# Sandbox for Proposed Challenges

This "sandbox" is a place where Code Golf users can get feedback on prospective challenges they wish to post to main. This is useful because writing a clear and fully specified challenge on your first try can be difficult, and there is a much better chance of your challenge being well received if you post it in the sandbox first.

Sandbox FAQ

## Posting

Write your challenge just as you would when actually posting it, though you can optionally add a title at the top. You may also add some notes about specific things you would like to clarify before posting it. Other users will help you improve your challenge by rating and discussing it.

When you think your challenge is ready for the public, go ahead and post it, and replace the post here with a link to the challenge and delete the sandbox post.

## Discussion

The purpose of the sandbox is to give and receive feedback on posts. If you want to, feel free to give feedback to any posts you see here. Important things to comment about can include:

• Parts of the challenge you found unclear
• Problems that could make the challenge uninteresting or unfit for the site

You don't need any qualifications to review sandbox posts. The target audience of most of these challenges is code golfers like you, so anything you find unclear will probably be unclear to others.

If you think one of your posts requires more feedback, but it's been ignored, you can ask for feedback in The Nineteenth Byte. It's not only allowed, but highly recommended! Be patient and try not to nag people though, you might have to ask multiple times.

It is recommended to leave your posts in the sandbox for at least several days, and until it receives upvotes and any feedback has been addressed.

## Other

Search the sandbox / Browse your pending proposals

The sandbox works best if you sort posts by active.

To add an inline tag to a proposal, use shortcut link syntax with a prefix: [tag:king-of-the-hill]. To search for posts with a certain tag, include the name in quotes: "king-of-the-hill".

Get the Sandbox Viewer to view the sandbox more easily!

# Is the array sorted?

Inspired in part by Creative ways to determine of an array is sorted.

Given an array of integers, find out if the array is sorted. This challenge is as simple as it sounds. I'm surprised that I couldn't find a question like this on PPCG. I'm aware that a lot of golfing languages will have short solutions, but perhaps this question could allow you to showcase a language that's more esoteric than it is practical?

## Input

An array of integers, or a sequence of integers if your language doesn't support arrays.

## Output

Truthy if the array is sorted, falsey otherwise

## Examples

[1, 2, 3] => true
[3, 2, 1] => false
[1, 1, 1] => true
[] => true
[1] => true
[-3, -2, -1] => true
[-1, -2, -3] => false
[1, 2, 1] => false
[1, 1, 0] => false


Since this is , get ready to trim off some bytes! Happy golfing!

## For the sandbox

I have searched thoroughly for a question like this, but I haven't been able to find this question posted previously. If it exists, or if it is too similar to another question, please link the question or comment why this question should/shouldn't be posted.

• People don't post trivial challenges because it's not interesting to do in most languages. However having trivial challenges is not a too bad thing for esoteric languages like Brain-Flak. Dodos will still have great difficulty handling negative numbers. Dec 6, 2018 at 16:27
• Good point. Since this challenge will mainly be interesting in esoteric languages, I suggest restricting the input to positive integers. It won't make the challenge any different for languages with signed integer types, while allowing other languages to participate without implementing a new number type. Dec 7, 2018 at 13:46

# Compressed RSA keys

Silly cops-and-robbers idea.

As computers get more powerful, RSA keys have grown longer and longer to maintain security. This makes it difficult to:

• recite it over the telephone
• quickly scribble it down when you meet someone at a bar
• etc.

But we are all programmers (maybe even that someone that you met in a bar), so let's come up with RSA keys that can be generated by very short programs. We should be careful, though, that these keys remain resistant to cracking.

### Cops

Write a short program (or function) that outputs an RSA key. For this challenge, an RSA key is just any positive integer less than $$\2^{1024}\$$ that has a non-trivial divisor.

• take no input
• output the same number every time it's run
• finish quickly (a few seconds at most)

Post your program, its language, and what it outputs.

Also, prove that you can use it as an RSA key. Provide a SHA-256 hash of a string that lists all the prime factors of your key.

### Robbers

Crack the RSA keys! If an RSA key is the integer $$\n\$$; find any integer $$\1 that divides $$\n\$$.

### Scoring

A key is vulnerable to cracking for 7 days after it's posted. After that, it is considered safe and eligible for scoring.

Cops: The shortest uncracked program wins.

Robbers: Most cracks wins.

(Related: compress RSA public keys on crypto.SE)

• What's to prevent me from feeding random numbers into a 700-bit hash function until I get a semiprime, which I factor on huge amounts of cloud computing time and then post? No one can crack it unless they have a bigger computer. Dec 7, 2018 at 9:18
• Nothing, I guess. It might cost you a lot of money to do that, though.
– japh
Dec 7, 2018 at 9:37
• Also for what it's worth, not all semiprimes have large factors, so you'd have to factor many of them until you get a strong one. The robber only has to factor the one you posted.
– japh
Dec 7, 2018 at 9:39

Not a challenge, but not certain of the wording so trying it out here:

# Tips for golfing in JSLint

What general tips do you have for golfing in JavaScript restricted to satisfying JSLint with default options? I'm looking for ideas that can be applied to code golf problems in general that are at least somewhat specific to JSLint (e.g. "remove comments" is not an answer).

# Unjagged my array

Given a jagged array of integers, rearrange the integers such that

• The shape of the jagged array is unchanged, and

• The jaggedness of each subarray is minimized.

Here "jaggedness" is defined as the sum of the absolute differences in each subarray. If an array has only one element, then the jaggedness of that array must be 0, since there are no absolute differences to consider.

We only care about the differences in elements in the same subarray. So for example:

[[1, 2], [100]]


has a jaggedness of 1, since the first array has a jaggedness of $$\abs(1-2) = 1\$$ and the second array has a jaggedness of 0. We don't care that the difference between 2 and 100 is 98 since they are different subarrays.

Here's an example:

Input: [[16, 8], [15], [4, 42, 23]]
jaggedness of input: [abs(16 - 8)] + [0] + [abs(4 - 42) + abs(42 - 23)] =
8 + 38 + 19 = 65

One possible output: [[4, 8], [42], [15, 16, 23]]
jaggedness of this output: [abs(4 - 8)] + [0] + [abs(15 - 16) + abs(16 - 23)] = 4 + 1 + 7 = 12


Of course, this is one possible output. Any of these outputs would also have a jaggedness of 12:

[[4, 8], [42], [23, 16, 15]]
[[8, 4], [42], [15, 16, 23]]
[[8, 4], [42], [23, 16, 15]]


## Clarifications

• The input will always contain non-negative integers. The input will also always have a length >= 1, and all of the subarrays will have a length >= 1. This means that you must handle inputs as small as [[n]], where n is any non-negative integer.
• If the input is already arranged to be as smooth as possible, it is acceptable for your code to output the integers in a different order, as long as this new order has the same jaggedness. For example, if the input was

[[1, 2], [3, 4, 5]]


then your program may return the input unchanged, or it may return

[[5, 4], [1, 2, 3]]


or any of the other arrangements with jaggedness 3.

# Fastest code

This is a challenge, so the submission with the shortest run time for a hidden test case will win. I will use 2 different test cases with different shapes, and your score will be the sum of the time for each. (I haven't yet decided how large these test cases should be. I might adjust them once submissions come in so that it takes an appropriate amount of time.)

All submissions will be tested on my machine:

• Windows 10
• i7-4790k
• 8 GB RAM

You can use any language that is freely available on Windows 10. Please also provide instructions for running your submission. I will test each submission at my soonest convenience after it's posted, and then comment an unofficial score. After 1 week, I will test every submission, and post a leaderboard.

# Test Cases

These test cases were generated with and verified by a python brute-forcer.

Input: [[1, 2, 9], [10, 11]]
Output: [[9, 10, 11], [1, 2]] (jaggedness 3)

Input: [[4, 8, 15], [16], [23, 42]]
Output: [[4, 8, 15], [16], [23, 42]] (jaggedness 30)

Input: [[45, 98, 9, 46], [48, 53, 59], [13], [38, 94]]
Output: [[45, 46, 48, 53], [9, 13, 38], [59], [94, 98]] (jaggedness 41)


Note that none of these test cases show the only acceptable answer, merely one acceptable answer. For example, any output would remain valid if a subarray was reversed.

# Destroy the Computers with Magnets!

The robot apocalypse has finally arrived. But as we all know, robots are controlled by computers. And magnets are a convenient way of disabling computers.

Your task: place the magnet close enough to the computer so that it disables the computer.

A computer looks like this (always the exact same size):

 ____
|    |
|    |
|____|


The magnet looks like this:

o


The magnet needs to be touching the computer to do its work:

 o____o
o|    |o
o|    |o
o|____|o
oooooo


Note the lack of placements on the top; that is intentional. People use the computer on that side, and would easily see the magnet and remove it.

So if you are given a picture of some computers:

       ____
|    |
____ |    |
|    ||____|
|    |      ____
|____|     |    |
|    |
|____|


You should place 1 magnet for computer next to the computers:

       ____
|    |
____ |    |
|    ||____|
|    |  o   ____
|____|     |    |
o        o|    |
|____|


Magnets may touch computers, but there must be 1 touching magnet per computer.

Due to the apocalypse, resources are scarce, so your program to place magnets must be as short as possible.

Test cases:

Input
~~~~~
Example outputs (not only possible outputs)
=====
____
|    |
____ |    |
|    ||____|
|    |      ____
|____|     |    |
|    |
|____|
~~~~~
____
|    |
____ |    |
|    ||____|
|    |  o   ____
|____|     |    |
o        o|    |
|____|

____
|    |
____ |    |o
|    ||____|
|    |     o____
o|____|     |    |
|    |
|____|
=====
More soon....

• Is it permissible for a "component" to be adjacent to multiple voting machines? For example, with two voting boxes 1 space apart, would placing two components in that column be valid? Dec 15, 2017 at 20:03
• @KamilDrakari That is allowed, I will edit. Dec 16, 2017 at 0:27
• Would these be valid outputs? Also is the voting machine always the same size? Dec 16, 2017 at 20:24
• @BruceForte Those are all valid outputs except for number 4: there must be at least 1 physical component per voting machine (as it says in the challenge.) Voting machines are all the same size. I will edit. Dec 18, 2017 at 14:24

# Minimal Substring Lookup Table

A fairly common component of code golf challenges is taking input as a string, and finding the index of that string within an array. A common way to golf that action is to instead store a substring of each item in the array, and take the same substring of the input to use when looking up the index. For example, this question has some fairly large strings that need to be indexed, but this answer found that each item could be uniquely identified by its 4th letter.

### Challenge

Find the shortest substring that, when applied to each string in the input, results in a new set of unique strings.

### Input

• Input is list of strings in any default format
• The strings will consist of printable ASCII characters.
• The strings will not have trailing spaces
• There will be no duplicate strings

### Output

• Output should be a list of the shortened strings after applying the substring
• If outputting indexes, 0-indexing or 1-indexing may be used.
• When strings have different lengths, substrings exceeding the length of some lines may be handled any of these ways:
• Truncate to the available characters, e.g. ["abc","acd","ab"] => ["bc","cd","b"]
• Pad shorter strings with spaces, e.g. ["abc","acd","ab"] => ["bc","cd","b "]. Note that input strings can contain spaces which need to be taken into account, so ["ab c","acdc","ab"] must not become ["b ","cd","b "].
• Wrap back to the start of the shorter string, e.g. ["abc","acd","ab"] => ["bc","cd","ba"]. Note that even with this method it is not necessary to look for substrings where every string wraps; the starting index should always be within the length of the longest input string.

Sandbox questions

• Should I restrict the input strings to only a subset of printable ASCII?
• Should I enforce one specific behavior for handling shorter strings? Requiring wrapping would let the challenge look for even shorter solutions requiring full wrapping, e.g. ["aba","ab"] => ["a","b"], but I don't want languages with different default substring implementations to need a lot of work just to get the wrapping working.
• I used the term "substring" as something that is "applied" to a string to get the shortened version, but I know that's extremely wrong. Is there a better way to phrase this?
• This challenge is related, but uses arbitrary substrings rather than aligned substrings. Should I mention it somewhere?
• "Should I enforce one specific behavior for handling shorter strings?" - Personally, I'd say yes. Dec 20, 2018 at 19:02
• I think it would be better to output the position and length of the substring, rather than every substring. Dec 30, 2018 at 1:24
• Alternatively, you could just say that all input strings have the same length.
– japh
Jan 16, 2019 at 8:43

## Piano Scale Fingering

There are 12 different major scales on the piano, which have similar, but not identical, fingering. Furthermore, the fingering for each hand is slightly different.

The fingers of each hand are numbererd starting at 1 for the thumb. Each note usually takes the next finger in turn. For the left hand playing an ascending scale, this means that the numbers decrease, while for the right hand the numbers increase.

Of course you will run out of fingers before you get to the end of the scale. For the left hand, this means that when you get to the thumb you need to start again with one of the other fingers, while for the right hand this means that when you get to a certain finger you need to start again with the thumb.

Although there are eight notes in an octave, the part that repeats in the middle of a scale is seven, which is a prime number. Scales are therefore normally played using the repeatying sequnces 4321321 in the left hand and 1231234 in the right hand; the little finger is only used to avoid switching to the thumb on the first or last note respectively. Other adjustments also apply at the very start and end of the scale.

There are a couple of rules which work for most scales. The first rule is that you try to play the tonic with the thumb except at the appropriate end of the scale when you use the little finger. If you can't follow that rule then with the right hand you use the thumb when switching from a black note to a white note; with the left hand you aim for the thumb to land on a white note just before a black note.

Here is a full list of all 12 keys and their fingerings:

 #  Key Left Hand       Right Hand
-6  Gb  432132143213212 234123423412345
-5  Db  321432132143212 231234123123412
-4  Ab  321432132143212 231231234123123
-3  Eb  321432132143212 212341231234123
-2  Bb  321432132143212 212312341231234
-1  F   543213214321321 123412341231234
0  C   543213214321321 123123412312345
1  G   543213214321321 123123412312345
2  D   543213214321321 123123412312345
3  A   543213214321321 123123412312345
4  E   543213214321321 123123412312345
5  B   432143213214321 123123412312345
6  F#  432132143213212 234123123412345


Please write a program function which takes as input the key (or #, which is the number of sharps in the key, or the negation of the number of flats in the key) of a scale and a flag (e.g. L or R) for which hand, and output the fingering for two octaves as above. (For longer scales you simply repeat the middle 7 digits so I don't see the utility of asking you to output an arbitrary number of octaves.)

Note that Gb is enharmonic to F# on the piano but you might find it easier to support one rather than the other. Please indicate which one your answer supports.

As an alternative to writing a single function I will also allow two named functions (one for each hand) but not two lambdas. The 1 byte per extra file penalty applies if you submit two full programs.

• Why not two lambdas? Dec 24, 2018 at 9:42
• @wizzwizz4 You're not allowed to encode the answer in your choice of boolean flags, and IMHO allowing lambdas would violate that loophole, but for named functions the answer isn't encoded in the name, so that's OK. (You could of course just pay the penalty for naming your lambdas.)
– Neil
Dec 24, 2018 at 10:08
• Some languages don't feature named lambdas, e.g. a LISP dialect without defvar. It seems completely arbitrary. Dec 24, 2018 at 10:10
• What's wrong with lambdas? That rule is completely arbitrary. Dec 28, 2018 at 19:29

# Front lines go front! code-golfarray-manipulation

Given two lists of positive integers, bring the elements in the first one that are in the second to the far left, keeping their order the same as in the first list. You can assume that the second list's elements will all be unique.

These approaches are forbidden.

For example, let's call the first list $$\A\$$ and the second $$\B\$$. Here are some test cases:

$$\A=[1,3,4,1,2,1,4,3,2,1,3,2,4,1],B=[1,3]\rightarrow[1,3,1,1,3,1,3,1,4,2,4,2,2,4]\$$

$$\A=[],B=[1,4,5]\rightarrow[]\$$

$$\A=[1,2,3,4,5,6,7,8,10],B=[]\rightarrow[1,2,3,4,5,6,7,8,10]\$$

$$\A=[],B=[]\rightarrow[]\$$

$$\A=[1,2,2,1,3,4,4],B=[1,2]\rightarrow[1,2,2,1,3,4,4]\$$

$$\A=[3,4,4,1,2,1,5],B=[1,5,2]\rightarrow[1,2,1,5,3,4,4]\$$

$$\A=[1,2,3,2,1,2,3,2,1],B=[4,5,6]\rightarrow[1,2,3,2,1,2,3,2,1]\$$

$$\A=[2,2,1,2,1,3,3,1,4],B=[1,2,3,4]\rightarrow[2,2,1,2,1,3,3,1,4]\$$

Note: the integers aren't necessarily single-digit.

• @JonathanFrech Yes. How could it be used anyway? Dec 27, 2018 at 11:42
• @JonathanFrech Clarified. Dec 27, 2018 at 15:52

### Do not not not not make a filter that doesn't not not not get rid of these words!

Tags: code-golf

Your program has to scan each run of "not"s in a sentence, and if the number of "not"s is odd, then replace it with only one "not". If it is even, replace it with the empty string.

Notes:

• The "not"s are case-insensitive, so your input may be "not" or "Not" or more.
• The input may only be taken in as a string.

Test cases:

 Don't not not -->  Don't
not not not a not not  -->  not a
NoT nOt not NOT noT a Not nOt noT  -->  not a not
hello notnot anot not anota not not not --> hello a anota not

• We've had a very similar challenge before. On my phone so can't search properly for it to determine if it would be similar enough to be a dupe target. Dec 21, 2018 at 19:27
• Oh, let me see that then. Dec 21, 2018 at 19:40
• So, input and output as a string? Would list of words be acceptable input/output? Dec 23, 2018 at 14:20
• @BMO List of words is not a valid format Dec 25, 2018 at 23:28
• So, should I make this a question? Dec 25, 2018 at 23:30
• What exactly counts as a not? Can it be part of another word? Do the nots have to be separated from each other by a single space? Dec 30, 2018 at 1:10
• @12Me21 it can be part of another word Dec 30, 2018 at 3:07

In radiation hardening challenges, the goal is to write a program whose behavior doesn't change when an arbitrary character is removed from its source. This KotH explores an adversarial variant on this challenge in which two programs compete to survive the most character deletions. The twist is that each program is provided the source code of its opponent and outputs which character should be deleted from its opponent's source code next.

## Gameplay

Suppose program A and B are competitors.

1. Each program is run and the output is recorded. Let the output of program A be a and the output of program B be b. If one program fails to output a valid character index of its opponent's source code, the other program will be declared the winner of the match. If neither program outputs a valid index, then a tie is declared.
2. The a'th character is removed from program B and the b'th character is removed from program A.
3. Repeat steps 1 and 2 until there is a winner or a tie.

## Submissions

Submissions should be a complete javascript or python program which:

• Begins with a shebang (#!) indicating which language it is written. For python entries, this should be #!/usr/bin/python. For javascript entries this should be #!/app/.heroku/node/bin/node.
• Takes three command line arguments: the code for the opponent program, the start of the range of deletable characters (x), and the end of the range (y). In python these can be accessed using sys.argv, while in javascript they can be found in process.argv.
• Prints a number in the range [x,y) to stdout and terminates in less than 10 seconds.

Note: The index of the first deletable character of your opponent's program (x) is the index of the first character after the shebang. y is the length of the opponent program in characters (so y-1 is the last deletable character).

## Sample Submissions

Python:

#!/usr/bin/python
import sys

# vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
print (sys.argv[3]-1)
# vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv


Javascript:

#!/app/.heroku/node/bin/node
// vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
console.log(Math.round(Math.random() * process.argv[4]));
// vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv


## Controller

I've hosted a web based controller at https://radiationshowdownkoth.herokuapp.com/. It should conform to the behavior specified in the Gameplay section. Let me know if you find any bugs.

## Rules

• Standard loopholes apply.
• The use of preexisting interpreters and their components (parsers, lexers, etc) is banned. This includes, but is not limited to, calling the python and node binaries from your program (eg os.popen("python -c ...")), using builtin eval functions (eg python's exec), and using relevant modules (such as python's astlib). However, you may use regular expressions.
• No malicious code.

## Logistics

It will be a Round Robin Tournament. Each submission will be matched against every other submission. For each victory a submission achieves, it will be awarded 1 point. Likewise, ties are worth 0.5 points and losses are worth 0 points. The submission with the most points after all rounds have been completed wins this KotH. The winner will receive a bounty from yours truly, and the most popular entry will be awarded accepted answer. This KotH will be open to entries for a week following its posting, at which point the tournament will be run and a winner declared.

## Meta

• Duplicate?
• Clarifications?
• Other Language Suggestions besides Python & Javascript?
• Are there any trivial strategies that would be overwhelmingly effective?
• Any other foreseeable problems?
• Would an empty program that outputs 0 auto win? Dec 30, 2018 at 3:31
• Also, removing the shebang would kill the program. Dec 30, 2018 at 3:31
• @user202729 Yes, I believe an empty program that outputs 0 would always win as the rules are currently written. Is it possible to write such a program in Javascript or Python? Dec 30, 2018 at 3:35
• @user202729 Actually, I am mistaken. If a program outputs 0 it will lose because 0 is not in the valid range of deletable characters (passed in as the second and third arguments to the program). This range starts after the end of the shebang in order to prevent the second exploit you pointed out. I will try to clarify this in the spec. However, a 0 character program that printed its second command line argument would auto win. As far as I know, it is not possible to write such a program in Javascript or Python, but please let me know how if it is. Thanks for the exploits :). Dec 30, 2018 at 3:53
• How are x and y computed? Dec 30, 2018 at 3:54
• @user202729 x is the index of the first character after the shebang. y is the length of the program in bytes. I've edited the spec to include this information. Dec 30, 2018 at 4:07
• I like it! I'm not very experienced, but I'd say it's ready to go. Dec 30, 2018 at 4:10
• Hm... in that case I'd say that because Python and Javascript are both "sensitive" (?) to radiation, making a program that would never lose is pretty easy. (just remove the first non-comment character or similar) Dec 30, 2018 at 4:12
• @user202729 One way to defend yourself against a program that deletes the first non-comment character would be to define a variable with a really long name on the first non-comment line of the program. But I see your point. Any ideas for better choices of languages? Dec 30, 2018 at 4:28
• It might be better to make the challenge "given a Python/JS program, delete a character so that it throws an error." Because AFAIK it's always possible, and such a program will never lose. Jan 1, 2019 at 14:59

# Unfactor a list code-golfjellymathnumberprimes

Jelly has an interesting built-in, ÆẸ. It's the inverse of ÆE. ÆE returns the exponents of the prime factors of an integer n, where each exponent corresponds to the prime number at its index, so, for example, [3, 5, 1] represents $$\2^3\cdot3^5\cdot5^1\$$. If a prime isn't included in the prime factorization of n, but there is at least one prime after it that is included, a zero is put in its place ($$\p^0=1\$$). So, for example, [3, 0, 5, 1] represents $$\2^3\cdot5^5\cdot7^1\$$, and [0, 4, 4, 50, 0, 7] represents $$\3^4\cdot5^4\cdot7^{50}\cdot13^7\$$.

Your job is to implement ÆẸ, ÆE's inverse. That is, you'll be given a list returned by ÆE, and your job is to find n.

You must always return a positive integer (strictly greater than zero). You may assume the input doesn't have negative or float exponents or trailing zeroes.

Note that you're encouraged not to use ÆẸ or your language's equivalent (if any) in your answer.

Using any of the standard loopholes is prohibited.

Test cases (you don't have to support exponents, an output or an input's length beyond your natural signed integer type's limit):

[] -> 1
[0, 4] -> 81
[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1] -> 1111
[10] -> 1024
[3, 5, 1] -> 9720
[3, 0, 5, 1] -> 175000
[31] -> 2147483648
[32] -> 4294967296
[63] -> 9223372036854775808
[64] -> 18446744073709551616
[0, 4, 4, 50, 0, 7] -> 5713082599062385095715588395684396863698753591853983125


Make more test cases:

To generate the input that corresponds to a given output, use this program.

To generate the output that corresponds to a given input, use this program.

• @Adám Hm, I included the stuff for clarity, but I can remove it as well. EDIT: Removed the fluff. Jan 7, 2019 at 22:49

# Where can I get to from each location? (Transitive Closure)

Given an undirected graph (network) $$\G\$$, construct a new graph in which vertex (node) pair $$\(u,v)\$$ is an edge (are connected) if and only if a path $$\(u,w_1,w_2,...,w_k,v)\$$ exists in $$\G\$$ for some $$\k\ge 0\$$. This is known as $$\G\$$'s transitive closure. If this is clear to you, you're ready to get started. Otherwise, just read through the below sections. It is actually a very simple problem.

## I/O formats

Take for example the graph (network)

1──2  3──4
│     │
5  6──7──8


We can represent¹ it as an adjacency (connection) matrix:

 │1 2 3 4 5 6 7 8
─┼───────────────
1│  ┘
2│
3│      ┘
4│
5│┘
6│            ┘
7│    ┘         ┘
8│


which is:

[[0,1,0,0,0,0,0,0],[0,0,0,0,0,0,0,0],[0,0,0,1,0,0,0,0],[0,0,0,0,0,0,0,0],[1,0,0,0,0,0,0,0],[0,0,0,0,0,0,1,0],[0,0,0,1,0,0,0,1],[0,0,0,0,0,0,0,0]]


Note that here only one of $$\(u,v),(v,u)\$$ is represented, but we consider both directions as valid edges (connections).

Or as a list of of edges²:

[[1,2],[1,5],[3,4],[5,1],[6,7],[7,3],[7,8]]


Or for each vertex³ (node), the list of its adjacent vertices:

[[2,5],[1],[7,4],[1],[7],[3,8],[8],[]]


Or as a dictionary⁴:

{"1":[2,5],"3":[7,4],"7":[6,3,8]}


Any of these, and any other reasonable input and output formats (you optionally may use one format for input and another for output) are allowed, but you must state what your formats are. However, it is required that your formats support under-representing (e.g. [3,7] but not [7,3]) and over-representing (e.g. both [3,7] and [7,3]).

## Walk-through

Let's use the representation [[1,2],[1,5],[3,4],[5,1],[6,7],[7,3],[7,8]]. Since 1 is connected to 2 then 2 is also connected to 1, so we add (it doesn't matter where) this edge (connection):

[[1,2],[1,5],[2,1],[3,4],[5,1],[6,7],[7,3],[7,8]]

It is also possible to travel [1,5] in reverse, but that pair is already represented further in the list. Now note that it is possible to find a path from 2 to 5, and vice versa, via 1, so we add these two edges:

[[1,2],[1,5],[2,1],[2,5],[5,2],[3,4],[5,1],[6,7],[7,3],[7,8]]

This completes the left side of the graph. Similarly, we process the right side by adding the reversals of [3,4], [6,7], and [7,3]:

[[1,2],[1,5],[2,1],[2,5],[5,2],[3,4],[5,1],[6,7],[7,3],[7,8],[4,3],[7,6],[7,3]]

Two-step paths via 3 are possible, so we add [7,4] and [4,7]. Similarly, two-step paths via 7 are [6,3], [3,6], [6,8], [8,6], [3,8], and [8,3]:

[[1,2],[1,5],[2,1],[2,5],[5,2],[3,4],[5,1],[6,7],[7,3],[7,8],[4,3],[7,6],[7,3],[7,4],[4,7],[6,3],[3,6],[6,8],[8,6],[3,8],[8,3]]

Finally, we add the three-step paths [6,4], [4,6], [4,8], and [8,4]:

[[1,2],[1,5],[2,1],[2,5],[5,2],[3,4],[5,1],[6,7],[7,3],[7,8],[4,3],[7,6],[7,3],[7,4],[4,7],[6,3],[3,6],[6,8],[8,6],[3,8],[8,3],[6,4],[4,6],[4,8],[8,4]]

And this is our answer. It could of course be in any order.

1. Though the connections do not have direction, I've only put in one entry in the table for each connection, and obviously the diagonal is all-true too, as every node is reachable from itself.
2. Since the connections in this challenge do not have a direction, [1,5] and [5,1] are the same connection. This serves to illustrate that such may occur in the given data.
3. Here, each node must have its own list, as the lists are paired to their points by their position in the data. However, each list need not be exhaustive as long as all connections are represented somewhere.
4. Here, we can omit entries that are fully covered by the other entries.
• Am I over complicating things by allowing all the various I/Os?
Jan 7, 2019 at 13:26
• All the input formats you mentioned seem reasonable to me and I'm sure different languages could use different formats when golfing. And if you restrict it to one specific format, you'll get people asking about using other formats anyways. Jan 7, 2019 at 14:27
• why not state it in graph-theoretic terms - vertex, edge, path, graph, incidence matrix - instead of the more vague location/place/point, connection, journey, network, connection table? "[1,5] and [5,1] are of course the same connection" - it would be just as valid and interesting a problem if it were about directed graphs, so it's not "of course" :) tag "repeated-transformation" is unnecessary - this can be solved in O(n^3) time without the use of ⍣≡.
– ngn
Jan 7, 2019 at 14:27
• @Emigna Fixed. Thanks.
Jan 8, 2019 at 7:25
• @ngn Reason: I am not educated enough. Can I bother you to edit those in? I'll remove "of course".
Jan 8, 2019 at 7:26
• @Adám sure, i'll edit it. it's nice you put "transitive closure" in the title. many challenge authors (with or without cs/maths degrees) like to reinvent terminology, obscure the problem with backstories, etc. i find that mildly annoying. example - seems to be an interesting challenge but it takes so long to figure out what the words mean.
– ngn
Jan 8, 2019 at 13:43
• @Adám how does it sound now? i'm not sure if i didn't go to far with that pseudo-academic lingo...
– ngn
Jan 8, 2019 at 14:21
• @ngn Thank you so much. I've added in parenthesised layman's terms upon first usage in each section, and added a bit to the intro to keep people from fleeing. I also inserted the word "pair" on the very first line before (u,v) as it seemed to me that it was missing. Correct me if I'm wrong.
Jan 8, 2019 at 15:05
• @Adám i forgot to delete the word "vertex". inserting "pair" should be fine too
– ngn
Jan 8, 2019 at 15:12
• Jan 9, 2019 at 11:40

# Introduction

This is my first entry here, so please be gentle if it's poorly stated. I'll need a bit of guidance to make it a good submission.

I am a typewriter maker, and I want to prove that the design of my competitor, Mr. Crown, is very inconvenient to use. To do that, I want a list of the most cumbersome words to type out on a Crown typewriter.

# Challenge

Write a program that takes an argument "n" as a positive integer. You may load the Unix words list from any source (ignore case; discard any word containing anything outside of A-Z, such as "O'Brien", "L'vov", etc.). Given the layout shown below, output the "n" most cumbersome words to write, one result per line. Each result is the cumbersome word, some form of separator (eg. a space), then their word score (see below) with two decimals of precision (and your choice of decimal separator). Bonus if results are sorted in descending order by their word score, and further bonus if ties (by score) are also sorted in alphabetic order.

X Q K G B P M C O F L A N D T H E R I S U W Y J V Z

# Word scoring

Assume the pointer starts in the leftmost position, on "X". Count each movement and each hammer strike as one unit, but divide the sum by the number of characters in the word. Hence, sample word scores are:

• THEIR: 4.8 -- 24 units (14 movements, strike "T", movement, strike "H", movement, strike "E", 2 movements, strike "I", movement, strike "R") divided by 5 characters
• LAND: 4.25 -- 17 units (10 movements, strike "L", movement, strike "A", movement, strike "N", movement, strike "D") divided by 4 characters
• FOWL: 9.50 -- 38 units (9 movements, strike "F", movement, strike "O", 13 movements, strike "W", 11 movements, strike "L") divided by 4 characters
• BARBARA : 8.00 -- 56 units (4 movements, strike "B", 7 movements, strike "A", 6 movements, strike "R", 13 movements, strike "B", 7 movements, strike "A", 6 movements, strike "R", 6 movements, strike "A") divided by 7 characters
• X: 1 -- just hit the hammer once and you're done

Note: These examples are not the most cumbersome words.

Input:

5

Standard output:

ANNA 4.25

THEIR 4.80

LAND 4.25

BARBARA 8.00

FOWL 9.50

Bonus output:

FOWL 9.50

BARBARA 8.00

THEIR 4.80

ANNA 4.25

LAND 4.25

# Scoring submissions and determining the winner

I'm really not sure what the best measure of success is. Definitely not code golf, because I can never distinguish the damn things from line noise or pure black magic. But what then? I'm not very familiar with which competition types are available.

If two solutions are tied for a position, but only one of them uses the bonus output format, then that one wins the tie.

• Tak for at du bruger sandkassen!
Jan 4, 2019 at 13:19
• You can find a list of all winning criteria here, though I think code-golf is actually the best fit. When you are bothered by illegible answers, try asking for an explanation, most users are happy to explain how their code works and which tricks they used. Jan 6, 2019 at 8:32
• As for the challenge, it would be helpful to provide a transcript of the order of the letters from the image. I personally like challenges to focus on their core, in this case calculating the score of a word, and to avoid additional tasks, like sorting a list and printing the top n entries, as those can distract from the main challenge. (E.g. in a language without a build-in sort method the challenging part could become implementing a sorting algorithm even though that's presumably not the intention of your challenge.) Jan 6, 2019 at 8:51
• @Laikoni thank you for the reference! I am leaning towards [popularity-contest], but otherwise I guess I can settle for [code-golf]. Jan 8, 2019 at 9:11
• popularity-contest challenges are very hard to get right, read e.g. here. I'd strongly advise you against using it, as it is not a good fit for this otherwise nice challenge. Jan 8, 2019 at 13:28
• Is the word list common for all solutions or may the answerer choose their word list? Jan 10, 2019 at 1:28
• I can restrict it to the Unix Words file if that helps. Jan 10, 2019 at 9:31
• If not a [popularity-contest], then what's the best way to score this, given that I desire code readability (so not [code-golf])? Jan 10, 2019 at 9:32
• Imo, the heart of the challenge is to determine the score of a word: I consider the rest as unnecessary, reading words from files has been done and will only make it difficult (or even impossible) for some languages, sorting with a custom function too, has been done. So I would extract the heart of this challenge and state it as Given a word, output its score. Jan 10, 2019 at 14:01
• Now, since you don't want a code-golf, popularity-contest is not really the right thing, the challenge is not computationally hard (ruling out fastest-code or the like) that kind of leaves us with: atomic-code-golf or code-challenge. For the former one I don't see a sensible way to define the challenge which leaves us with the latter and gives rise to a start of a really interesting challenge: Make it a code-challenge where the scoring is according to the word's score ("cumbersome-ness"). The problem will be characters outside of that range: Jan 10, 2019 at 14:08
• Ideas (for all I'm assuming case-insensitivity): 1) You extend the range to include @!…⅓ and exchange the character _ or the unreadable one after Z with a space. Then you'll only allow submissions using this extended range. [imo the most sensible, it does restrict the submissions heavily but in an interesting and sensible way so] 2) Before scoring the submission, remove all characters outside of the range. [people will abuse this and only use characters outside of that range] 3) Only XQ…Z are valid in a submission. [this is a really tight restriction for most languages] Jan 10, 2019 at 14:14
• 4) Instead of ignoring characters outside that range, penalize them with a constant (kind of expensive, but not too expensive). Jan 10, 2019 at 14:21
• @BMO, thanks a ton for your feedback! I'm not sure I follow (= I'm pretty sure I don't follow) your "ideas" comment. Could you explain further? Jan 10, 2019 at 15:28
• If you go the route of code-challenge with score being the "cumbersomeness" of the code itself, it's not clear how you'll treat characters which aren't in the range XQ…Z and there would be need of a way to score these. The ideas (1-4) would be different options to define that. Jan 10, 2019 at 16:17

# My Number is Bigger Than Yours: KoTH

The concept of this game is simple; each turn, two bots submit a positive integer (in Python, so no upper bound on its size). Whichever bot submits a larger number wins a certain number of points based on the scoring put forward in the next section. This process is iterated a certain number (100?) times to form the competition, with the bot which has the highest score winning the competition between the two bots.

However, scoring is based on some metric of the difference between the submitted values such that the larger the difference is, the smaller the amount of score gained is. I'm trying to decide some metric which heavily incentivizes making a guess which narrowly beats your opponent's guess, maybe with a metric such as 1/(a-b)**2. However, as the numbers will likely grow throughout the iterations, maybe 1/(a/b)**2 might be better.

Example bot whose goal is to beat opponents who pick a constant number:

def constant_beater(my_score, opp_score, my_numbers, opp_numbers):
if len(opp_numbers) == 0:
return 1
return opp_numbers[-1] + 1


The largest concern I see with this challenge is a bot which just increments much quicker than all other bots so that it wins every game with a minuscule amount every time. Maybe to curb this a score that is below a certain threshold for a game is rounded to zero to encourage more competitive choices.

• I feel like this would be a much better challenge if bounded a bit, from maybe 1-100 for example Jan 11, 2019 at 22:57
• I was thinking of some sort of bounding, but I'm struggling to then find an incentive for a bot to choose anything except for the upper bound. Jan 11, 2019 at 23:02
• Make it a round robin tournament? Jan 11, 2019 at 23:19
• Numbers from 1 to 100. 100 rounds between 2 bots, higher number wins, can only play each of the 100 numbers 1 time each. Score is the number of rounds won? The play method doesn't work well for difference between numbers. Jan 13, 2019 at 1:43
• This sounds extremely abusable, even more if they try to tap into the others' functions. However, it still sounds real fun, reminds me of the Robot Roulette but with much better winning chances. Maybe you could add some restriction like preventing to use the same number X amount of rounds with a maximum number cap (like @fəˈnɛtɪk said but without too much limiting) Jan 15, 2019 at 22:38
• Maybe instead of a maximum number cap a maximum sum cap? Jan 16, 2019 at 21:44
• If there's a maximum number cap I don't think there needs to be a maximum sum cap. However, there needs to be an incentive so bots don't go spamming the maximum number 24/7 (or worse, having 2 bots spam max number to infinity and ending in a 0 point tie) Jan 17, 2019 at 23:23
• Thinking about this: Perhaps the max is 100, and bots score 100-guess if they win, and -(guess/100) if they lose? Jan 31, 2019 at 22:20

# Drawing $$\K_n\$$ (complete graph with $$\n\$$ vertices)

In graph theory the complete graph with $$\n\$$ vertices, often written as $$\K_n\$$, is the graph where every vertex is connected to every other vertex. For example, $$\K_3\$$ with $$\V = \{A,B,C\}\$$ has the edges $$\\bigr\{\{A,B\},\{B,C\},\{C,A\}\bigl\}\$$ - forming a triangle.

## Challenge

Given an even $$\n \geq 4\$$ you will generate/display an image of the graph $$\K_n\$$ as follows:

• pick an image width $$\\texttt{width} \geq 100\$$
• pick two distinct colors $$\\texttt{vertex_color}\$$ and $$\\texttt{edge_color}\$$
• draw $$\1\$$ vertex with color $$\\texttt{vertex_color}\$$ in the middle of the image
• draw the remaining $$\n-1\$$ vertices evenly distributed on a circle around it
• connect every vertex with color $$\\texttt{edge_color}\$$

## Rules

• the background color doesn't matter (transparent is fine), but it must not coincide with the colors $$\\texttt{vertex_color}\$$ and $$\\texttt{edge_color}\$$
• the radius of the circle should be $$\\texttt{width/4}\$$
• the center of the circle should the middle of the image
• the shape of the vertices doesn't matter, but they must be at least $$\3\$$ pixels wide
• the edges must not overlap the vertices
• the edges may be dashed, dotted etc.

## Examples

### Example output for $$\K_{32}\$$:

• "the edges must not overlap the vertices" is incompatible with "draw 1 vertex ... in the middle of the image" when n is odd. Jan 18, 2019 at 16:21
• @PeterTaylor: Good call, thank you! Restricted the input to only even integers. Jan 18, 2019 at 16:48
• a bit too many rules for me but... still very interesting and fun Jan 31, 2019 at 2:58
• @donbright: What do you mean by too many rules? Since it's graphical output, I have to set rules otherwise it'll become an art contest unfortunately. I'm open to less restrictive suggestions though. Jan 31, 2019 at 14:46
• its just my personal taste. like i dont care how wide the image is, how wide vertices are or whether edges overlap, etc etc. i upvoted anyways. good luck. Feb 2, 2019 at 2:55

# Chain round the number!

A decimal place is a digit of a number's location in a number. In the number 987654321, each digit corresponds to its decimal place. For example, in the number 16111, the 6 is in the 4th decimal place.

Normally when rounding to n decimal places, one checks the digit at n-1 (heretofore called x). If x >= 5, then the digit at n is rounded up. Otherwise the digit at n is rounded down (stays the same). Then, all digits including and after x are turned into 0.

For example, take the number 158, rounded to the 3rd decimal place. The digit at the 2nd decimal place (3rd decimal place minus one) is 5. Because of this, the 1 at the third decimal place gets rounded up. This results in the number being 258. However, the 5 and the 8 both get turned into 0, and so the final result is 200.

Chain rounding is like normal rounding, except that you do it to every digit starting with the digit at decimal place 2. Once you round to the decimal place 2, then you round to decimal place 3, and so on until decimal place n.

For example:

Input: 24472
24472    (2 < 5, so the 7 rounds down)
24470    (7 >= 5, so the 4 rounds up)
24500    (5 >= 5, so the 4 rounds up)
25000    (5 >= 5, so the 2 rounds up)


The challenge here is to chain round a number to the highest decimal place possible for that number.

Test cases: (all inputs will be positive whole numbers)

24472 -> 30000
1999 -> 2000    ("1 9 9 9" -> "1 9 10 0" -> "1 10 0 0" -> "2 0 0 0")
9945 -> 10000
2001 -> 2000
9444 -> 9000
9445 -> 10000
13579 -> 10000
24681 -> 30000
1337 -> 1000
5 -> 10
1 -> 1

• Good challenge, but I think this needs more explanation on what is happening. It's probably a good idea to fix an example with digits $ab$ and explain how the new rounding works (ie. compare $b$ to $5$ and so on), at first I did not understand where the $5$ comes from. You should also address how carries are treated, it's not clear atm. Jan 22, 2019 at 19:17
• @BMO I tried to clarify, is that better? (Thank you for the help). Jan 22, 2019 at 19:56
• Yes, looks good to me. Make sure to make a test-case with carries, eg. when rounding $9996$ or the like. And also specify whether numbers can be zero or even negative. Jan 22, 2019 at 20:15

# A Turtle Finds a Portal

The turtle wants to move along the grid to get to his food. He wants to know how many moves it will take for him to get there.

As well since he is slow he has teleporters set up around his domain that he will utilize if it shortens his path. Or avoid them if it lengthens his path.

## Meet the turtle

🐢

The turtle lives on a grid $$\begin{matrix} X&X&X&X&X\\ X&X&X&X&X\\ X&X&🐢&X&X\\ X&X&X&X&X\\ X&X&X&X&X\\ \end{matrix}$$ The turtle can move to any adjacent square... $$\begin{matrix} X&X&X&X&X\\ X&\nwarrow&\uparrow&\nearrow&X\\ X&\leftarrow&🐢&\rightarrow&X\\ X&\swarrow&\downarrow&\searrow&X\\ X&X&X&X&X\\ \end{matrix}$$

However, the turtle cannot move to a square with a mountain $$\begin{matrix} X&🌄&X&X&X\\ X&\nwarrow&\uparrow&\nearrow&X\\ X&🌄&🐢&\rightarrow&X\\ X&🌄&\downarrow&\searrow&X\\ X&🌄&X&X&X\\ \end{matrix}$$

The turtle wants to eat his straw berry, and would like to know how long it will take to get to his strawberry $$\begin{matrix} X&🌄&🍓\\ 🐢&🌄&X\\ X&🌄&X\\ X&X&X\\ \end{matrix}$$ This example would take the turtle $$\5\$$ turns $$\begin{matrix} X&🌄&🍓\\ \downarrow&🌄&\uparrow\\ \searrow&🌄&\uparrow\\ X&\nearrow&X\\ \end{matrix}$$ To get around mountains the turtle uses his teleporter. There are two teleports on the grid that map to each other. Stepping on the teleporter immediately moves the turtle to the corresponding teleporter. $$\begin{matrix} 🔵&🌄&🍓\\ 🐢&🌄&🔴\\ X&🌄&X\\ X&X&X\\ \end{matrix}$$ It is now faster for the turtle to move up twice. Now the turtles shortest path is $$\2\$$ $$\begin{matrix} 🔵&🌄&🐢\\ \uparrow&🌄&🔴\uparrow\\ X&🌄&X\\ X&X&X\\ \end{matrix}$$

## The challenge

Given an initial grid configuration output the number of moves it will take the turtle to reach his strawberry.

## Rules

• You may assume that the input grid has a solution

• The input grid may be entered in any convenient format

• The shortest path does not need to make use of the portal

• The turtle cannot pass into mountain tiles

• The turn that the turtle moves onto a teleporter square he is already on the corresponding teleporter. He never moves onto a teleporter and stays there for a move

• You may use any ASCII character to represent mountains, turtle, empty gird square, strawberry.

• You may use either the same character or two different characters to represent the teleporter pairs

• A grid can have more than one path with the same shortest path length

• Each grid will only have one strawberry and two portals

• This is .

## Test Cases formatted as lists

[ ['T', 'X', 'X', 'S', 'X'], ['X', 'X', 'X', 'X', 'X'], ['X', 'X', 'X', 'X', 'X'] ] --> 3
[ ['T', 'M', 'X', 'S', 'X'], ['X', 'M', 'X', 'X', 'X'], ['X', 'X', 'X', 'X', 'X'] ] --> 4
[ ['T', 'M', 'X', 'S', 'O'], ['O', 'M', 'X', 'X', 'X'], ['X', 'X', 'X', 'X', 'X'] ] --> 2
[ ['T', 'M', 'X', 'S', 'X'], ['O', 'M', 'X', 'X', 'X'], ['O', 'X', 'X', 'X', 'X'] ] --> 4
[ ['T', 'M', 'S', 'X', 'O'], ['X', 'M', 'M', 'M', 'M'], ['X', 'X', 'X', 'X', 'O'] ] --> 7



## Test Cases formatted for humans

T X X S X
X X X X X
X X X X X --> 3

T M X S X
X M X X X
O X X X O --> 4

T M X S O
O M X X X
X X X X X --> 2

T M X S X
O M X X X
O X X X X --> 4

T M S X O
X M M M M
X X X X O --> 7


Credits

Design and structure via: Hungry mouse by Arnauld

Proposed Challenges Edit Advice: Kamil-drakari, beefster

• How many teleporters at most? Jan 18, 2019 at 0:33
• @Quintec Since they say pairs, then... only two at most, because there's no way to represent multiple pairs Jan 18, 2019 at 3:28
• @Quintec. Yea one at most. I'll add to the rules. Jan 18, 2019 at 10:13
• You should clarify how diagonal moves work near mountains. The first diagram showing the basic movement options includes diagonal moves, but the example with the strawberry would only be two moves regardless of portals if a diagonal move was used. Jan 18, 2019 at 19:29
• @kmail. Very true. Will add condition when on computer. Jan 19, 2019 at 15:16
• I would suggest taking the "gun" out of the title because "portal gun" suggests being able to place your own portals. I would also allow input to be in any convenient format rather than requiring it to be a list of lists with a predefined key. I think for your examples, you should use space and newline separated grids. And one last thing: rather than "The grid must have a possible solution to reach the strawberry", you should state "You may assume that the input grid has a solution". Good challenge though. Just a couple tweaks will make it a great challenge. Jan 24, 2019 at 17:46
• I would like a testcase where the portal is in the way of the turtle, meaning that the turtle has to use the portal twice. Something like "TOSMO","MMMMX", "XXXXX" or "TOSXX", "MMMMX", "OXXXX" (both evaluate to 4 moves) Feb 3, 2019 at 20:50
• @BlackOwlKai I'm considering making portals single use? Thoughts? Feb 3, 2019 at 21:31
• This is also an option and could allow some more interesting golf approaches. Also, the single-use option would make brute-force algorithms more lengthy, because they need to keep track of whether the portal has been used. Either way, it should be clarified in the rules. But I still would like to add a testcase "XXXMO", "XMXMX", "XOXMX", TMSMX", where the portal is between the turtle and the strawberry, even if portals are single-use. Feb 3, 2019 at 21:53
• Will do. I'll add this to rules aftter I get off my bus ride. As I won't have internet to edit. Cheers, Thanks for the suggestion. Feb 3, 2019 at 22:36
• Support for the emoticons is not guaranteed, and they aren't in the fonts on my Linux system. IMO an image would be preferable. Feb 4, 2019 at 14:36
• @PeterTaylor Will do. I'll do this once home from work. If you want to try doing it now I would approve the edit. Cheers. Feb 4, 2019 at 14:52
• You have a typo on rule 6: "empty gird square" should be "empty grid square". Also, why does it need to be an ASCII character value? Why can't I use an int? Feb 4, 2019 at 17:33
• Will fix the spelling. As far as I'm aware integers are ascii characters? Feb 4, 2019 at 17:37
• @Beefster I'll specify in the post though. Feb 4, 2019 at 17:38
• A worked example or 2 would be useful; I missed, on my first readthrough, that we must take the highest available tile <= the score rather than any tile <= the score and was therefore confused by a couple of the test cases. Although, I'm still confused by the 3rd last one - either the output is wrong or (more likely!) I'm too tired/drunk to figure it out. Nice challenge, though :) Feb 1, 2019 at 1:18

# SLIPpery Packets

The Serial Line Internet Protocol is an early internet protocol, used to essentially escape any "packet END" bytes that may appear in a packet. It has since been replaced by the more sophisticated Point-to-Point protocol, however it is still preferred on microcontrollers and low-level devices due to its simplicity.

Your goal is to implement simplified encoding and decoding functions following this protocol as a function/program/subroutine/etc.

The following table lists the special bytes used:

Hex    Dec   ISO 8859-1   Abbrev
0xC0   192   À            END
0XDB   219   Û            ESC
0xDC   220   Ü            ESC_END
0xDD   221   Ý            ESC_ESC


(The ISO 8859-1 characters are for visualizing where non-ASCII bytes are in the string.)

Encoding follows this system, iterating through each byte:

• If the END byte appears in the packet, write ESC, ESC_END instead
• If the ESC byte appears in the packet, write ESC, ESC_ESC instead
• Otherwise, write byte verbatim
• At the end, an END byte is appended to the message, and it is returned

Some implementations, such as the reference C implementation from RFC 1055, prepend an END byte to the string to be sent. This is omitted here for simplicity.

Decoding follows in reverse:

• If ESC, ESC_END appears in the packet, write END instead
• If ESC, ESC_ESC appears in the packet, write ESC instead
• Protocol violations that involve ESC followed by a "bad byte" that isn't ESC_END or ESC_ESC should be handled by writing the "bad byte"
• The first END byte seen (not following ESC) is interpreted as the end of the packet. The decoded packet should be returned without the END byte. You are guaranteed to have at least one such END byte appear.

## Input/Output

For encoding, input is a packet (string of bytes) and output is the encoded packet (string of bytes).

For decoding, input is a "byte stream" (string of bytes) and output is the (first) decoded packet (string of bytes).

## Test cases

Inputs

''                --> 'À'
'test\n'          --> 'test\nÀ'
'testÀ'           --> 'testÛÜÀ'
'teÛst'           --> 'teÛÝstÀ'
'À'               --> 'ÛÜÀ'
'ÛÜÛÝÀÛ'          --> 'ÛÝÜÛÝÝÛÜÛÝÀ'


Outputs

'À'               --> ''
'testÀblah blahÀ' --> 'test'
'teÛÝstÀ'         --> 'teÛst'
'testÛÜÛÜÀ'       --> 'testÀÀ'
'ÛÝÜÛÝÝÛÜÛÝÀ'     --> 'ÛÜÛÝÀÛ'
'ÛÀÀ'             --> 'À'
'ÛµÀ¶'            --> 'µ'


## Scoring

Your score is the sum of the number of bytes in your encoding and decoding functions (or programs, etc.)

I have also considered requiring functions to take a length of packet in bytes which is used in encoding and decoding (similar to RFC 1055), not requiring handling protocol violations (ex. RFC 1055), and not testing decoding packets that have an END byte as not the last byte. Your opinions on how restrictive this challenge should be are appreciated.

• regex would not be too hard actually.
– qwr
Jun 20, 2019 at 14:42

# One circle, two inputs, eight outputs

Given the values of only two quantities from the following eight quantities ​​(the rest is unknown):

1. Radius of circle: $$\\displaystyle R\$$
2. Diameter of the circle: $$\\displaystyle D=2\times R\$$
3. Circumference of the sector: $$\\displaystyle C=2\times\pi\times R=\pi\times D\$$
4. Area of the circle: $$\\displaystyle A=\pi\times R^2\$$
5. Angle of the sector: $$\\displaystyle\alpha\$$
6. Length of arc: $$\\displaystyle LA=\frac{C\times \alpha}{360°}\$$
7. Area of the sector: $$\\displaystyle AS=\frac{A\times \alpha}{360°}\$$
8. Perimeter of the sector: $$\\displaystyle PS=LA+2\times R=LA+D\$$

compute the rest of quantities if possible, else put special output (-1, False,...) if at least one of the quantities can not be computed.

## Input

Two quantities with there values (non negative integers)

## Output

The values of the eight previous quantities (decimal with any precision or rounded integer).

## Example

In this example the inputs is the ordered list of the previous quantities. Values are integers, angles in degrees, and outputs as rounded integers.

[16, ?, ?, ?, 100, ?, ?, ?]   --> [16, 32, 100, 804, 100, 27, 223, 59]
[24, ?, ?, ?, ?, ?, ?, 58]    --> [24, 48, 150, 1809, 23, 10, 120, 58]
[28, ?, ?, ?, ?, ?, ?, 188]   --> [28, 56, 175, 2463, 270, 132, 1848, 188]
[?, 18, ?, ?, ?, ?, ?, 41]    --> [9, 18, 56, 254, 146, 23, 103, 41]
[?, 28, ?, ?, ?, ?, ?, 808]   --> [14, 28, 87, 615, 3191, 780, 5460, 808]
[?, 88, 220, ?, ?, ?, ?, ?]   --> False
[?, ?, 88, ?, ?, 84, ?, ?]    --> [14, 28, 88, 616, 343, 84, 588, 112]
[?, ?, 220, ?, ?, 66, ?, ?]   --> [35, 70, 220, 3851, 108, 66, 1155, 136]
[?, ?, ?, 50, 50, ?, ?, ?]    --> [3, 7, 25, 50, 50, 3, 6, 11]
[?, ?, ?, 254, ?, ?, 18, ?]   --> [8, 17, 56, 254, 25, 4, 18, 21]
[?, ?, ?, 707, 75, ?, ?, ?]   --> [15, 30, 94, 707, 75, 19, 147, 49]
[?, ?, ?, ?, 99, 24, ?, ?]    --> [13, 27, 87, 606, 99, 24, 166, 51]
[?, ?, ?, ?, 140, ?, 175, ?]  --> [11, 23, 75, 450, 140, 29, 175, 53]
[?, ?, ?, ?, 324, 85, ?, ?]   --> [15, 30, 94, 709, 324, 85, 638, 115]
[?, ?, ?, ?, 324, ?, 3465, ?] --> [35, 70, 219, 3850, 324, 197, 3465, 267]
[?, ?, ?, ?, ?, 66, 88, ?]    --> False


## Rules

• The input and output can be given in any convenient format.
• Specify in the answer the special output when quantities can not be computed all.
• Specify in the answer whether the angle is in degree or radian or other unit.
• π can be set to 3.14 or 22/7 or 355/113 or any more precise value.
• No need to handle invalid input values.
• Either a full program or a function are acceptable. If a function, you can return the output rather than printing it.
• If possible, please include a link to an online testing environment so other people can try out your code!
• Standard loopholes are forbidden.
• This is so all usual golfing rules apply, and the shortest code (in bytes) wins.

# Gregorian-Hijri(Islamic) Calendar Conversion code-golfcalendar

## Introduction

The Hijri, or Islamic calendar is a lunar calendar used in the Islamic countries to determine the dates of traditional events. The year in the Hijri calendar is usually denoted AH xxxx. The Hijri calendar has an epoch on 16 July 622 on Julian calendar.

Like other lunar calendars, in the Hijri calendar a year consists of 12 months, each month consists of 29 or 30 days. A year is thus 354 days, or in leap years there are 355. On average there are 11 leap years every 30 years.

There is a tabular variant which uses arithmetic calculations in place of astronomical observations to determine the dates. There is also a solar version of the Hijri calendar, which is used in Iran and Afghanistan.

In this challenge, only the tabular variant will be discussed.

## The tabular Hijri calendar

All western dates below will be shown in (possibly proleptic) Gregorian even if the date is before 15 October 1582.

There are 12 months in each year in the Hijri calendar, with alternating 29- or 30-day periods. The odd months have 30 days, and the even months have 29 days (except for the 12th month Dhu al-Hijjah in leap years which also has 30 days). The month names are as follows (according to this Wikipedia template):

#    Name              Number of days
1    Muharram          30
2    Safar             29
3    Rabi' al-awwal    30
4    Rabi' al-Thani    29
7    Rajab             30
8    Sha'ban           29
10   Shawwal           29
11   Dhu al-Qidah      30
12   Dhu al-Hijjah     29 / 30 (in leap years only)


There are 11 leap years every 30 years. There are 3 versions to determine the leap years, but the most common version is to set the years with $$\y\in\{2,5,7,10,13,16,18,21,24,26,29\}(\text{mod }30)\$$ as leap years. (Be reminded that the calendar stars from year 1.)

The Hijri calendar has its first day (i.e. 1 Muharram 1) on 19 July 622.

As an example, we now convert the date 8 February 2019 into Hijri (Here we define a cycle to be 30 years):

1. Date: 8 February 2019
2. Dates from the Epoch: $$\1396 \times 365 + 339 + 166 + 38 = 510083\$$
• There are 1396 ($$\2019 - 623\$$) full years, so $$\1396 \times 365\$$ days
• There are 339 ($$\349 - 14 + 4\$$) leap years between 2 dates
• There are 166 days between 19 June 622 to 1 January 623, and 38 days between 1 January 2019 to 8 February 2019
3. Cycle: $$\510083 \div 10631 = 47\text{ (cycles) and }10426\text{ (days)}\$$
• One ordinary year is 354 days and there are 11 leap years in a 30-year cycle, so total days in a cycle is $$\354\frac{11}{30} \times 30 = 10631\$$
4. $$\10427 - (354 \times 29 + 11) = 150\$$, so this this is the 150th day of the 30th year in the 48th cycle, which is $$\47 \times 30 + 30 = \text{(AH) }1440\$$.
• The subtraction has 18 $$\354\$$s and 11 $$\355\$$s, but there is not enough space to write it all.
5. $$\150 - 30 - 29 - 30 - 29 - 30 = 2\$$, so this is the 2nd day of the 6th month in the year AH 1440.

As a result the complete Hijri date of 8 February 2019 is 2 Jumada al-Thani 1440.

## Challenge

Write a program or function that converts the (Gregorian) input date into the corresponding Hijri date.

The program or function may receive the input date in any reasonable format, but it must output the Hijri date in the format of [date] [month name] [year]. The italics above shall be ignored.

The input month and day may be 0-indexed for convenience, but you must state so if you use this convention.

You may assume that the input is always a valid Gregorian date, and is always not earlier than the Gregorian epoch of the calendar, 19 July 622.

## Sample I/O

Input:  "19 July 622" / "622-7-19" / [622, 7, 19] / (622, 6, 19) (0-indexed month)
Output: "1 Muharram 1" (The epoch date)

Input:  "8 February 2019" / "2019-2-8" / [2019, 2, 8] / (2019, 1, 8) (0-indexed month)
Output: "2 Jumada al-Thani 1440" (The example date)

Input:  "12 August 2019" / "2019-8-12" / [2019, 8, 12] / (2019, 7, 12) (0-indexed month)
Output: "10 Dhu al-Hijjah 1440" (The start date of Eid ul-Adha in AH 1440)

Input:  "9 August 2021" / "2021-8-9" / [2021, 8, 9] / (2021, 7, 9) (0-indexed month)
Output: "30 Dhu al-Hijjah 1442" (The next leap day)


## Winning Criteria

As this is a , shortest answer of each language wins. Standard loopholes are banned by default.

• What is "47 R 10426"?
Feb 8, 2019 at 8:16
• @Adám The quotient is 47, with remainder 10426. When I was in primary school I was taught like 5/3 = 1...2 and heard in western countries it'd be written as 5/3 = 1 R2 Feb 8, 2019 at 8:17
• This starts off really well, explicitly addressing the usual ambiguities with Gregorian calendar-related questions and the particular complications of the Islamic calendar. But for me it starts to fall apart at the example, which has a lot of unexplained magic numbers and introduces new terms like cycle. I suggest that you should take the example slower: which number in step 2 is the number of leap days? Define a cycle as (I assume) 30 Iunar years. Feb 8, 2019 at 21:34
• I also suggest that you reconsider the requirement to output month names. That really makes it two barely related challenges (a calendar conversion and a kolmogorov-complexity for an array of 12 strings) packaged as one. Feb 8, 2019 at 21:35

# Unique Skittle Pairs

This is my first question, so hopefully it's OK.

I like to eat Skittles. However, I only like to eat them in pairs, and and those pairs must not have two Skittles of the same color. Your task is to find the most pairs that you can that match these requirements.

# Challenge

Input (taken from standard input) is a set of Skittles formatted like so: pyyyggoooorr, where p is purple, y is yellow, g is green, o is orange, and r is red. The number of letters for each color corresponds to the number of Skittles that are that color. The only colors used in this challenge are purple, yellow, green, orange, and red. The input will always be even in length and lowercase.

Output is the most possible pairs of Skittles that I like to eat given the input. Different outputs don't matter as long as they are valid, i.e. for an input of pygo, both py go and po gy are valid outputs. Output for invalid input doesn't matter.

The output must be in the same format as the test cases.

# Examples/Test Cases

Permutations of these solutions are valid. You do not have to output all valid outputs. Multiple output lines here means multiple valid outputs.

> ppyy
py py

> yggooo
yo go go

> pyygorrr
pr yr yg or
py yr gr or

> yggoor
yg go or
yo go gr
go go yr

> yggor
(anything, including no output/errors)

> gygoro
yg go or
yo go gr
go go yr

> ygoOor
(anything, including no output/errors)

> ggorrrrr
gr gr or

> pppyyygggooorrrr
py py py ro ro ro rg
po po po ry ry ry rg
py po py ro ry ro rg
etc.


Least bytes wins.

• Thanks. I'd forgotten about that. Feb 27, 2019 at 19:16
• I believe this is a dupe. The only differences are the slightly more inconvenient input/output of this challenge and the different group size. However, forcing the output to be the actual pairs prevents using the most dominant strategy of simply calculating the number of groups. Personally I think they are too close anyway, but some people may disagree. Mar 1, 2019 at 20:14

# Sort according to cyclic order

Cyclic order on a set is a function f taking three distinct elements of a set, returning bool and having following properties:

1. If f(a, b, c) then f(b, c, a)
2. If f(a, b, c) then not f(c, b, a)
3. If f(a, b, c) and f(a, c, d) then f(a, b, d)
4. One of f(a, b, c) and f(c, b, a) is true

Values of f when arguments are not distinct don't matter. You are given a list of dictinct numbers and a function f posessing the above properties when arguments are taken from the list. Your task is to arrange the numbers in such an odrer that

f(a, b, c) if and only if k > j > i or i > k > j or j > i > k

(i.e. numbers a, b, c come in that order modulo cyclic permutaion)

Score is the total number of calls to the function f on some dataset.

It's not very clear to me how to create the dataset and how to count calls to f across different languages. Is it better to just give a list of n*(n-1)*(n-2) values of f as input (and score by bytes)?

• Hi and welcome to PPCG. Thank you for using the Sandbox. Please don't specify a challenge using a specific (and unstated) programming language, as many people will have a hard time understanding that. At the very least, explain the notation.
Mar 5, 2019 at 23:49
• @Adám Thank you for your comment. I've removed all language-specific notation Mar 6, 2019 at 0:09
• Is 4. OR or XOR?
Mar 6, 2019 at 5:51
• @Adam 2. says that one of f(a, b, c) and f(c, b, a) is false, 4. says that one is true. Maybe it would be best to combine them into just 2. Exactly one of f(a, b, c) and f(c, b, a) is true Mar 6, 2019 at 20:49
• Neat challenge idea. For the scoring, maybe you can ask for the expected number of calls to f on a random n-element set. It should be easy to Monte Carlo simulate scores, as long as they're different enough. One issue is that people can pick an element e and then only do queries of the form f(e,x,y), to reduce this to a standard sorting problem. Apr 2, 2019 at 20:31

# Venn of $$\n\$$ number of sets

Given positive integer $$\n\$$, output $$\n\$$ 2D bool images with same width and height such that:

1. each image should be 4-connected, i.e. for each two pixels that are true, you can start from one and go up, down, left and right for some times to the other pixel, only passing through true pixels.

2. each image should have no hole, i.e. the component should be 4-connected

3. If we choose itself, its component or neither for each image, their intersection should be non-empty and 4-connected

## Examples (only a finite amount of solution shown, but there are infinitely many more)

Input: 1

Output:

.....
.***.
.**..
..*..


Input: 1

Output:

.....
...*.
.....
.....


Input: 2

Output:

..... .....
.***. ..**.
..... ..**.
..... .....


Input: 2

Output:

..... .....
.**.. ..**.
..**. ..**.
..... .....


Input: 3

Output:

..... ..... .....
.**.. ..**. .....
.**.. ..**. .***.
..... ..... .***.


Input: 4

Output:

...**.. .***... ....... .......
...**.. .***... ..***.. .......
...**.. .***... ..***.. ..****.
....... .***... ....... ..****.
....... ....... ....... ..****.
....... ....... ....... .......


Shortest code in each language win

Reference

• im a little slow... what would be an example of an non-valid output? Mar 23, 2019 at 3:31
• So each image represents one Venn circle? This is a cool challenge. I would (1) say explicitly that each image represents one Venn circle, to make the question easier to understand, and (2) use a different term than "4-connected", which already has a different mathematical meaning. Apr 2, 2019 at 20:25
• @Lopsy I mean the 4-c here
– l4m2
Apr 4, 2019 at 4:04
• Posted
– l4m2
Jun 20, 2021 at 5:31

# Data.List.nub :: Eq a => [a] -> [a]

Tags: , ,

O(n^2). The nub function removes duplicate elements from a list. In particular, it keeps only the first occurrence of each element. (The name nub means essence'.) It is a special case of nubBy, which allows the programmer to supply their own equality test.

> nub [1,2,3,4,3,2,1,2,4,3,5]
[1,2,3,4,5]


## Goal

If you ever golfed in Haskell and you kind of needed to deduplicate a list, chances are that you chose a different route or were annoyed that there seems to be no solution shorter than importing Data.List.

In this challenge you will write a function [a] -> [a] (unnamed or named) which removes duplicates from a list, the order does not have to be preserved.

## Scoring

Since I don't think it's possible to get something shorter than 20 bytes and Haskell sometimes is really annoying to golf with if you want to use lengthy functions, the scoring mechanism will count tokens, please use this program to count the tokens and use the generated template.

• If your solution is a named function or operator, it must not be defined in another file/module (ie. banning imports).

• Each solution may impose type constraints, each set of type constraints counts as its own language, compiler flags can be used without any restrictions.

• Your solution may not assume that the input is non-empty, however if you have an elegant solution which fails on empty input, feel free to include it in your post.

### How are tokens counted?

Tokens are counted by using GHC's lexer which you can import Language.Haskell.Extes.Lexer, except that newlines will count as a token too.

White space is ignored, every token of the language will count as one, no matter how long it is so ma >>= f will count as the three tokens ma, >>= and f the same holds true for string or number literals (eg. "abc" or 1.0 each count as 1).

## Test cases

[] -> []
[1] -> [1]
[1,1,1] -> [1]
[3,1,1,2] -> [1,2,3]
[1,2,3,3,2,1] -> [1,2,3]

• I think I'm missing something, but what stops the import Data.List;nub solution from winning with score 4?
– xnor
Mar 16, 2019 at 1:26
• @xnor: No you didn't, my bad. Fixed, I hope. Mar 16, 2019 at 2:49
• Hmm, this way it's still no good. Using nubBy (==) is still going to be hard to beat. Guess I need to ban imports completely :( Mar 16, 2019 at 2:53
• Yeah, I think banning imports is the way to go. I know past token-counting challenges have had "exploits" where the code payload is put into a huge string or number payload, the decompressed and executed by a general-purpose program. I don't see offhand a good way to do that in Haskell without imports though -- do you?
– xnor
Mar 16, 2019 at 3:03
• I tried writing a solution and the best general-type solution I got is the same as the one I'd use for golfing with imports banned, with 17 tokens. I'm unfortunately not seeing a way to take advantage of long built-ins like the scoring encourages. I see a shorter approach for a restricted type that feels both optimal and hard-to-improve to me. If you don't have some clever better method up your sleeve, I'd suggest a more complex task. Nice token-counting script by way, the number labelling makes it very clear.
– xnor
Mar 16, 2019 at 3:36
• @xnor: Without imports one could still embed an interpreter, I guess. I was thinking earlier about this and I think there's nothing wrong with counting literals as multiple tokens. However yeah, probably it's best to come up with a more complex task. Thanks for your feedback. Mar 16, 2019 at 3:59
• I don't really see how counting tokens is too much better, and it makes the challenge language specific. Mar 19, 2019 at 22:26

A simple Code Golf challenge in the likes of the Markdown parser.

The parser should ignore normal text and other markup (without an # before it).

Some valid headers are: ### Header 3, # Header1, Text # Header

Sample input:

# Hello
Lorem # Hello2
Hi this is normal text

Lorem ipsum

### Subsub
Solor di amet

# Hello again
This is more text!


sample output:

Hello
Hello2
--- Subsub
Hello again

• What about [linebreak]# text in code[linebreak]? Or with indentation? What about where there is no space (still formats) : #no space? Or inside quote: > # header/spoiler: >! #header? Please specify these and any other edge cases. Apr 12, 2019 at 15:45
• I'd personally recommend trying to avoid so many edge cases. If you give a precise list of all of the cases that the parser has to care about and then just say something like "ignore any other markup" I think your challenge will be better. Apr 15, 2019 at 16:10

# Growth of the Jackpot (Tentative Name)

### (Gauging interest)

There is a money pot that gains $1 for every player currently in the game each turn. At any time, a player can decide to take the money and run or stay in the game hoping for a larger pot. If more than one player decides to leave at the same time, the money will be divided evenly among them, rounded down, with the remainder being put back in the pot. If after 100 rounds, there are still players left in the game, they will split whatever is in the pot at that time. Whoever has the most money at the end of the game wins. Write a bot to play this game. You will always know the size of the pot, how many players are currently playing, how many were initially in the game, and the current first place amount of money. On the surface, the optimal strategy is to bail out on turn $$\\lceil{100n \over {2n - 1}}\rceil\$$ (where $$\n\$$ is the number of players) because that turn guarantees that nobody can do better than you. But then if everyone does that, they'll split the pot and lose, so the best action to take is not immediately obvious. One possible variant to round limits would be to have it cost each player$1 per turn they stay in the game, with each player starting with \$100 and being forced to take the money and run when they have none left to contribute. Then the game could be run for several rounds.

• Kind of a one-shot form of Reaper Apr 17, 2019 at 19:47
• I like this. It allows for multiple simple strategies, but has a lot of depth and an interesting metagame aspect. To encourage complicated answers, I would recommend having bots play a few games in a row, to allow a bot to learn optimal timings vs given opponents. Jun 23, 2019 at 4:59
• If a bot may play several games, is the object to win as much money as possible, or to win as many games as possible? Aug 1, 2019 at 15:08

# Hungry Monster 9

My first challenge. Feedback is appreciated.

A monster shaped like 2 is hungry and wants to become a 9. To do so he will eat food, that is shaped like the number 1, seven times.

0000000000
0000100000
0100000000
0000001000
0000020000
0100000000
0000000100
0010000000
0000001000
0000000000


# Challenge

It starts with a grid filled with 0, containing the digit 2, and seven times the digit 1. The 2 (aka the monster) needs to move to the closest 1. Upon "eating" the number, 2 will change to 3, and continue to move to the next closest number.

Write a program that outputs the next correct state.

# Input specifications

• The input grid is a rectangle and can be of any size. It doesn't need to be a square.
• The input can be at any state. For example a grid with a 6 and three times 1 left.
• The grid will contain exactly one digit higher than or equal to 2 (the "player").
• The grid will contain seven times or less the number 1.

# Rules

• When the monster gets on the position of where a 1 is, its digit increases by 1.
• The monsters previous position becomes 0.
• When the monster is 9, it does not move.
• The monster can move in 8 directions (horizontally, vertically, diagonally).
• The monster moves to the closest 1. If there are multiple closest, you may choose any of those. It is not required to be random.

# Example Input and Output

This is an example of the first 4 states (and example inputs/outputs):

0000000000
0000100000
0100000000
0000001000
0000020000
0100000000
0000000100
0010000000
0000001000
0000000000

0000000000
0000100000
0100000000
0000003000
0000000000
0100000000
0000000100
0010000000
0000001000
0000000000

0000000000
0000100000
0100030000
0000000000
0000000000
0100000000
0000000100
0010000000
0000001000
0000000000

0000000000
0000400000
0100000000
0000000000
0000000000
0100000000
0000000100
0010000000
0000001000
0000000000


The output is allowed to be an array of lines. ["0000000000", "0100000000", "0000080000"]

• If there are multiple closest, it can choose any of those. Does it have to randomly choose, or can it (for example) always take the north-most? Apr 19, 2019 at 13:27
• @AdmBorkBork It does not have to be random. Your example is allowed. Thanks, I'll make that more clear. Apr 19, 2019 at 16:45

# Low diversity quine.

A low diversity quine is a quine (program that outputs it's source code) that uses a low amount of distinct characters.

## Scoring:

As a code-golf challenge the program with the lowest score wins.

Score is calculated as: NumberOfDistinctCharacters x ByteCount

## Rules:

Standard quine rules apply. Most notably: No source code reading

• Whitespace and brainfuck have a clear advantage :P. Also, the standard quine for a lot of languages is optimal here. Notably V. Apr 20, 2019 at 18:02
• Pretty sure we've had (something very close to) this before, but I could be wrong. Apr 20, 2019 at 23:21
• @Shaggy I thought the same, but I'm not so sure anymore. I'm probably confusing it with Eww, those bytes are gross, where answers use the least amount of distinct bytes as well, but when the input contains any of the bytes in your source code, they should remain in the output. So it's related to the source code and distinct bytes, but not a quine. Alternatively I could've confused it with Write a Quine Suite, where answers contain multiple quines that share none of the same characters Apr 23, 2019 at 15:47
• that uses a low amount of distinct characters usedthat uses a low amount of distinct characters or with a low amount of distinct characters used Apr 23, 2019 at 21:32

# Sort a list using a stack and a queue

Consider an unsorted list of length n, containing the integers 1 through n. Your job is to sort them using a stack (FIFO) and a queue (LIFO) and a single register. To do this, you have access to six commands

i    Input the next integer onto the register
o    Output the value of the register
Q    Place the value of the register at the tail of the queue (enqueue)
q    Place the value of the head of the queue in the register (dequeue)
S    Place the value of the register at the top of the stack (push)
s    Place the value of the top of the stack in the register (pop)


A few examples (sorting in ascending order):

Input:
[1 2 3 4]
Commands:
ioioioio

Input:
[3 2 4 1]
Commands:
iQiSiQiosoqoqo


## Challenge

Write a program or function that, given some (unsorted) list, returns a sequence of commands that outputs an ordered list. However, the number of commands should grow as $$\o(n^2)\$$, i.e., strictly smaller than $$\\mathcal{O}(n^2)\$$. That is, for sufficiently large $$\n\$$, you must always use less than $$\cn^2\$$ commands (with $$\c\$$ some positive constant) to sort the list.

• The input is flexible. The list of size $$\n\$$ may start from $$\0\$$ or from $$\1\$$ and will always contain all integers from $$\0\$$ to $$\n-1\$$ or from $$\1\$$ to $$\n\$$ respectively. You may take $$\n\$$ as a separate input if desired.
• The output is flexible. You may choose a different set of unambiguous command identifiers instead of ioQqSs. You may output as a string (which is allowed to contain superfluous characters such as delimiters, linebreaks, etc), or as an array or list of commands.
• Your entry must indicate the worst-case output size, e.g., $$\\mathcal{O}(n)\$$, $$\\mathcal{O}(n\log(n))\$$, which must of course be strictly smaller than $$\\mathcal{O}(n^2)\$$.

A few notes:

• Things like ii, qi, oo should never happen. You only have one register, and overwriting it would mean you can no longer output the full sorted list.
• A naive approach would be to put the entire input in the queue iQiQiQ... and then cycling through the queue qQqQqQ... until the register contains the next integer in order. However, this approach would mean you have $$\\mathcal{O}(n)\$$ cycling commands for $$\\mathcal{O}(n)\$$ numbers, which would be $$\\mathcal{O}(n^2)\$$ and thus does not meet the complexity requirement.
• The restricted-complexity tag means that your code must have a restricted complexity in at least one aspect, not that the output has a restricted growth rate. As it is, I can simply brute force and find the optimal output in all cases, but I don't think that's what you want. Apr 29, 2019 at 12:28
• @EriktheOutgolfer Thanks for raising that point. I'm not sure if brute forcing will always be the shortest in byte count, so I would like to think that different approaches are still competitive. Perhaps I could change it such that the entire algorithm must run in o(n^2) (which would automatically disqualify O(n^2) output)? Or perhaps o(n^2) for the output, O(n^3) for the algorithm. What are your thoughts? Apr 29, 2019 at 13:04
• Unfortunately, most of the time, brute-forcing ends up being the shortest approach. :( Also, yeah, requiring the algorithm to run in o(n^2) but still managing to get O(n^2) output means the answer is... weird. However, the other problem is that your code doesn't know which commands to execute from the beginning, and I'm half-confident that o(n^2) will be enough, but not really confident. I don't think the output can really be o(n). As for the specific complexity, well... no specific opinion yet. :P Apr 29, 2019 at 13:38
• Are you sure this is in $O(n^2)$? Something tells me the complexity might be bigger than that when all you have is a stack and a queue. You should also use LaTeX to format your big-O notation: $O(n)$ $O(n \log n)$ $O(n^2)$` Apr 30, 2019 at 0:26
• @Beefster Like explained in the notes, you can do this in $\mathcal{O}(n^2)$ by just using the queue, but I'll have to check if $o(n^2)$ is possible; I'll of course check before posting. Thanks re. MathJax, I forgot this was turned back on. Apr 30, 2019 at 11:27
• I think you need to work out $c$- i.e. you should probably put a hard cap on the number of commands. The worst-case naive approach comes out to $n(n-1) \over 2$ cycle commands for a total of $n(n-1) + 2n$ total commands. You can definitely always do better than that since the naive queue-only approach worst case is a reversed sequence, which is handled trivially by the stack. May 1, 2019 at 19:12
• BTW, I came up with the formula by noticing that it is a shrinking queue and you therefore need at most $l-1$ cycle commands to get to the correct element from any point in the queue. (where $l$ is the queue length) May 1, 2019 at 19:19
• Also I think I have the formula off by a bit: It should be $n(n-1) + 4n$ since you need $2n$ commands for both filling and emptying the queue. May 1, 2019 at 19:21
• In addition, skipping the queue for the first element cuts it down to $(n-1)(n-2) + 4n - 2$ total commands. May 1, 2019 at 19:25
• Following this mental thread, I suggest the following rule: "For any input sequence with length $n \geq 3$, you must output fewer than $n^2 + n$ commands." May 1, 2019 at 19:40
• Your queues work back to front compared to mine: I enqueue to the tail and dequeue from the head. Perhaps it would be cleanest to just talk about pushing to the queue, popping from the queue, pushing to the stack, and popping from the stack. May 2, 2019 at 11:28
• I'm beginning to suspect that the worst case is linear. For $4 \le n \le 8$ the worst case is $4, 5, \ldots, n, 2, 3, 1$ at cost $6n - 10$. May 2, 2019 at 13:55
• @PeterTaylor I was thinking the same, but I didn't have time yet to verify this. And... have you never heard about left-hand-drive queues? May 2, 2019 at 15:59