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


To post to the sandbox, scroll to the bottom of this page and click "Answer This Question". Click "OK" when it asks if you really want to add another answer.

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.


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
  • Comments addressing specific points mentioned in the proposal
  • 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 needs more feedback, but it's been ignored, you can ask for feedback in The Nineteenth Byte. It's not only allowed, but highly recommended!

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.


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!


3548 Answers 3548

74 75
77 78

Output a number tent


Given an integer, output a number tent. (The number tent is just called as such.)

The input determines:

  • whether the tent is upside down (negative) or right way up (positive),
  • whether the entrance is on the left side (even) or the right side (odd), and
  • the size of the tent (abs(input/2) http://mathurl.com/hwkkw3f.png with integer division).

For input -4:
\  /   /

Assume that the input will never be 1, 0 or -1.

TODO Clarify.

Your code should be as short as possible.

  • 1
    \$\begingroup\$ Input of 1 should return an empty string (as should -1 and, of course, 0, although we should clarify that size is abs(input/2) with integer division). Tag suggestion: ascii-art \$\endgroup\$ Sep 7 '16 at 18:18

Find all paths in a matrix maze.

A maze is represented as a two-dimensional array of 0s and 1s. Zero indicates a closed door and one indicates an open door. An example representation of maze is given below.


to avoid loops input square matrix should not contain a square of once like

111      11
101  or  11

Write an application either in Java,C,C++,Python,C# or Javascript. It takes as input a maze(mxn array) and prints out all the possible ways out.

• No database should be used.

• Yes...Way out is any edge containing 1.

• Enter from any edges of the maze

• Exit from any edges of the maze

• Should not enter and exit from the same door

• Move can be one unit distance either top/bottom/left/right/diagonal

For the above maze, one possible way out is: [0,1] -> [1,2] -> [2,3] i.e you start at [0,1] and then move to [1,2] and then to [2,3]. There are other possible ways as you can see.

  • \$\begingroup\$ Probable dupes: One, Two \$\endgroup\$
    – Emigna
    Sep 8 '16 at 15:10
  • 4
    \$\begingroup\$ Also, restricting challenges to specific languages are frowned upon. An objective winning criterion is also needed. \$\endgroup\$
    – Emigna
    Sep 8 '16 at 15:10
  • \$\begingroup\$ @Emigna, No Actually, here we need to find all the paths not the shortest path or straight path and here we can also traverse diagonal in the matrix. \$\endgroup\$
    – Ali786
    Sep 8 '16 at 15:15
  • \$\begingroup\$ okay will add few more languages that I know so that I can evaluate ... \$\endgroup\$
    – Ali786
    Sep 8 '16 at 15:16
  • 1
    \$\begingroup\$ 1. "Handle error cases like a production application" is not a clear requirement. Different production applications have different standards for how to handle errors. 2. "prints out all the possible ways out" runs into trouble when there's an infinite number, and I don't see any guarantee that there won't be loops. 3. The example doesn't make the input format very clear. What's the non-rectangular border for? 4. Path-finding with a trivial input format is definitely a dupe of at least one previous question. \$\endgroup\$ Sep 8 '16 at 15:31
  • \$\begingroup\$ @PeterTaylor, I have modified my question, can you plz check once. \$\endgroup\$
    – Ali786
    Sep 8 '16 at 16:04

Auto golf a string

Your objective is to make a program (written in a language of your choice) that takes arbitrary strings as input and outputs a program in a constant target language that, when run, prints that string. The original program does not have to be golfed, but the output programs should be in some way optimal or nearly optimal.

Voters are encouraged to keep the following points in mind when voting:

  1. How difficult is the target language to program in? Is it on par with languages such as Brainfuck and whitespace, or is it something trivial, like Python or Ruby?
  2. How efficient is the golfer? That is, how does it compare to any existent auto string golfers? And how does it compare to hand-golfing?


Okay, so this isn't our usual type of challenge, but I think this could really turn out well. However, I think this is underspecified as it is, and would appreciate feedback.

  • \$\begingroup\$ This may overlap with some other metagolfing questions... That will mean that any future kolmo meta-golfing qs will be dupes. \$\endgroup\$
    – Beta Decay
    Sep 11 '16 at 20:00
  • \$\begingroup\$ @BetaDecay Yeah, but only in reference to strings \$\endgroup\$ Sep 11 '16 at 21:06
  • \$\begingroup\$ This is a bad fit for a site with a 30000 character limit to answers. I would have to golf 12kB off my GolfScript-Kolmogorov program to fit it in an answer even without any explanation beyond the comments in the code. \$\endgroup\$ Sep 12 '16 at 11:12
  • \$\begingroup\$ @PeterTaylor that is certainly impressive. How substantial are the reductions? \$\endgroup\$ Sep 12 '16 at 11:14
  • \$\begingroup\$ See e.g. codegolf.stackexchange.com/a/42400/194 , codegolf.stackexchange.com/a/11568/194 . Also related. \$\endgroup\$ Sep 12 '16 at 11:25

Balanced use of the alphabet

Write a program that counts from 1 to 100 and back down and ends by the sentence "I finished counting":

1, 2, 3, ... 97, 98, 99, 100, 99, 98, 97, ... 3, 2, 1, I finished counting

The output can be of any kind: array, comma separated, line separated, ...

In order to do that, your code reviewer (who has a wierd OCD) won't accept code that doesn't contain the same amount of each letters of the alphabet.

He also doesn't accept code with no letter at all because, you know... letters are cool.


Your score will be codeLength * (mostRepresentedLetter / leastRepresentedLetter).

Example of scoring

(Bad) code:

for(int i=1;i<=100;i++){abcdefghijklmnopqrstuvwxyz();}

Length: 39
Most represented letter: i (5 times)
Least represented letter: a (once)
Score: 39 * 5 / 1 = 195

NOTE: if a letter is not at all present, divding something by zero will make an inifinite score. As this is , it is not something you want.

  • \$\begingroup\$ As written, what prevents folks from just putting any extraneous letters behind a comment? How would code that doesn't use ASCII, like APL or Jelly, be scored? \$\endgroup\$ Sep 12 '16 at 13:40
  • \$\begingroup\$ Since the scoring isn't based solely on code length, maybe [code-challenge] is more suitable. \$\endgroup\$
    – Geobits
    Sep 12 '16 at 20:07

Validate the traversals

Given inorder and preorder traversals of a binary tree, return a truthy or falsy value depending on whether they could be from the same tree.

Input can be strings of characters or arrays of characters or numbers, whatever is the most convenient. You can assume both inputs are the same length.


inorder: A
preorder: A
result: true

inorder: AB
preorder: AB
result: true

inorder: AB
preorder: BA
result: true

inorder: ABC
preorder: BCA
result: false
possible inorder trees:
    C   C   B   A   A
   /   /   / \   \   \
  B   A   A   C   C   B
 /     \         /     \
A       B       B       C

This is , so the shortest solution wins.

  • \$\begingroup\$ The test cases should include some which repeat characters. \$\endgroup\$ Sep 12 '16 at 13:39
  • \$\begingroup\$ @PeterTaylor I'll have to make it clear in the question that the characters won't be repeated, as they're supposed to refer to different nodes in the tree. \$\endgroup\$
    – Neil
    Sep 12 '16 at 13:54

Illuminati is Illuminati

This is a popularity contest inspired by a dedicated scratch programmer who made the program here: https://scratch.mit.edu/projects/117836320/. What the program does, is it takes in a string, and links the string to the Illuminati. This scratch program is especially popular because it demonstrates meta-satire on linking things to other things.


Your job is to more or less produce something of similar function and form to the one linked above.


Your program can take anything as an input, however, it has to be able to take something. My suggestion is a string.


Your program must output a series of links that make some sense, and eventually link your input to the illuminati.
The last output of your program must be some variation of [input] + " is Illuminati confirmed."


This is a popularity challenge. Your code isn't supposed to look great, or even execute well, so long as it has the most up votes. You must use a code that has a free compiler, and you must post your source code along with your program.

Tips and tricks:

This section will be updated, as people post suggestions in the comments.

  • The Illuminati music in the background could add some spiff to your program. you can download it here: http://www.aiomp3.com/download.php?mp3=hAAlDoAtV7Y
  • Unlike what the program listed above does, there are many more links to have done besides the number of letters an a string. try looking for anagrams, or similar words.
  • \$\begingroup\$ A question should ideally be self-contained. Without reading the external link, which could 404 tomorrow, I have no idea what you want the program to do. \$\endgroup\$ Sep 14 '16 at 10:03
  • \$\begingroup\$ @PeterTaylor the link will not 404. it's a permalink..... and you should know what I want it to do, I put in my post "What the program does, is it takes in a string, and links the string to the Illuminati." and "Your job is to more or less produce something of similar function and form to the one linked above." \$\endgroup\$
    – user56309
    Sep 14 '16 at 14:03
  • \$\begingroup\$ Seems humorous, give some examples. Also be more specific on what "links the string to the illuminati" means. \$\endgroup\$ Sep 14 '16 at 14:43

Translate an SVG path

An SVG path consists of a number of components. Each component begins with a letter, which may be upper case for an absolute position or lower case for a relative position. The component then has a variable number of parameters. Parameters may be separated by commas or spaces.

For the purposes of this question, you will not need to support the H, L or A commands. This means that each component accepts an even number of parameters, and that alternate parameters refer to the X and Y coordinates.

Given an SVG path and X and Y displacements, please output the translated path.

Input and output should be in any reasonable format, as long as you are consistent, i.e. both coordinates should be in the same format, or they can use a Point type, while the output path should be in the same format as the input path.

Please avoid floating-point errors e.g. adding 0.1 to 0.1 and getting 0.199996 like Inkscape does when I asked it to do this.


M 6.4 12.8 L 19.2 25.6 l 6.4 -6.4 z

M 12.8 19.2 L 25.6 32 l 6.4 -6.4 z

This is , so the shortest program wins.

  • \$\begingroup\$ Does "exact arithmetic" mean use of floating point is forbidden? \$\endgroup\$
    – trichoplax
    Sep 14 '16 at 14:05
  • \$\begingroup\$ @trichoplax By "exact arithmetic", I mean that you should ensure that you don't introduce floating-point accuracy errors in your output. The input list may be absolute or relative depending on whether the command letters are in upper or lower case. Relative coordinates do not need to be adjusted, of course. \$\endgroup\$
    – Neil
    Sep 14 '16 at 14:53
  • \$\begingroup\$ It's worth including a summary of what commands need to be supported, and how each works. \$\endgroup\$
    – trichoplax
    Sep 14 '16 at 15:14
  • \$\begingroup\$ Using floating point variables will introduce floating point errors for some values. Do you want to specify a minimum required accuracy, or do you want to insist on only number types that do not share the problems of floating point? \$\endgroup\$
    – trichoplax
    Sep 14 '16 at 15:17
  • \$\begingroup\$ The numbers won't need huge amounts of accuracy, so I don't care how you represent the numbers internally, as long as the output contains no floating-point errors. \$\endgroup\$
    – Neil
    Sep 14 '16 at 15:56
  • 1
    \$\begingroup\$ The problem is it's hard to avoid floating point errors without knowing what accuracy the inputs will have. \$\endgroup\$
    – trichoplax
    Sep 14 '16 at 16:01

Pointfree Generator -- this is a draft.... --

What is this?

From Wikipedia's page:

Tacit programming, also called point-free style, is a programming paradigm 
in which function definitions do not identify the arguments
(or "points") on which they operate


In this challenge, you will have to generate Haskell code in pointfree style. You can provide an answer in any language of your choice as long as it generates Haskell code.

The functions to generate will be polynomial function of multiple variables with integer coefficients like : f : x,y,z -> x^8*y*z + 3*y + 12

for example, the function f : x -> x² + 3x can be written in pointfree Haskell like this (3*)>>=flip((+).(^2)) or (+3)>>=(*)


The input is a list of list of coefficients and exponants : For example, the f : x,y,z -> x^8*y*z + 3*y + 12 function will be defined as [[1,8,1,1],[3,0,1],[12]]



Haskell function/operators allowed are the following : (+),(-),(*),(.),(^),flip,(=<<),(>>=)





Pitchers and Rivers

Credit for the idea for this challenge goes to Dan Garcia of UC Berkeley.

Many of you may have heard of any number of variations on the pitcher problem. For those who haven't, or need a refresher, the generalized form of the pitcher problem is as follows:

Given a list of pitchers of known size and a river filled with infinite water, how can you obtain a specific amount of water?

For a more specific example:

Given a 3-liter pitcher, a 7-liter pitcher, and a river, you can obtain exactly 2 liters as follows:

  1. Fill the 3-liter pitcher.

  2. Pour the 3-liter pitcher into the 7-liter pitcher.

  3. Repeat steps 1 and 2 again twice.

  4. If you didn't spill any extra water when filling the 7-liter pitcher the third time, the 3-liter pitcher will now contain 2 liters.

You cannot pour out part of the pitcher unless you can measure how much you poured out by filling something else up, e.g. another pitcher.

Your goal, now, is to determining which amounts of water you can get from your pitchers, given their sizes... but generalized to any number of pitchers.


You will receive the pitcher sizes in any reasonable format. A (non-exhaustive) list of examples are:

  1. A newline-separated list. 5\n13\n532

  2. A Pythonic list. [3,7,13,22]

  3. Prompt for input, like BF's .[>.]

You do not have to handle invalid input (e.g. negative sizes, empty lists)


A list of obtainable (measurable) amounts of water, in any reasonable format. You may include 0, but you do not need to. You do not need to list the steps by which you obtain these quantities.

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

[3,3] -> [3,6]

[4,6] -> [2,4,6,8,10]

[7, 5, 3, 2] -> [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17]

[30, 15, 5] -> [0, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50]

[5,0] -> [0, 5]

[-1] -> Undefined behavior; sizes are always positive 

Reference solution in Pyth, barely golfed (100 bytes)

  • 1
    \$\begingroup\$ I would suggest [4,6] -> [2,4,6,8,10] as another test case. \$\endgroup\$ Sep 14 '16 at 1:25
  • \$\begingroup\$ This has been asked before, although it's closed and definitely abandoned. It would be nice to ensure that your spec is compatible with the two existing answers and see whether the mods will merge the old question into the new one. \$\endgroup\$ Sep 14 '16 at 7:37
  • \$\begingroup\$ @PeterTaylor The goal of this challenge seems very different from the one linked; that one asks for the list of steps to get a specific target goal and mine asks what target goals are reachable. Also, my question allows for multiple pitchers to be combined in order to get the measurements, while that challenge requires that all the water end up in one pitcher. I believe these are two completely different challenges. \$\endgroup\$
    – Steven H.
    Sep 14 '16 at 9:26
  • \$\begingroup\$ "that one asks for the list of steps to get a specific target goal and mine asks what target goals are reachable" are fundamentally the same thing. The algorithm is identical: the only difference is the output format. "Also, my question allows for multiple pitchers to be combined in order to get the measurements" Does it? I can't see that in the question. \$\endgroup\$ Sep 14 '16 at 10:06
  • \$\begingroup\$ The major difference is that mine doesn't require you to compute the steps (as there are characterizations of these lists that do not involve computing individual steps; part of the challenge may be finding these characterizations.) I didn't include anything saying that you may combine the pitchers because, at the time, I hadn't thought that the distinction would need to be made. However, in the sample outputs you can notice that I include values up to the sum of the pitcher sizes, as opposed to up to the maximum pitcher size. \$\endgroup\$
    – Steven H.
    Sep 14 '16 at 10:18
  • \$\begingroup\$ Is the solution just multiples of the gcd up to the sum of the inputs? If not, you need test cases that show this. \$\endgroup\$
    – xnor
    Sep 14 '16 at 17:17
  • \$\begingroup\$ @xnor, to be honest, I don't know for sure. What test cases would you recommend, though? \$\endgroup\$
    – Steven H.
    Sep 14 '16 at 17:23
  • \$\begingroup\$ I'd strongly suggest figuring out a solution or having a reference implementation to generate lots of cases before posting this to avoid disagreements as to whether a submitted solution works. \$\endgroup\$
    – xnor
    Sep 14 '16 at 17:33
  • \$\begingroup\$ @xnor A reference solution has been added. Anything else I should add? \$\endgroup\$
    – Steven H.
    Sep 14 '16 at 23:27
  • \$\begingroup\$ @StevenH. You should add more test cases, especially ones with more inputs, ones that given unexpected results, ones that break the gcd pattern if they exist, etc \$\endgroup\$
    – xnor
    Sep 14 '16 at 23:36
  • \$\begingroup\$ All the examples I tried on the Pyth app follow the gcd pattern. You could try iterating or generate random examples to look for exceptions. I found proofs that it holds for 2 jugs, maybe one can induct the same onto n jugs. (Edit: This paper proves it) If it holds, I think that's unfortunate because it gives a simple direct solution that doesn't require doing anything specific to the bucket problem. Maybe you could require more output. \$\endgroup\$
    – xnor
    Sep 14 '16 at 23:45
  • \$\begingroup\$ @xnor how about mapping all the reachable bucket states to their total amount of water? I'd submit the (slightly modified) Pyth solution as an actual answer in that case. \$\endgroup\$
    – Steven H.
    Sep 14 '16 at 23:56
  • \$\begingroup\$ What mapping do you mean? Could you give an example? \$\endgroup\$
    – xnor
    Sep 16 '16 at 3:29
  • 1
    \$\begingroup\$ Unfortunately it seems that someone posted almost the same challenge now (presumably unknowingly). \$\endgroup\$ Sep 22 '16 at 18:14

Code Golf: Convert to base 6 and back

If given a number, convert it to base 6. You can assume it is positive. You can get the number through any of these methods: STDIN, or ask for user input. If i has a 'o' at the end of it then assume the number is base 6 and convert it to a decimal number. Assume i is a string.

Example: Given i convert it. So if i = 12 then return 20. Given i convert it. so if i = 12o then return 8

  • 6
    \$\begingroup\$ I don't think this really adds anything to the dozens of base-conversion challenges we already have and is likely to be closed as a duplicate of one of the simpler ones. \$\endgroup\$ Sep 17 '16 at 11:49

Quine and Antiquine

Your task is to make two different program, A and B. Both are nonempty proper quine. However, there is a restriction.

  1. AB and BA produce no output.
  2. CABD and CBAD will do the same thing as CD, where C and D are any (potentially empty) strings made up of copies of A and B. E.g. ABBABA and ABBBAA should behave identically (here, C = ABB and D = A).

The score is the sum of the length of A and B.


A program written in DJam:

program A :


program B :


Then, these output are sastified

>> <empty output>
>> abcd
>> abcd
  • \$\begingroup\$ Are proper quines required? Do A and B have to be different? \$\endgroup\$
    – Dennis
    Sep 16 '16 at 18:40
  • 1
    \$\begingroup\$ "where C and D is any string, if C and D consist entirely of A and B." I don't understand this part, would you mind rephrasing it? \$\endgroup\$ Sep 17 '16 at 11:48
  • \$\begingroup\$ @MartinEnder Sorry if my English is not good. But I mean string S where S = e | S A | S B where e is empty string. \$\endgroup\$
    – Xwtek
    Sep 18 '16 at 7:52

Periodic Pillbox Problem

Pill-popping Patty is prescribed to take a particular pharmaceutical every P days. Presently, Phil has punked Patty's pill-box, punching holes in particular weekday's plastic pockets. Patty proceeds to plan pills P days apart whenever possible, but places pills in the preceding day's plastic pocket if the preferred pocket is punctured. Produce a pill planning chart presuming Patty popped her prior pill on the penultimate day of the previous week.

In case it's not clear, this is the kind of pill-box I am referring to.

The number P, followed by the 1 to 6 weekdays that Phil punctured. Beginning with Sunday(N) the weekdays are N M T W R F S. For instance 5MRF means Patty aims to wait 5 days after each pill, but cannot plan pills on Monday, Thursday, or Friday.

You can have the input with a single space after the number (e.g. 5 MRF) or with spaces after the number and each day (e.g. 5 M R F).

Given 5MRF Patty's first pill is taken on the first Wednesday (5 days from last Friday). Her second pill should be taken on the second Monday(5 days from the first Wednesday), but is moved up to the second Sunday because Monday is punctured. Her third pill should be taken on the second Friday(5 days from Sunday, not Monday), but is moved up to the second Wednesday because Friday and Thursday are both punctured. She continues in this manner indefinitely.

Note on the first week Patty's pillbox contains one pill on Wednesday, on the second and future weeks it contains pills on Sunday and Wednesday.

The pill planning chart shall indicate the state of Patty's pill-box each week, one week per line. Days she does not take her pill must have a dot ·. Days she takes her pill must have a capital X. The chart must stop as soon as the cycle occurs (i.e. in the fewest lines possible) with an arrow > pointing the first week in the cycle on the left. All weekdays must align in singly spaced columns with Sunday on the left.

You may assume Phil was not malicious enough to puncture P consecutive days of the weeks, and that P is between 2 and 7 inclusive.

If Patty can't take her pill on the first Sunday, she takes it on the previous Saturday. Her pillbox was not damaged last week.

A cycle has occurred when pills are being taken on the same weekdays as in an earlier week.

Patty never waits more than P days to take her pill.

Testcase 1: 5MRF

  · · · X · · ·
> X · · X · · ·

Testcase 2: 5MR

> · · · X · · ·
  X · · · · X ·

Testcase 3: 5MF

  · · · X · · ·
> X · · · X · ·
  · · X · · · ·

Testcase 4: 6NTFS

  · · · · X · ·
> · · · X · · ·
  · X · · X · ·

A reference solution in C++ is provided here.

For The Sandbox...
First Challenge. Hooray! My biggest concern is whether or not there is anything with a similar process, I'm not sure if there is a generalization or a name for this sort of elimination thing. Barring that, is a good golfing question, can it be made a better golfing question, is clear what has to be done, etc. Really any advice is welcome.

  • \$\begingroup\$ 1. "For instance 5MRF means Patty aims to wait 5 days after each pill, but cannot plan pills on Monday, Thursday, or Friday." So what does that mean in terms of the actual days in which she takes the pill? I think that she ends up taking them too frequently rather than averaging out in the long-term, but I had to try to figure that out by reverse engineering the test cases. 2. "followed by the 1 to 6 weekdays that Phil punctured ... P is between 2 and 7 inclusive" seems contradictory. \$\endgroup\$ Sep 21 '16 at 10:54
  • \$\begingroup\$ @PeteTaylor, Okay... I've added a plain English example of the process. Yes, Patty takes her pills too often. To point 2, I don't see the contradiction, P is the days she waits to take a pill, not how many days Phil punchs. Her process would never work with say 1W, or 7NMTWRFS. \$\endgroup\$
    – Linus
    Sep 21 '16 at 16:02

Checkmate (aka the urinal problem) V2

inspired by Checkmate (aka the urinal problem)

given a non-negative integer n, assume there are n urinals.

also assume that you are not allowed to use an urinal when

  • it is out of order.
  • or a neighboring urinal is occupied.

If no urinal is available, the situation is considered checkmate.


Create a program or function that finds and prints (or returns, or yields) all possible checkmate combinations for n urinals.

Input: STDIN, command line argument or function argument

Output: Any output that makes the combinations clear is accepted.

TODO: add visual and test cases.

a thought:
This cries for string operations; how about a bonus for a calculational solution? Do bonuses suck?

  • 4
    \$\begingroup\$ You say "out of order" but then don't mention it again. How does that change things and how is it given in input (if it is)? \$\endgroup\$
    – Geobits
    Sep 21 '16 at 20:12
  • \$\begingroup\$ @Geobits I hope the edit clarifies. \$\endgroup\$
    – Titus
    Sep 21 '16 at 20:20
  • 1
    \$\begingroup\$ Yes, bonuses suck. \$\endgroup\$
    – m-chrzan
    Sep 22 '16 at 2:26
  • \$\begingroup\$ Do you have a reference implementation? Have you checked for the sequence on OEIS? (I ask the second question because once I saw the OEIS page for yesterday's urinal challenge I was tempted to vote to close as a dupe of a Fibonacci question because the sequence was so similar). \$\endgroup\$ Sep 22 '16 at 17:17
  • \$\begingroup\$ @PeterTaylor: not yet. and: I don´t even know OEIS. Why should I check it? \$\endgroup\$
    – Titus
    Sep 22 '16 at 18:36
  • \$\begingroup\$ The Online Encyclopaedia of Integer Sequences is the biggest repository of integer sequences that I know of. Whenever an integer sequence question is asked, people will look there for formulae. Now: having re-read this question and on the assumption that any number of urinals can be out of order, this is asking for sequences of length n over the alphabet 01 avoiding the substring 11. They're counted by the Fibonacci numbers, so there's a good chance that there's an earlier question close enough to be a duplicate. \$\endgroup\$ Sep 22 '16 at 20:38
  • \$\begingroup\$ I believe f(n) = f(n-1) + 3f(n-2) + 2f(n-3), with some boundary conditions. \$\endgroup\$
    – m-chrzan
    Sep 22 '16 at 21:01

Compute The anti-derivative of a non-constant function

Most of us know how anti derivatives work. They are basically the sums of infinite number of very small elements (or units).

The challenge here is to calculate anti derivative of a function (NON CONSTANT). using any programming language except the ones which calculate anti derivatives directly (Mathematica for instance).

Anti derivatives which have been calculated must be done by the usual summation method which is basically the definition of the integral. Summation of y.dx over some interval of x)

You basically have to find the anti derivative WITH limits i.e. definite integral of a non constant function in between two user specified limits

  • 2
    \$\begingroup\$ As it stands this would get closed very quickly as "Unclear what you're asking". Even at the very basic level of whether you want symbolic or numeric integration, there are hints pulling in opposite directions. Either way, you need a specification of the input format; if you want quadrature you need to specify constraints on the rule; and you need to avoid duplicating codegolf.stackexchange.com/q/2072/194 and codegolf.stackexchange.com/q/66714/194 \$\endgroup\$ Sep 23 '16 at 20:06
  • 1
    \$\begingroup\$ I though the challenge is about finding a sybolic anti-derivative, not a numeric definite integral. \$\endgroup\$
    – Vi.
    Sep 24 '16 at 6:33

Download a file

Your boss wants you to make a comprehensive download utility. She uses Microsoft Bob and thinks shorter code is better code.

The Gist: In as few bytes as possible, given an input URL, output the file to stdout.


  • You will be given a fully qualified (w/ http/https) URL.
  • Download link may redirect. Follow all redirects.
  • Any non 2xx-3xx status code should be interpreted as an error. Exit with a status code of zero.
  • Verify SSL certificates - exit with 0 if invalid


  • Do not use a preexisting utility already designed for downloading files, such as wget. This is considered cheating.
  • \$\begingroup\$ I'd consider this a chameleon challenge. Validating an SSL certificate is harder than all other parts combined. It also needs some clarification. Which SSL certificates are considered valid? \$\endgroup\$
    – Dennis
    Sep 25 '16 at 17:32
  • \$\begingroup\$ @Dennis hm, libraries should exist... \$\endgroup\$ Sep 25 '16 at 18:03
  • \$\begingroup\$ 1. For those languages which do have SSL libraries (which isn't all of them), there are often options. E.g. there could be an option to accept or reject self-signed certs. There could be an option to accept or reject certs whose chain includes MD5 or key sizes smaller than 256 bits. Those options may vary between libraries or distros. So Dennis is right to ask for a clear definition of "valid". 2. What's the distinction between libraries which do that level of checking and "a preexisting utility"? \$\endgroup\$ Sep 25 '16 at 21:41
  • \$\begingroup\$ Someone might download a virus... \$\endgroup\$ Sep 28 '16 at 12:42

Handwriting Recognition

The MNIST dataset is a series of handwritten digits used as a standard testbed for machine learning, pattern recognition techniques. Each image is of a single digit, 0-9; as a 28x28 pixel grayscale matrix with values from 0-255.

MNIST example images

The challenge is to create a classifier for MNIST that scores an Error Rate of less than [TBD] in the least number of bytes possible.

Your program must take a 28x28 2d array in whatever format is applicable for your language representing a single image. For example, the input for the first digit might be:

[[  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  3, 18, 18, 18,126,136,175, 26,166,255,247,127,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0, 30, 36, 94,154,170,253,253,253,253,253,225,172,253,242,195, 64,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0, 49,238,253,253,253,253,253,253,253,253,251, 93, 82, 82, 56, 39,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0, 18,219,253,253,253,253,253,198,182,247,241,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0, 80,156,107,253,253,205, 11,  0, 43,154,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0, 14,  1,154,253, 90,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,139,253,190,  2,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, 11,190,253, 70,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, 35,241,225,160,108,  1,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, 81,240,253,253,119, 25,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, 45,186,253,253,150, 27,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, 16, 93,252,253,187,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,249,253,249, 64,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, 46,130,183,253,253,207,  2,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, 39,148,229,253,253,253,250,182,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, 24,114,221,253,253,253,253,201, 78,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0, 23, 66,213,253,253,253,253,198, 81,  2,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0, 18,171,219,253,253,253,253,195, 80,  9,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0, 55,172,226,253,253,253,253,244,133, 11,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,136,253,253,253,212,135,132, 16,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0],
 [  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0]]


  • This is code golf. The shortest piece of code that meets the criteria wins.
  • The code must take a provided 28x28 input and attempt to classify it.
  • Testing will take place on my computer at 12pm AEST on Saturday the . I will run each classifier over a set of 1000 images. To be considered, it must correctly classify [TBD] of them. If I can't get your code to run, it wont be counted, so help with loading the images in your language would be appreciated.
  • Standard loopholes are not permitted.

Questions for Sandbox

  • Overall thoughts on the challenge?
  • Has anything been done like this before? Did it work?
  • Any ideas on a good cutoff for the classifier? I was thinking around 60% correct. Though was going to have a go at it myself to see what I could reasonably achieve.
  • Does the testing clause make sense? Is it reasonable? Should I put a limit on the languages?
  • Since barrier to entry is a bit high (knowing how to get hold of the images, possibly some ML experience), is there anything extra I should do to make it easier to start the challenge.
  • This is my first suggestion for a challenge, is there anything I'm missing?
  • \$\begingroup\$ I actually found a very similar question from back in 2014 (using MNIST). It had a different scoring system and some slightly different criteria and expectations. Is this still worth it? Is it late enough for others to have another go? \$\endgroup\$
    – SCB
    Sep 26 '16 at 3:36
  • \$\begingroup\$ I don't think it's sufficiently different. If you want to provoke others to have another go, you can do so in the chat or by placing a bounty on the existing question. \$\endgroup\$ Sep 26 '16 at 13:44
  • \$\begingroup\$ 60% correct is way too leniant; current state-of-the-art techniques on this dataset "easily" reach 99+% of correct classification. \$\endgroup\$
    – Fatalize
    Sep 28 '16 at 11:48
  • 1
    \$\begingroup\$ Unrelated, but I would like to congratulate you on submitting the 3000th answer to the Sandbox! (That number includes all deleted submissions.) \$\endgroup\$
    – PhiNotPi
    Sep 28 '16 at 15:12
  • \$\begingroup\$ @Fatalize I was thinking 60% because I wanted the emphasis to be on the code-golfing, not the machine learning. If it's lenient enough I'm hoping someone can figure out some really, really simple way that would just scrape in. All in the spirit of code golf. \$\endgroup\$
    – SCB
    Sep 28 '16 at 23:02
  • \$\begingroup\$ @PeterTaylor I was hoping that the idea of a hard limit would switch the challenge around so it's not focused on building a function that works well, but instead building a very short function that works, barely. In the original a slight improvement in code length that decreased the correct matches would be bad, whereas in this challenge decreasing the code length is always good, as long as you still sit above [TBD] correct images. \$\endgroup\$
    – SCB
    Sep 28 '16 at 23:09

Gif - Jif, Jif - Gif

The point of this challenge is to (not) settle debate on the pronunciation of "gif" http://38.media.tumblr.com/11bc6092d29c8c6840f9e10184ca03fd/tumblr_mlksz8SLjn1qz4h3co1_1280.gif

The pronunciation of gif is debated and while it's supposed to be (and should be) pronounced jif, it's still commonly disputed.

In this challenge you will be given a set of words that have a g or j, representing the sound that the word is pronounced with. You'll also get some text in which you have to fix the incorrect spellings of gif.

Because this is the internet and everyone always wrong. It's (not) common courtesy to correct them.

Duty Calls

An example of a wrong pronunciation would be:

There was a gif of a mouse eating a burrito

The wrong spelling of gif? Unacceptable! This must be corrected immediately:

There was a jif (as in jar) of a mouse eating a burrito

Are we done? Nope, you're still wrong.

You're always wrong http://travelsummary.com/wp-content/uploads/2013/03/Im-Right-Youre-Wrong-e1363993322818.jpg

This must work the other way:

In the jif the cat wore a hat of a cat

This obvious misspelling must be fixed, we shall correct this to:

In the gif (as in graphic) the cat wore a hat of a cat


  • You may assume all words that contain the letter g have the hard g sound (as in gravy) and all words that contain the letter j have the j sound (as in jam)
  • Words will never have both g and j (Will add more specific rules later)
  • Words must be picked at random each item with the same chance of being picked
  • Words may be supplied through an array or the closest alternative for your language.
  • Case must be preserved E.g. GiF -> JiF.
  • You may write a program or a function


Input and output separated by a single line:

graphic, jar, jam, gram
I saw a jif of how to pronounce gif that showed gif is pronounced jif

I saw a gif (as in graphic) of how to pronounce jif (as in jar) that showed jif (as in jam) is pronounced gif (as in gram)

gravy, jeff
G is for gif, h is for jif, i is for gif, j if for jif

G is for jif (as in jeff), h is for gif (as in gravy), i is for jif (as in jeff), j is for gif (as in gravy)
  • 1
    \$\begingroup\$ Is the assumption that, if there's more gif/jifs than input words, that you cycle through the g and j words independently (i.e. for gravy, jar, jam, gram, you cycle through gravy, gram for jif and jar, jam for gif)? And do we have do anything special for soft g words, e.g. generate? \$\endgroup\$
    – Sp3000
    Oct 23 '15 at 12:28
  • \$\begingroup\$ @Sp3000 forgot to specify that. I've made it so all words containing "g" are the hard g sound \$\endgroup\$
    – Downgoat
    Oct 23 '15 at 22:49
  • 2
    \$\begingroup\$ "Words must be picked at random" from what source? \$\endgroup\$ Oct 24 '15 at 8:49
  • \$\begingroup\$ Has this challenge been posted? If not, may I take control of it? \$\endgroup\$
    – Neil A.
    Jun 5 '17 at 23:09
  • \$\begingroup\$ @NeilA. You may but I don't think it'll be well received because it's a simple find/replace \$\endgroup\$
    – Downgoat
    Jun 6 '17 at 21:47
  • \$\begingroup\$ @Downgoat: I will make a new answer with the same core idea but a few changes. Feel free to delete this if you wish. \$\endgroup\$
    – Neil A.
    Jun 6 '17 at 21:52

Largest Number Given Digits and Operators


Given a list of digits (0-9) and operators (+, -, /, *), output the largest number that can be formed using those digits and operators, as well as the mathematical formula used to create this number.


  • All digits and operators must be used once and only once
  • Parenthesis may be used in the outputted mathematical formula
  • The outputted mathematical formula must be valid in an interactive Python shell (Note, this means ["--", "---", etc.], "//", "**", and ["++", "+++", etc.] are all valid operations that can be used in the output)
  • Base 10 will be used
  • For multiple different formulas that result in the same maximum number, output whichever one
  • Input may contain any number of digits and operators
  • Answers can be a function or program, and will be scored in bytes
  • No loopholes or built-ins

Example I/O

Input: [1, 4, 3, *, +]
Output: +31*4

Input: [1, 2, 0, '+']
Output: +210
  • \$\begingroup\$ won't all pluses be at the front of the output? \$\endgroup\$ Sep 28 '16 at 2:30
  • 2
    \$\begingroup\$ This is rather similar to a number of existing questions, and may be closed as a duplicate. If you want to try to get it ready to post regardless, the use of the word "operators" in the first sentence is misleading given the later reference to combining those characters to form operators like **; and the question needs a complete list of permitted operators and what they do. People shouldn't have to find a Python language reference to know what // means, for instance. \$\endgroup\$ Sep 28 '16 at 8:47

The one-looper ASCII box


I saw this one in one of my university programming textbooks as an "advanced" exercise and I thought it would fit very well here.


Your program have to ask for an input number and draw a box with the size of the given number and a cross in it.
Example if the number is 7:

|\     /|
| \   / |
|  \ /  |
|   X   |
|  / \  |
| /   \ |
|/     \|

Example if the number is 10:

|\        /|
| \      / |
|  \    /  |
|   \  /   |
|    \/    |
|    /\    |
|   /  \   |
|  /    \  |
| /      \ |
|/        \|

The minimum width is 1 and it looks like this:


Notice that with an odd size, there is an X in the middle.

The hard part

In your program only one function is allowed (main) and in the main function there can be only one loop and that loop can only have one statement which can't be if. The shortest answer wins!

note: Sorry for the messy description but I had to translate it from my native language.

  • 1
    \$\begingroup\$ I'd recommend removing the "C-only" part. \$\endgroup\$
    – wizzwizz4
    Sep 29 '16 at 18:23
  • 1
    \$\begingroup\$ Language specific challenges are discouraged, as per meta.codegolf.stackexchange.com/a/8058/48538. Good luck. \$\endgroup\$
    – user48538
    Sep 29 '16 at 18:27
  • 2
    \$\begingroup\$ do X without Y is confusing, especially when some internal things in languages use loops to achieve certain functionallity \$\endgroup\$ Sep 29 '16 at 20:42

Interpret loose ranges

ListSharp is an interpreted programming language that has many features, one of those features is a 1 index based range creator that works like this:

You define a range as (INT) TO (INT) or just (INT) where both or the single int can go from min to max int32 value

Then you can use those ranges to extract elements of an array without fearing to overstep it's boundaries


1 TO 5 generates: {1,2,3,4,5}

3 generates: {3}

Ranges can be added up using the AND operator

1 TO 5 AND 3 TO 6 generates: {1,2,3,4,5,3,4,5,6}

remember this works with negative numbers as well

3 TO -3 generates: {3,2,1,0,-1,-2,-3}

The challenge is the following:


A character array and the previously defined range clause as a string


The elements at the 1 index based locations of the range (non existing/negative indexes translate to an empty character)

How to win

Create the program with the shortest byte count

Test cases:

input array is:
{'H','e','l','l','o',' ','W','o','r','l','d'}

range clause:
"1 TO 3" => {'H','e','l'}
"5" => {'o'}
"-10 TO 10" => {'','','','','','','','','','','','H','e','l','l','o',' ','W','o','r','l'}
"0 AND 2 AND 4" => {'','e','l'}
"8 TO 3" => {'o','W','','o','l','l'}
"-300 AND 300" => {'',''}
  • \$\begingroup\$ What is the winning criterion? \$\endgroup\$
    – acrolith
    Sep 29 '16 at 20:49
  • \$\begingroup\$ correct, i forgot that! its code-golf and i dont know how to add the tag \$\endgroup\$ Sep 29 '16 at 20:50
  • \$\begingroup\$ You can place [tag:code-golf] near the title. \$\endgroup\$
    – acrolith
    Sep 29 '16 at 20:51

Balanced Columns

I have a number of paragraphs of various heights which I would like to be distributed between some columns as evenly as possible. Here evenly means that you should minimise the total height of the tallest column, and then maximise the height of the shortest column, excluding empty columns.

The input will be an integer, representing the number of columns to be output, and an array of integers, representing the paragraph heights in lines. There is more than one way to represent the output; it could be an array (one for each column) of arrays of paragraphs, or it could be an array (one for each paragraph) of columns. (Please indicate whether your arrays are 0- or 1-based.) You may use any convenient format for (e.g. the number of columns as a command-line parameter and the array on STDIN) for I/O.

Because the paragraphs are double-spaced, there will be at least two lines in each paragraph, but if you prefer, your answer can take the double-spacing into account internally.

Within each column, paragraphs should be listed in ascending order, and the columns should be listed in ascending order of their first paragraph. Examples:

2, [2, 3, 4, 5, 6] -> [2, 3, 5], [4, 6] or [0, 0, 1, 0, 1]
3, [2, 3, 4, 5, 6] -> [2, 5], [3, 4], [6] or [0, 1, 1, 0, 2]
4, [2, 3, 4, 5, 6] -> [2, 3], [4], [5], [6] or [0, 0, 1, 2, 3]
5, [2, 3, 4, 5, 6] -> [2, 3], [4], [5], [6] or [0, 0, 1, 2, 3]

Note that in the last example the height could not be reduced by adding the extra column so in this case only four columns could be utilised.

  • \$\begingroup\$ If it's not a requirement to use every column (as indicated by your last example), that should be called out explicitly rather than inferred. \$\endgroup\$ Sep 30 '16 at 14:02
  • \$\begingroup\$ @TimmyD "maximise the height of the shortest column, excluding empty columns" not explicit enough for you? \$\endgroup\$
    – Neil
    Sep 30 '16 at 17:25
  • \$\begingroup\$ Yeah, as I didn't understand that to mean that I could optionally choose to have empty columns if it made the output more even. Maybe "explicit" wasn't the right word in my comment, but it could be more clear. \$\endgroup\$ Sep 30 '16 at 17:57

No, this challenge is not this

Where is my error?

My compiler has broken down and now it is only showing the index of the character which the error is on.

In this challenge you will write a program to point to the given character.



if (goat == Downgoat) {


1 | if (goat == Downgoat) {
2 |     print(Downgoat));
  |                    ^
3 | }



  • \$\begingroup\$ I was going to suggest giving an {x, y} pair as input, but that would make the challenge substantially easier. \$\endgroup\$ Sep 30 '16 at 22:40
  • \$\begingroup\$ What if there are more than ten lines? \$\endgroup\$ Oct 5 '16 at 1:14

Is this Rubik's Cube Solvable?

The challenge: find out if a rubik's cube configuration is solvable.

Takes input as a multidimensional array. An example for a solved cube would be


starting on the sides, going round, then top, then bottom. This input would give an output of 1, since it is solvable (very, very solvable). A (small) net of this solved cube would therefore be:

 _ _|4|_
 ¯ ¯|5|¯

The result of the program should be a truthy/falsey value, printed to STDOUT or similar. This is code-golf, so shortest answer in BYTES wins. Standard loopholes apply.

  • \$\begingroup\$ What is the winning criterion? \$\endgroup\$
    – acrolith
    Oct 12 '16 at 17:16
  • \$\begingroup\$ shortest code. I'll add code-golf as a tag now. \$\endgroup\$ Oct 12 '16 at 17:20
  • \$\begingroup\$ I think you've got an extra 2 side in your example. \$\endgroup\$ Oct 12 '16 at 17:55
  • \$\begingroup\$ whoops. Thanks for pointing that out. I will add some test cases soon. \$\endgroup\$ Oct 12 '16 at 17:57
  • \$\begingroup\$ Add for people less familiar with the cube group that it has 12 = 2 * 2 * 3 orbits, 2 because you can't only exchange edges, 2 because you can't only flip an edge and 3 because you can't only turn a corner \$\endgroup\$
    – Ton Hospel
    Oct 12 '16 at 22:25
  • \$\begingroup\$ codegolf.stackexchange.com/q/10768/194 . There's no fundamental difference between solving it and testing whether it can be solved. \$\endgroup\$ Oct 13 '16 at 7:21
  • \$\begingroup\$ @PeterTaylor I'd say there is a big difference. Trying to solve a cube is pointless if it's impossible to begin with. This is about spotting which cubes have been tampered with. \$\endgroup\$ Oct 13 '16 at 9:27
  • \$\begingroup\$ Do you have a reference implementation which uses an approach other than trying to solve it and seeing whether you get stuck? (Also, if it is possible to do it without solving, I strongly recommend trying to find a way to prohibit brute-force solving, because that can be very short). \$\endgroup\$ Oct 13 '16 at 10:01
  • \$\begingroup\$ Not yet, but I'm working on it \$\endgroup\$ Oct 13 '16 at 12:52

Image Smoothing

Input into your program a black and white image. (png, jpg, etc)

Convert the image to a 2-d matrix of pixels. Where each index is the grayscale value of the image.

Take each index of the matrix and average the 8 pixels around it. Then replace that index with the average.

For example if you had:

120 80  60
20  10  30
40  100 05

If you were finding the average for 10, the result would be 56.875. Round to up to the nearest integer (57) and replace the index. The result of this iteration would be

120 80  60
20  57  30
40  100 05

If you are finding the average for a pixel on an edge then only take values that are valid. In the example above if you were finding the average for '20' You would only consider 112, 80, 57, 100, and 40, which is 79.4 rounded to 80.

You may start at any index of the graph but it must be in order, for example of you must go right - left, left - right, up - down, or down - up. You may not go something like randomly pick the indices to average.

Once the averaging is complete, convert the matrix back into a new image.

display the new image as the output of the program. (Or save it and add it in your solution). The result of this process should be a normal crisp looking image and the result would be a smoothed out image.

Please provide your code and start and end image (SFW please) and remember this is code-golf so the shortest code wins!

Good Luck and have fun!


Implement BozoCrack

Designed to show why MD5 is a bad solution to password hashing, BozoCrack, for the uninitiated, is a rather simple and efficient method of cracking MD5 hashes. It doesn't brute force, use rainbow tables or indeed, any sort of math at all.

Instead, it finds the key by searching google for the hash and comparing the hash of each word in the results with the input.

Your task is simple: given at least one valid MD5 hash in a reasonable input format, find their source texts using Google and return or print to stdout the results.

In fact, you could probably arrive at the answer by finding the word in the page with the most occurrences -- you needn't use the results for a dictionary attack, but you must use Google.

The inputs can be given as an array of strings, as program arguments, space or otherwise delimited strings on standard input or in a file, etc -- whatever's easier for you, but make sure you specify in your answer how the input is taken.

Example inputs / outputs:

Input                            : Output
fcf1eed8596699624167416a1e7e122e : octopus
bed128365216c019988915ed3add75fb : passw0rd
d0763edaa9d9bd2a9516280e9044d885 : monkey
dfd8c10c1b9b58c8bf102225ae3be9eb : 12081977
ede6b50e7b5826fe48fc1f0fe772c48f : 1q2w3e4r5t6y

Of course, you may use your language's library function for computing the md5sum of a string -- but please, don't use any functions or libraries designed to solve exactly this task, e.g. Python's PyBozoCrack and the original BozoCrack.rb are disallowed.

In the unlikely event there are no results for a sum, print "NO RESULTS" to stdout instead.

  • \$\begingroup\$ Nitpicking: 1. There's nothing special about MD5 here. Using any hash function without a salt has the same problem. 2. It sorta does use rainbow tables, just not locally stored ones. \$\endgroup\$
    – Dennis
    Oct 18 '16 at 16:46
  • \$\begingroup\$ @Dennis The github description says Instead of rainbow tables, dictionaries, or brute force,.., which is even more incorrect, so I made it a little less wrong :) It does work for any hash algo but it works best for MD5 \$\endgroup\$
    – cat
    Oct 18 '16 at 17:02


(Slight paraphrased) Extract from Wikipedia:

AlphaGo is a computer program developed by Google DeepMind in London to play the board game Go and is the first computer Go program has beaten a 9-dan professional without handicaps.

In order to appreciate the Go art more, let's try to simulate what happens to the game when a move is made!


The input will be:

  1. a 4x4 board, (Here . denotes empty, o and x denotes different color of stones held by the 2 players respectively for illustration. Any standard input formats (2d array, array as sequence of alternative moves...) are welcomed as we like nice algorithms!)

  2. a move to be made (Like A4, B2. You may also like to use 0-indexed and both numeric indices like (0,0)..(3,3)) For simplicity's sake, the move to be made is always with o (i.e. always the same color).


The board after the move is made. Stones of the opposite color (i.e. x) will be taken away iff there are no more connected x that has at least 1 . in its 4-neighbor (Up, Down, Left, Right).


  Input | Move | Output
  ABCD  |      |
1 ....  | A1   | o...   (case 1: all stones remains on board)
2 ....  |      | ....
3 ..x.  |      | ..x.
4 ....  |      | ....
  ....  | B1   | .o..
  oxo.  |      | o.o.   (case 2: x taken away: the only . is taken)
  .ox.  |      | .ox.
  ....  |      | ....
  ..o.  | B1   | .oo.
  oxxo  |      | o..o   (case 3: >1 x taken away: the only . is taken)
  .oxo  |      | .o.o
  .xo.  |      | .xo.
  o.xo  | B1   | oo.o
  ..o.  |      | ..o.   (case 4: Edge case)
  ....  |      | ....
  ..o.  |      | ..o.
  x.o.  | B1   | .oo.
  ooxo  |      | ooxo   (case 5: Corner case)
  ...o  |      | ...o
  ..o.  |      | ..o.
  xxxx  | B2   | ....
  x.xx  |      | .o..   (case 6: Enclosed by Board case)
  xxxx  |      | ....
  xxxx  |      | ....
  .oxx  | C2   | .o..
  ox.x  |      | o.o.   (case 7: Enclosed by o case)
  oxxx  |      | o...
  .ooo  |      | .ooo
  .xo.  | A1   | o.o.
  xo..  |      | .o..   (case 8: taking multiple groups case)
  o...  |      | o...
  ....  |      | ....


  1. This is thus the shortest code wins.
  2. Please kindly provide explanations and ways to test the solutions!
  3. The Input board can be assumed to be correct: there will be neither x that without a connected x with ., nor a group of o surrounded by x without .
  4. The Input move can be assumed to be legitimate: either it won't be a position surrounded by x, or it must take away some x to free . for the o to live. (ref. case 6, 7, 8)
  5. The output must be the board, it may or may not be in the same format as the input (if you take input as 2 arrays of o and x positions, say). Output that only shows the stones that are taken away after the move is made will not be accepted.
  6. Again, feel free to play with the standard input formats. Hope to see interesting 1 and 0 playing Go instead of o and x :P

Meta Questions

  1. I don't know if this may be too hard. Would it be better if I allow golfers to fail some test cases and get penalty instead?
  2. Should I stick with a 4x4 board or I can award minus scores if the solution can work with a 19x19 board?
  3. If it is a yes to any of the above questions, which should be the number for the penalty / award?


  • \$\begingroup\$ If it can work with a 4x4 board it's probably scalable. I don't see any reason to have both; 4x4 seems less daunting, but 19x19 would be more satisfying to program for, even if it is pretty much the same thing. \$\endgroup\$
    – wizzwizz4
    Oct 18 '16 at 17:42
  • \$\begingroup\$ The explanation of what exactly the program should do is rather poor. I think I can work it out, but only because I've played Go in the past. IMO the question would be greatly improved by removing the current introduction and replacing it with an explanation of liberties (which in essence is half of the rules of Go, isn't it? That plus ko). As far as size, don't do bonuses; just require answers to handle a board of any rectangular size, with a guarantee that each line is the same length. Or maybe make each dimension at least two. Then add test cases to cover the minima. \$\endgroup\$ Oct 19 '16 at 11:52

Categorize the asymptotic behavior of a consumer/producer system

Given two functions P(t) and C(t), output the correct category for the behavior of P(t)-C(t) as t goes to positive infinity:

  1. Ideal: P(t)-C(t) goes to 0.
  2. Stable: P(t)-C(t) oscillates, with the time spent as a positive value equal to the time spent as a negative value within a single oscillation period.
  3. Hungry: P(t)-C(t) either converges to a negative value, or oscillates, with more negative values than positive values in one oscillation period.
  4. Starved: P(t)-C(t) diverges, approaching negative infinity.
  5. Saturated: P(t)-C(t) either converges to a positive value, or oscillates, with more positive values than negative values in one oscillation period.
  6. Oversaturated: P(t)-C(t) diverges, approaching positive infinity.


  • The expected output will always be one of the above cases - you don't have to worry about other cases
  • P(t) and C(t) will always be real-valued continuous functions that only consist of the following operations: addition, subtraction, multiplication, division, exponentiation (for constants a and b and parameter t, a^b, t^a, and a^t are all allowed), principal roots (real roots, only when they exist, and only the positive root when there are 2 real roots), logarithms, and basic trigonometric functions (sine and cosine of arbitrary values, and tangent for values that are not odd multiples of pi/2)
  • Input may be in any reasonably format (string, function object/pointer, mathematical function object, etc.)
  • You may use any unambiguous set of 6 values for output (like [1, 2, 3, 4, 5, 6], [0, 1, 2, 3, 4, 5], [Ideal, Stable, Hungry, Starved, Saturated, Oversaturated], or [0, ~, -, -∞, +, +∞])

Test Cases


Sandbox Notes

  • Are the definitions and rules clear?
  • Are there any I/O methods that should be explicitly allowed or forbidden that aren't covered by the current rules?
  • \$\begingroup\$ I suspect that this is extraordinarily difficult and possibly formally undecidable. See Richardson's Theorem. \$\endgroup\$
    – xnor
    Oct 20 '16 at 4:22
  • \$\begingroup\$ @xnor Richardson's Theorem only applies to finding zeroes of functions and deciding if two functions are equal everywhere. It doesn't have anything to do with asymptotic behavior. As for the challenge being extraordinarily difficult - it's not meant to be an easy challenge. \$\endgroup\$
    – user45941
    Oct 20 '16 at 4:55
  • \$\begingroup\$ Actually, I can prove it's undecidable. Let A be an elementary function for which you want to decide whether it is everywhere 0, in contradiction of Richardson's theorem. Then, A(e^t * sin(t)) goes to 0 as t->∞ if and only A is identically 0. This is because any x for which A(x) is nonzero would correspond to an infinite number of t with x=e^t*sin(t), causing the limit to be nonzero. So, by running an algorithm for your problem and checking whether it outputs (1) or not, we'd be able to solve an undecidable problem. \$\endgroup\$
    – xnor
    Oct 20 '16 at 5:18
  • \$\begingroup\$ @xnor Good catch, I didn't consider that. Would it be possible to limit the types of functions such that it would be deciable, and all of the cases would be possible? \$\endgroup\$
    – user45941
    Oct 20 '16 at 5:44
  • \$\begingroup\$ I don't think so. Without sin or cos, you can't get oscillation, but that's basically enough for the undecidability to apply. \$\endgroup\$
    – xnor
    Oct 20 '16 at 5:46
  • \$\begingroup\$ @xnor Shame. I think I'll rewrite this using discrete functions, in a manner so it is decidable. \$\endgroup\$
    – user45941
    Oct 20 '16 at 5:48

(Need a title.)

Your task is to write 3 programs (or functions) that check if the input is a string of balanced brackets satisfying all the following criteria:

  • The number of []s is always less than or equal to the number of ()s surrounding each position.
  • The number of {}s is always less than or equal to the sum of the numbers of ()s and []s surrounding each position.
  • The number of <>s is always less than or equal to the product of the number of () [] and {}s surrounding each position.

You can assume that:

  • There are only the characters ()[] in the input of your program 1.
  • There are only the characters ()[]{} in the input of your program 2.
  • There are only the characters ()[]{}<> in the input of your program 3.

Your score is the product of (the length of each program + 1). Lowest score wins.


(To do.)

  • 1
    \$\begingroup\$ I don't understand the first tie-breaker. What is a "program number"? \$\endgroup\$ Oct 19 '16 at 14:13
  • \$\begingroup\$ @PeterTaylor: You can choose to make up to 3 programs. Program 1 handles 2 bracket-types, Program 2 handles 3 bracket-types and Program 3 handles 4 bracket-types. If person A has made programs 1 and 2 and person B has made programs 2 and 3. Person B wins the tie-breaker as his program numbers (3,2) are higher (and thus more advanced). \$\endgroup\$
    – Emigna
    Oct 20 '16 at 9:46
  • \$\begingroup\$ If Emigna is correct that this is a task where you can choose to not do the hardest two thirds of it, then that's stupid. I had assumed that the "up to three programs" was to allow people to write a single program to maximise code reuse between the three subtasks, not to allow skipping the subtasks entirely. \$\endgroup\$ Oct 20 '16 at 10:21
  • \$\begingroup\$ In any case, the first 2 tie-breakers is unlikely to see use as I'm sure someone will do all 3 and then only the program lengths will determine the score. \$\endgroup\$
    – Emigna
    Oct 20 '16 at 10:47
  • \$\begingroup\$ Yes, removed that part. \$\endgroup\$
    – jimmy23013
    Oct 21 '16 at 7:24

Quine tree

Inspired by that really weird quine challenge

For this challenge, you must construct a quine tree, which is a thing I made up, specifically an infinite binary tree.

How a quine tree works:

A quine tree has infinite nodes.

  • Each node is associated with a program

    • Let X, Y, Z, be the programming languages. This program is a polyglot in X and Y. In all but the root node, it is also a polyglot in Z. Note the

    • see below for more details

  • Each node has exactly two children

  • Every node but the root has a parent. Their parent has the node as a child (obviously).


  • When a node's program is run in X, it produces the program of the first child of that node.
  • When run in Y, it produces the program of the second child of that node.
  • The programs produced must be different (node foo's program, cannot have the same output in X and Y. This output must also be different than the nodes program).
  • If not the root node, when run in Z, it produces the parent of the node.


Imagine that we have the (fake) programming languages Hello, World, and Foobaz. Say the root node is $QQ$QQ$;:

  • when run in Hello, perhaps it produces $QQ$QQ$;;
    • When this is run with Hello again, perhaps produces $QQ$QQ$;;;
    • When this is run with World, perhaps produces $QQ$QQ$;;Q
    • Both these programs must produce, run in Foobaz, $QQ$QQ$;; (Output of root node run in Hello)
  • when run in World, perhaps it produces $QQ$QQ$;Q
    • When this is run with Hello, perhaps produces $QQ$QQ$;Q;
    • When this is run with World again, perhaps produces $QQ$QQ$;QQ
    • Both these programs must produce, run in Foobaz, $QQ$QQ$;Q (Output of root node run in World)
  • Both these programs, run in Foobaz, must produce $QQ$QQ$; (root program).
  • \$\begingroup\$ This seems like a really interesting challenge. I'm not sure how possible it is, though! Substitution languages like /// might be easiest to solve this in. \$\endgroup\$
    – wizzwizz4
    May 27 '17 at 8:20

Fastest Cashier in the West

You're trying to design an automatic cashier that is able to dispense the correct amount of change in any denominations as quickly as possible. Your goal is to implement a program that accepts a target integer T ≥ 0 which is the amount you're trying to make change for, and a set of positive integers {c1, c2, ..., cn} which are the values of each coin where you have an infinite number of each, and output all possible solutions for making change with those types of coins.

For example, if you're trying to make change for 123 and the coins available are {12, 16, 20, 27}, then there are 8 possible solutions

12 16 20 27  (123)
 0  1  4  1  means zero 12's, one 16, four 20's, one 27
             also 0*12 + 1*16 + 4*20 + 1*27 = 123
 0  6  0  1
 1  4  1  1
 2  2  2  1
 3  0  3  1
 4  3  0  1
 5  1  1  1
 8  0  0  1


The input will be formatted as


where T is the target value, N is the number of coins, and each cK is the value of coin k. Also, the coins will be listed in ascending order.

For example,


The output will be format as

x11 x12 ... x1N
x21 x22 ... x2N
xM1 xM2 ... xMN

where M is the number of solutions and N is the number of coins. Each line will contain the number of each coin needed to reach the target value separated by spaces. You may have trailing whitespace on each line.

For example, the output to the input above would be

0 1 4 1
0 6 0 1
1 4 1 1
2 2 2 1
3 0 3 1
4 3 0 1
5 1 1 1
8 0 0 1

You may output the solutions on each line in any order, but each solution must be unique.


  • This is so your only goal is to make your program as fast as possible.
  • Builtins or libraries that solve this problem directly are not allowed.
  • In the interests of having fine submissions, the deadline will be by December 12, 2016, which is almost two months.
  • Your program must be single-threaded. However, you are allowed to use SIMD instructions and any other features of the CPU.
  • The code will be ran on my machine which uses Fedora 24 x64 and has an i7-920 with 18 GB of memory.
  • Please provide compilation instructions and version information if your program requires it.
  • Please do not obfuscate your code unnecessarily. If I am suspicious of certain code and/or libraries, I reserve the right to ignore your submission.
  • This is optional but if you are using advanced mathematical strategies, commenting and explaining your program would be very much appreciated.


The test cases will be created and provided as a download if there is interest.

For scoring, there will be a hidden dataset that is ran when the deadline is reached for final scoring purposes. After this is complete, the hidden dataset will be released along with the winner.

For extra incentive, should I increase the best answer reward by including a bounty. Or maybe for the three fastest?

Is it okay to ban commercial products as I may not have access to some of them. It wouldn't be fair to allow Mathematica and not Matlab/Maple/etc.

@PeterTaylor: I have an implementation in Python that uses a memoized recursive function and I would say its performance is decent at best. I do agree that I/O could be a bottleneck but I could choose test cases that have few or no solutions. Using methods to calculate the Frobenius number would help find those, but I would still like to see a few cases with a large number of solutions. I see two options. First, I could require the output to be ordered so that results could be piped into another program and avoid hitting the disk while being verified. The second is to test each program twice, first with the output enabled to check the accuracy, and if accurate, the second run would be with the output disabled to measure the execution time.

  • \$\begingroup\$ Have you written and profiled a reference implementation? I would guess that for many inputs the bottleneck is going to be I/O rather than calculation, and that should be taken into account when timing. \$\endgroup\$ Oct 22 '16 at 18:34
  • \$\begingroup\$ @PeterTaylor My response was a bit long so I added it to the question. \$\endgroup\$
    – miles
    Oct 22 '16 at 18:55
  • \$\begingroup\$ Also, here is a reference implementation. \$\endgroup\$
    – miles
    Oct 22 '16 at 19:16
  • \$\begingroup\$ Please state the upper bounds for number of coins, cost, coin value, number of solutions... \$\endgroup\$
    – feersum
    Oct 23 '16 at 5:45
74 75
77 78

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .