The Nineteenth Byte

Martin Büttner

7:00 PM

this might be a nice instalment for random golf of the day :)

orlp

@AlexA. BACD is not BCAD, but not legal either

@MartinBüttner meh

Martin Büttner

lol

orlp

everyone will just use the O(N!) sampling with rejection

Martin Büttner

you can easily rule that out if you know something better exists

Alex A.

@orlp Oh right

orlp

7:01 PM

@MartinBüttner why?

I don't understand why

Martin Büttner

well I suppose there could be non-uniform multiplicities

orlp

I mean, look at the element at the top left

it has equal chance to end up in any position

the same goes for any other element

so why isn't it uniform?

Martin Büttner

@orlp except the one where it started

wait, not true

orlp

@MartinBüttner nope

you see my mental problem?

2

I can't explain to myself why it's not uniform

Sherlock9

@orlp I was running your code on (4,8) and it has now reached 18/32

But it detected
00011011
01101100
10110001
11000110
and not
00110011
11001100
00110011
11001100

For tiling the plane

orlp

7:09 PM

@Sherlock9 the latter is invalid

diagonals mate

Sherlock9

Where in the diagonals is there three in a row?

orlp

or maybe the formatting is putting me off

@Sherlock9 is the program done?

Sherlock9

Nope

Still somewhere between 18 and 19

I have it print when it's starting a new number

orlp

ah that's because it's still my old version

once it has found a better best_s it will never print something older

even if it tiles

Sherlock9

Ah fair enough

New code?

orlp

7:12 PM

gist.githubusercontent.com/orlp/7583b57f25c343a62098/raw/…

Sherlock9

Oh it finished

orlp

at?

Sherlock9

I think it didn't check 19, so it must be 18, except the only one it printed that tiled the plane is 16/32

Martin Büttner

@orlp I can't even explain why it would give a finite probability for every possible latin square

finite == non-zero

orlp

@MartinBüttner every latin square has one of ABCD (for N=4) once in every row/column

each row/column is a permutation of ABCD

let's say I have

ABCDE
BCDEA
CDEAB
DEABC
EABCD

can you find a latin square which can not be generated by permuting rows/columns of this one?

Sherlock9

7:16 PM

You know what might help? Sort row_possibs by s

Would add extra overhead, but it would help with determining where in the heck it could work

orlp

@Sherlock9 do you know how simple that is?

I'm dumb that I didn't do it earlier

row_possibs = (r for r in itertools.product([0, 1], repeat=DIM[1]) if valid_row(r))

row_possibs = (r for r in itertools.product([1, 0], repeat=DIM[1]) if valid_row(r))

well that doesn't quite do it, but it's already better :)

yep, after doing that it finds 20/36 for 6x6 in like 5 sec

so...

Sherlock9

:D

orlp

proving the bound is going to be annoying

Sherlock9

Code please

orlp

@Sherlock9 I showed you

literally swap 0, 1 for 1, 0

Sherlock9

7:19 PM

Oops sorry

orlp

at these sizes you're going to have to start looking towards C or something though

Sherlock9

I'm considering it

I think trying to code C at 2:20 in the morning might be a tad suicidal though

orlp

@MartinBüttner oh I think I can explain it

or nope...

nvm

Martin Büttner

@orlp not being able to find one is no prove that it doesn't exist ;)

orlp

@MartinBüttner you misunderstood me

if you could find one that can't be derived means you have proven it :)

Martin Büttner

7:26 PM

sure, but I was hoping you'd have a prove that each one can be derived :P

Sherlock9

Did your (6,6) get you anything that tiles the plane, @orlp?

orlp

no I stopped it as soon as it found > 18/36

we already have tilings for 1/2

what's the point in finding more?

Martin Büttner

finding ones for > 1/2

Sherlock9

I'm trying to find a tiling > 1/2

ninja'd

orlp

I don't believe it exists

Sherlock9

7:29 PM

Half the point of this insanity is finding a tiling > 1/2

Very well, I'll keep looking

orlp

actually

can't we exclude a ton of numbers

by using LCM?

we don't need to check multiples

e.g. 4/8 is a lot easier to check than 8/8

feersum

How would that work.? Couldn't you have an 8/16 with a minimum period of 4x4?

orlp

@feersum I guess if you find a partial tiling (horizontal or vertical) for NxM with size > 1/2 you don't need to check Nn x Mm for any n,m anymore

if you're looking to make the bound 1/2 tight

feersum

you're trying to show that a 1/2 tiling exists for every period ?

orlp

no, I'm trying to find a small N x M to bruteforce that has an optimal solution of 1/2

because that would show that the infinite plane is optimal at 1/2

Martin Büttner

7:38 PM

if you have a 4x8 solution, that doesn't imply that 8x8 can handle the same density

because what we find for 4x8 is not necessarily a tiling

orlp

@MartinBüttner no, but if it's partially tilable it does imply that 8x8 can't be worse than that

e.g. if we find that 3x6 can tile horizontally for 5/9, then we don't need to check 6x6 anymore

since that can also have density 5/9

Martin Büttner

Of course

feersum

I don't see how this proves anything about the infiite plane.

Martin Büttner

That's why we stopped checking 3xN

orlp

@feersum If I can brute force, say, 6 x 8, and find that it can't contain more than 1/2 on squares

then that means the infinite plane can't have more than 1/2 on squares

feersum

7:41 PM

6x8 tiling or non?

orlp

doesn't matter

or well - I guess non

feersum

seems implausible to find a max 1/2 with non-tiling

surely you can always have more at teh edges

orlp

nooo

I'm trying to find one where nothing above 1/2 can exist

oh wait, that's what you meant with 'max' isn'tit

feersum

I would conjecture that you can always find > 1/2 for a non-tiling finite rectangle

Sherlock9

And I would conjecture that a tiling > 1/2 exists. Just for the sake of have a conjecture out there. The evidence thus far does not in any way support this conjecture.

feersum

7:52 PM

Hmm how many people here are Lembik's sock?

Sherlock9

I'm sorry. What is a sock? Secondary account?

feersum

Correct

orlp

@feersum did I miss some drama?

why would he be using sockpuppets?

Sherlock9

Why sock? And why is our chat called "The Nineteenth Byte". I assume there are stories behind this

feersum

I suppose this comment only makes sense if you're in this chat room way too much

Sherlock9

7:54 PM

I just wonder at the origin of the terms

Alex A.

@Sherlock9 We score things in bytes and there are nineteen holes in a golf course.

Sock is short for sock puppet, i.e. an account that's controlled by another user, as one would control a sock puppet.

Sherlock9

Thanks @AlexA.

And aren't there usually 18?

Or am I being n00bish again?

VisualMelon

The 19th hole is the pub at the end of the golf course

Alex A.

Idk, I don't know actual golf, only code golf.

@VisualMelon Is that legit?

Sherlock9

XD @VisualMelon

Alex A.

7:57 PM

I'd believe it

VisualMelon

Nineteenth hole

In golf, the nineteenth hole is a slang term for a pub, bar, or restaurant on or near the golf course, very often the clubhouse itself. A standard round of golf has only eighteen holes, so golfers will say they are at the 'nineteenth hole', meaning they are enjoying a drink after the game. The concept is similar to Après-ski in skiing. The 19th hole on miniature golf courses is often a hole in which if a hole-in-one is scored, one receives a free game. == References in mediaEdit == The golf stories of author P. G. Wodehouse, which are narrated by his character, the Oldest Member, discuss ...

Alex A.

Regular golf seems sinfully boring; one could only hope that the reward at the end of the game is a nice beer.

@VisualMelon I had no idea.

Sherlock9

Thanks for going with me on this ball-whacking charade. Come on, let's grab a pint

VisualMelon

and you call yourself a mod? ;P

congrats on that, BTW, I don't think I was around when that happened

Alex A.

Oh thanks! :D

randomra

8:00 PM

alex-afraids-he-would-be-wrong

Alex A.

I think community consensus is that I am wrong. I can no longer live in fear of being wrong since clearly it's a perpetual state.

VisualMelon

Was I wrong to save your species then?

(Rather than let the foxes of truth hound you out)

Alex A.

At it turns out we've been hunted to extinction. I am the sole survivor.

VisualMelon

and you've evolved into a Magpie(esq being)

Alex A.

This is true.

As it turns out, geese are shitty creatures.

VisualMelon

8:02 PM

Geese are fantastic!

Alex A.

VisualMelon

is Bigtoes still around?

randomra

*fanatics

Alex A.

Have you not been around since he was Bigtoes?

@Geobits is alive and well.

VisualMelon

good to know, golfing C# is always more fun when there is a good Java answer to compete with

though I don't like this C# 6 nonsense... far too many golf-friendly features...

Alex A.

8:05 PM

I... never expected to hear that from someone who likes both C# and code golf.

Optimizer

C short

Alex A.

:O

Not a bad C#-based golfing language

VisualMelon

I love C# because it's simple, I love code-golf for some other reason, I'd rather they stopped adding features to C#

Optimizer

@AlexA. y :O?

Alex A.

@Optimizer I was using :O to denote a gasp of realization

VisualMelon

8:07 PM

(also I'm too lazy to download a C# 6 compiler, so I can't use the features anyway)

Alex A.

Doesn't that take like, I don't know, all of about 4 minutes?

:P

Optimizer

@AlexA. not everyone is in the luxury of good fast internet

VisualMelon

I did once try, but I couldn't find it or work out whether VS ships with it or not yet

I don't think my IDE of choice supports it yet, but that's just a feeling, I've not checked

Optimizer

@VisualMelon intellij?

VisualMelon

eurgh, no

SharpDevelop

and no-one mention Eclipse

Optimizer

8:12 PM

intellij is really nice, I dont think it deserves eurgh

VisualMelon

it's just the J on the end

Optimizer

J is really nice golfing language, you know!

VisualMelon

I hear it isn't nearly as painful as Eclipse, but I also know people who like Eclipse, but these people usually haven't used VS

Optimizer

also, why would an IDE named I C# code not have a C# compiler inbuilt?

Alex A.

@VisualMelon Wasn't planning on it. Eclipse is like trying to drive a semi truck, whereas other IDEs are like sedans and sports cars. :P

@Optimizer I agree, IntelliJ IDEA is the best Java IDE I've seen.

VisualMelon

8:13 PM

@Optimizer it does, but I don't like updating software

feersum

Why would VS not have the latest C#? Isn't Microsoft the main driver of that language?

Optimizer

intellij although being a bit heavy is really nice on having deep integration for language features for majority of languages

@AlexA. not just java, from my experience, JS too

Alex A.

@Optimizer Do you know J? It makes no sense to me. :P

Optimizer

i dont

VisualMelon

@feersum I imagine it does, but I couldn't work it out

Alex A.

8:15 PM

@Optimizer Yeah, and I think it has plugins for a variety of other languages as well.

Optimizer

languages and frameworks!

it has logical support for almost all major frameworks, web or backend.

Alex A.

That's awesome!

Have you tried Atom?

Optimizer

yeah

Alex A.

Do you like it?

Optimizer

but the issue with Atom/ST3/similar things is that they are just code editors with a bit of intelligence

they are not full blown IDE with all kinds of voodoo magic support

like intellij can understand my gulpfile.js and pull out the tasks as run configurations and what not. It can read my eslintrc config file and apply those rules itself to show the errors/warnings in the UI.

I liked ST better than Atom, mainly because of speed and support.

Alex A.

8:20 PM

ST is Sublime Text?

Optimizer

the only thing missing in intellij woudl probably be the advanced multi cursor support of ST

@AlexA. yes

Alex A.

Atom has a lot of plugins that can make it function like an IDE. I really like how extensible it is.

And it's FOSS. :D

Optimizer

have you tried brackets?

does atom understand your code?

Maltysen

atom is not really an ide, but as an editor it is AWESOME

Optimizer

like variable/function usage and related stuff?

Maltysen

8:23 PM

@Optimizer yes

Optimizer

how? using tern?

Maltysen

and you can navigate to the part of your code where they are defined

@Optimizer I'm not really sure

Optimizer

atom is JS based, right ? just like brackets..

Maltysen

yeah it runs in an entire modified instance of chrome

Alex A.

@Maltysen Uh, I don't know about that. Are you sure?

@Optimizer No, I haven't tried Brackets but it looks nice.

SuperJedi224

8:28 PM

Hey guys

Optimizer

@Maltysen yeah, that is the main concern with both atom and brackets. That is also why they cant match speed of ST

Maltysen

@AlexA. atom.io/docs/v1.2.4/getting-started-why-atom

Alex A.

> Atom is a specialized variant of Chromium designed to be a text editor rather than a web browser. Every Atom window is essentially a locally-rendered web page.

Wow! TIL.

Optimizer

they just took the browser out of the developer tools!

flawr

9:37 PM

Anyone here?

Nathan Merrill

hi

SuperJedi224

I am

@NathanMerrill Hi

flawr

I was thinking about doing a challenge where you get an ascii representation of rubberbands laying on a table, and you have to determine whether they are entangled or not.

But now I am not sure whether this is actually solvable=/

Peter Taylor

I don't understand what's going on with the chemistry questions and close/reopen votes. codegolf.stackexchange.com/q/64862/194 was closed today as a dupe of codegolf.stackexchange.com/q/11816/194 , which has just picked up two reopen votes, but the former adds more variety to the site than the latter.

@flawr codegolf.stackexchange.com/q/30292/194 ?

flawr

@PeterTaylor Yes that was the question I had in mind too, at first glance it seems about of equal difficulty compared to the problem of of finding out whether something is a knot or not (unknotting problem) right?

Peter Taylor

9:41 PM

I think it's the same problem.

Alex A.

^

flawr

I am not so sure about that.

Peter Taylor

Pick a segment from each such that nothing else overlaps; cut the segments and cross-link them; test whether you have a knot or an unknot.

flawr

You are saying that if you can solve the unknotting problem you can also solve the -lets call it- rubber band problem.

But I think the rubberband problem is actually quite a lot easier.

Peter Taylor

The reduction works the other way too.

flawr

9:50 PM

@PeterTaylor I do not completely understand this procedure, can you elaborate it? What do you mean by cross linking? (And by picking a segment that 'nothing' overlaps?)

New Main Posts

0

Q: Non-quadratic equation method of solving the roots of a quadratic trinomial

Like the question says, think back to your High-school algebra class. Write a program which solves for the roots of a Quadratic trinomial WITHOUT using the normal quadratic equation. (HINT: Think completing the square or factoring). The winner will be the smallest example (smallest in terms of by...

quartata

@NewMainPosts lmao

Field day for Mathematica

aditsu

the "normal quadratic equation" is completing the square

quartata

^

I didn't even notice he said that until I read it again.

Oh he edited it

> You may not just write a calculator using the standard quadratic equation (thats this: practiceit.cs.washington.edu/images/quadratic.gif

@flawr ffs I was literally just typing that

Maltysen

uhh he just deleted it, should I undelete, close vote?

quartata

9:59 PM

Awh man I was just trying to figure out how to do it in Mathematica with just the coefficients too

Alex A.

@Maltysen No

quartata

@Maltysen Nah, if he deleted it he wants it to stay like that

flawr

@Maltysen What benefit would we get?

Peter Taylor

@flawr To link two bands into one: pick one segment on each band such that you can draw a simple curve (i.e. topologically equivalent to a circle) through them which only contains the segments; suppose the segments are AB and CD. Delete those segments and add (AC and BD) or (AD and BC), whichever doesn't introduce an intersection.

Maltysen

no he specifically said, "Ok, prehaps I need to be a little more specific here. Give me a minute I'll update the rules."

Peter Taylor

9:59 PM

For the unlinking, it's exactly the same.

quartata

@Maltysen That was before this:

codegolf.stackexchange.com/questions/65015/…

Transcript for