
| 2021-07-29 18:29:53 | <roboguy_> | hm |
| 2021-07-29 18:30:04 | <motle> | you just do model extension putting tautologies everywhere and closing things and then adding decoration |
| 2021-07-29 18:30:07 | × | burnsidesLlama quits (~burnsides@dhcp168-019.wadham.ox.ac.uk) (Ping timeout: 268 seconds) |
| 2021-07-29 18:30:17 | → | doyougnu joins (~user@c-73-25-202-122.hsd1.or.comcast.net) |
| 2021-07-29 18:30:26 | <motle> | and out comes a half baked category theory!! |
| 2021-07-29 18:30:34 | → | burnsidesLlama joins (~burnsides@client-8-91.eduroam.oxuni.org.uk) |
| 2021-07-29 18:30:40 | <motle> | its a madness, damn ncatlab |
| 2021-07-29 18:30:41 | <juri_> | i'm not convinced this is a conversation. |
| 2021-07-29 18:31:06 | × | Atum_ quits (~IRC@user/atum/x-2392232) (Ping timeout: 250 seconds) |
| 2021-07-29 18:31:08 | <motle> | im just retorting to anyone mistakely trying to vex me with mention of AD |
| 2021-07-29 18:31:20 | <lechner> | maybe too much caffeine? |
| 2021-07-29 18:31:34 | <motle> | maybe too much prescision nudging! |
| 2021-07-29 18:32:04 | <Drew[m]> | <motle> "noo, its not a real concept!" <- I don't understand. Sure that example might be easy, but if you're combining a whole bunch of different differentiable functions it becomes not so easy pretty quickly surely. Why make the programmer do the work if we have the power to make machnes do it in at least some cases. |
| 2021-07-29 18:32:10 | <motle> | i was *trying* to talk about tolopogical differentiation of datatypes! |
| 2021-07-29 18:32:35 | <motle> | which as far as i can tell (.....) is all thats differentiationay about programs!! |
| 2021-07-29 18:32:54 | <motle> | i think AD is a misdirection in of itself, quit throwing things |
| 2021-07-29 18:33:29 | <motle> | nobody has demonstrated otherwise! the closest anyone can get is invoking the notion of normal differentiation |
| 2021-07-29 18:33:49 | <motle> | i have *actual* topological differention with these convolutional neighberhoods on pointer graph thing |
| 2021-07-29 18:34:01 | <motle> | none of your rubbish |
| 2021-07-29 18:34:16 | × | Atum__ quits (~IRC@user/atum/x-2392232) (Ping timeout: 258 seconds) |
| 2021-07-29 18:34:51 | <motle> | even topology would struggle to get that, so programming really helps here |
| 2021-07-29 18:35:32 | <Drew[m]> | Whatever concept of differentiation you're talking about, as far as I can tell automatic differentiation is a thing with a community agreed upon meaning and it's a useful thing and I just don't get why you were speaking ill of a useful thing |
| 2021-07-29 18:35:38 | <motle> | i just keep trying to get the class hierarchy, think it was pointer, navigate, comonad |
| 2021-07-29 18:36:33 | <motle> | Drew[m]: because sometimes when i try to learn things its just to find out if they are enough of a thing to learn about, or if they do what i think they infer they mean they do |
| 2021-07-29 18:36:50 | <motle> | i assumed they meant some kind of differentiation *other* than regular differentiation |
| 2021-07-29 18:36:55 | <motle> | apparently not |
| 2021-07-29 18:36:56 | × | Obo quits (~roberto@70.pool90-171-81.dynamic.orange.es) (Ping timeout: 252 seconds) |
| 2021-07-29 18:36:57 | <roboguy_> | I will say that the combativeness makes it difficult to engage in a discussion. |
| 2021-07-29 18:37:14 | <motle> | mate, iv been battling this AD thing for ages |
| 2021-07-29 18:37:35 | <motle> | its like, a main avenue of not a thing getting at me |
| 2021-07-29 18:37:41 | <Drew[m]> | Why would differentiation mean anything other than the normal sense of differentiating a function |
| 2021-07-29 18:38:01 | <motle> | well topology! |
| 2021-07-29 18:38:05 | <motle> | position in datatype |
| 2021-07-29 18:38:05 | <roboguy_> | I have worked with AD a bit myself. I've worked through that paper I mentioned carefully, read some of the references, looked through other implementations, read the followup paper where some similar techniques are connected to a generalized form of convolution |
| 2021-07-29 18:38:18 | × | merijn quits (~merijn@83-160-49-249.ip.xs4all.nl) (Ping timeout: 276 seconds) |
| 2021-07-29 18:38:38 | <motle> | how so? |
| 2021-07-29 18:38:46 | <roboguy_> | but it feels as though you do not want to hear any more about it. so that is that |
| 2021-07-29 18:38:48 | <motle> | i thought they werent allowed to be like differentiable datatypes? |
| 2021-07-29 18:38:52 | <motle> | nono |
| 2021-07-29 18:38:55 | <roboguy_> | you have made up your mined |
| 2021-07-29 18:38:56 | <roboguy_> | *mind |
| 2021-07-29 18:39:06 | <motle> | its a contradition, now yours seem topological all of a sudden |
| 2021-07-29 18:39:07 | <motle> | wft!? |
| 2021-07-29 18:39:29 | <motle> | boomerang of some sort |
| 2021-07-29 18:39:41 | <motle> | ok cool, so weer talking about the same thing |
| 2021-07-29 18:39:45 | <motle> | so stencils! |
| 2021-07-29 18:39:50 | <Drew[m]> | What does differentiation of datatypes even mean |
| 2021-07-29 18:39:58 | <motle> | zipper for the nav |
| 2021-07-29 18:40:10 | <motle> | presented in the SPJ as a suspended traversal |
| 2021-07-29 18:40:14 | <motle> | see a primative pointer |
| 2021-07-29 18:40:21 | <motle> | now graphs |
| 2021-07-29 18:40:28 | <roboguy_> | Drew[m]: http://strictlypositive.org/diff.pdf |
| 2021-07-29 18:40:28 | <motle> | nav |
| 2021-07-29 18:40:55 | <motle> | yeah that |
| 2021-07-29 18:40:58 | <motle> | thats a zipper |
| 2021-07-29 18:41:04 | <EvanR> | e.g. derivative of T^2 is 2*T |
| 2021-07-29 18:41:07 | <motle> | its a pointer of some kind see? |
| 2021-07-29 18:41:17 | <motle> | comonad follows |
| 2021-07-29 18:41:34 | <motle> | right so and then we need a class for many such things |
| 2021-07-29 18:41:50 | <motle> | and the nav |
| 2021-07-29 18:42:19 | <motle> | fine i guess its called AD whatever |
| 2021-07-29 18:42:33 | <motle> | but i never got stencils at the go |
| 2021-07-29 18:42:39 | <motle> | this is the API |
| 2021-07-29 18:43:02 | <motle> | because finite difference gradient approximations of various orders as common stencil uses |
| 2021-07-29 18:43:15 | <motle> | its more like, "because thats how differentiation works" |
| 2021-07-29 18:43:31 | <motle> | and the topology stuff is because we have places to put data not continuums of it |
| 2021-07-29 18:44:03 | <motle> | easyer to explain in programing actually, maths arent coping with it well - its called sheafification over there |
| 2021-07-29 18:44:09 | <motle> | stacks |
| 2021-07-29 18:44:31 | <motle> | and the heart of the stack sheaf is the topological neigberhood and constructing sections |
| 2021-07-29 18:44:47 | <motle> | which i didnt think was called AD... |
| 2021-07-29 18:45:00 | <motle> | but yeah, not records for lenses is how it hits here |
| 2021-07-29 18:45:20 | <motle> | regular grids as simple type level specifications for demo |
| 2021-07-29 18:45:32 | <lechner> | Hi, why do many people dislike 'printf' in favor of 'putStrLn', please? |
| 2021-07-29 18:45:38 | <EvanR> | me |
| 2021-07-29 18:45:57 | <EvanR> | oh i thought you said who dislikes printf in favor of putStrLn |
| 2021-07-29 18:46:29 | <EvanR> | printf has some good options but building the string yourself gives full control |
| 2021-07-29 18:46:52 | <lechner> | do you use ++ or <> ? |
| 2021-07-29 18:47:13 | <boxscape> | printf also gives you full control, you just need to use %s :P |
| 2021-07-29 18:47:17 | <roboguy_> | lechner: I believe there is, in effect, dynamic typing involved with printf |
| 2021-07-29 18:47:26 | <roboguy_> | unless I'm misremembering how the format string works |
| 2021-07-29 18:47:28 | <EvanR> | ++ or <>... doesn't matter. But concatting a list is another option |
| 2021-07-29 18:48:37 | <EvanR> | > concat ["This time is ", format theTime, " sweetheart"] |
| 2021-07-29 18:48:38 | <lambdabot> | error: |
| 2021-07-29 18:48:38 | <lambdabot> | • Variable not in scope: format :: t0 -> [Char] |
| 2021-07-29 18:48:38 | <lambdabot> | • Perhaps you meant one of these: |
| 2021-07-29 18:48:48 | <EvanR> | didn't mean to put > |
| 2021-07-29 18:48:52 | <roboguy_> | > printf "%s" True |
| 2021-07-29 18:48:54 | <lambdabot> | error: |
| 2021-07-29 18:48:54 | <lambdabot> | • Could not deduce (PrintfArg Bool) arising from a use of ‘printf’ |
| 2021-07-29 18:48:54 | <lambdabot> | from the context: PrintfType t |
| 2021-07-29 18:49:03 | <roboguy_> | > printf "%s" (123 :: Int) |
| 2021-07-29 18:49:05 | <lambdabot> | error: |
| 2021-07-29 18:49:05 | <lambdabot> | • Ambiguous type variable ‘a0’ arising from a use of ‘show_M693588695403... |
| 2021-07-29 18:49:05 | <lambdabot> | prevents the constraint ‘(Show a0)’ from being solved. |
| 2021-07-29 18:49:22 | <roboguy_> | that's different than what I get in GHCi... |
| 2021-07-29 18:49:29 | <EvanR> | > printf "%s" (123 :: Int) :: String |
| 2021-07-29 18:49:29 | <roboguy_> | > printf "%s" (123 :: Int) :: IO () |
| 2021-07-29 18:49:31 | <lambdabot> | "*Exception: printf: bad formatting char 's' |
All times are in UTC.