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!


3624 Answers 3624

97 98
100 101

How low can you go?

Time to play so ascii-art limbo!

Here's the bar:

|  |
|  |

Can you fit under it?


Write a program or function that takes an ascii string representing a some shape, and a positive integer representing a bar height.

Output the shape from the input after it has attempted to do the limbo.


In limbo you lean back to make yourself as small as possible to fit under the bar, and that is just what the input shapes will try and do.

If the input shape contains any repeating patterns in its lines, then you can remove all but the last of the repetitions that are in the pattern. In additions the pattern must start at the top line, and once the repeating pattern is broken no more sections can be removed.

If there is a repeating pattern that contains another repeating pattern, only the innermost pattern is stripped.

For example this is how the following inputs would look after "Leaning back":

1. XXX              2. xxx            3. xxx           4. xxx      
   YYY                 xxx       xxx     xxx              xxx      xxx
   XXX   -->   XXX     yyy  -->  yyy     xxx  -->         yyy      yyy
   YYY         YYY     zzz       zzz     xxx              yyy      yyy
   zzz         zzz     aaa       aaa     xxx      xxx     xxx  --> xxx
                                                          xxx      xxx
                                                          yyy      yyy
                                                          yyy      yyy

Note how in example number 4 there was a repetition ox xxx insinde of another repetition of xxx, xxx, yyy, yyy. In this case only the inner repeating lines got reduced.

The bars will be drawn as shown below for the given heights:

1 -> |--|    2 -> |--|   3 -> |--|  etc...
                  |  |        |  |
                              |  |

If the given input shape in its reduced form does not fit underneath the bar then the bar will be drawn on the ground like this |__|



a. height: 3      b.  height: 2     c.  height: 1
   shape: xxx         shape: xxx        shape: (emptystring)
          xxx                xxx
          xxx                xxx
          yyy                yyy


a. |--|           b. |  | xxx       c. |--| 
   |  | xxx          |__| yyy  
   |  | yyy


  1. You can assume only valid inputs will be given, handle invalid input however you want
  2. The input shape will be drawn 1 space after the bar, and the bottom of the input shape will always line up with the bottom of the bar.
  3. Extra whitespace after is fine as long as all the lines are properly aligned.
  4. If there is whitespace in the given input then that should be included in the output
  5. The shape will not necessarily always line up in a perfect rectangle as I drew it.

This is for code-golf so the answer using the fewest bytes wins.

Please let me know what you think / if anything is unclear and needs to be improved. Hope you guys like it!

EDIT: Would whoever down voted please explain whats wrong with it?

  • \$\begingroup\$ I'm not the voter, but I'd bet they voted because what you have right now is very difficult to understand. I had to read this three times to figure out what you wanted. It may be worth revisiting the concept of "leaning over" as it is deeply unintuitive to me at the moment. On top of that, this requires a lot of "boring" golfing for the required output format when true/false seems to be basically the same. I hope this is helpful! \$\endgroup\$ Jan 20 '20 at 20:24
  • \$\begingroup\$ @FryAmTheEggman Hmm, I thought I'd explained it fairly clearly - its just removing any lines that form a repetitive pattern, but I will try to update it. The output is slightly more than just T/F because of wanting to see the "limbo'ed" input shape with the shaved lines, and having to draw the bar different if it fails... though maybe that is what you meant by boring golf? \$\endgroup\$
    – Quinn
    Jan 20 '20 at 20:30
  • \$\begingroup\$ @FryAmTheEggman I tried updating the explanation, though I'm not sure if it is any clearer, please let me know if it makes more sense now \$\endgroup\$
    – Quinn
    Jan 20 '20 at 20:33
  • \$\begingroup\$ Explaining things is often harder than one would guess. Here I think a big problem is that your definition of bending is not what I would expect, so it makes the whole idea harder to grasp - particularly the rules about which repetitions happen first. Since I don't get the why I struggle to get the what. Maybe try explaining this to someone verbally to see if you can get rapid and direct feedback. What you have now is better, but I still think I'd have a hard time following on a first read. \$\endgroup\$ Jan 20 '20 at 20:35

Factorise a floating point number

Given a target floating point number, \$T\$ and a set of \$N\$ floating point numbers \$\{x_{1},..,x_{N}\}\$ and a permissible error \$tol\$ , find a set of integer coefficients \$A\$, \$\{m_{1},..,m_{N}\}\$ such that: $$ A\prod\limits_{i=1}^{N}x_{i}^{m_{i}} = T\pm{}tol $$


Input will be a target number, a set of real numbers and a tolerance as a decimal or percentage (note which) in any order or format required by your language.


Output should be \$A\$, \$\{m\}\$ and the corresponding error as a percentage. If multiple combinations are valid any or all sets of \$A\$ and \$\{m\}\$ are within tolerance.

General rules

  • This is , so shortest answer in bytes wins.
    Don't let code-golf languages discourage you from posting answers with non-codegolfing languages. Try to come up with an as short as possible answer for 'any' programming language.
  • Standard rules apply for your answer with default I/O rules, so you are allowed to use STDIN/STDOUT, functions/method with the proper parameters and return-type, full programs. Your call.
  • Default Loopholes are forbidden.
  • If possible, please add a link with a test for your code (i.e. TIO).
  • Also, adding an explanation for your answer is highly recommended.



15857.6 [3.2, 7.1] 0.05 => 3, [4, 2]


145.85 [2.7182, 3.1415] 0.1 => 2, [2, 2]


0.3 [3.14159, 0.1] 0.05 => 3, [0, 1]

The periodic table and the chemical symbols

Periodic table

The periodic table is a large tablecitation needed where we can find the chemical elements written out with their chemical symbols. For example, Helium shows up as He and Carbon shows up as C. You can read Wikipedia's article on the periodic table, if you want.

Your task

You have to write a function/program/procedure/... that, given a chemical element's name, returns its chemical symbol. For the purposes of this challenge, we will use the 118 elements listed below:

['Hydrogen', 'Helium', 'Lithium', 'Beryllium', 'Boron', 'Carbon', 'Nitrogen', 'Oxygen', 'Fluorine', 'Neon', 'Sodium', 'Magnesium', 'Aluminium', 'Silicon', 'Phosphorus', 'Sulfur', 'Chlorine', 'Argon', 'Potassium', 'Calcium', 'Scandium', 'Titanium', 'Vanadium', 'Chromium', 'Manganese', 'Iron', 'Cobalt', 'Nickel', 'Copper', 'Zinc', 'Gallium', 'Germanium', 'Arsenic', 'Selenium', 'Bromine', 'Krypton', 'Rubidium', 'Strontium', 'Yttrium', 'Zirconium', 'Niobium', 'Molybdenum', 'Technetium', 'Ruthenium', 'Rhodium', 'Palladium', 'Silver', 'Cadmium', 'Indium', 'Tin', 'Antimony', 'Tellurium', 'Iodine', 'Xenon', 'Cesium', 'Barium', 'Lanthanum', 'Cerium', 'Praseodymium', 'Neodymium', 'Promethium', 'Samarium', 'Europium', 'Gadolinium', 'Terbium', 'Dysprosium', 'Holmium', 'Erbium', 'Thulium', 'Ytterbium', 'Lutetium', 'Hafnium', 'Tantalum', 'Tungsten', 'Rhenium', 'Osmium', 'Iridium', 'Platinum', 'Gold', 'Mercury', 'Thallium', 'Lead', 'Bismuth', 'Polonium', 'Astatine', 'Radon', 'Francium', 'Radium', 'Actinium', 'Thorium', 'Protactinium', 'Uranium', 'Neptunium', 'Plutonium', 'Americium', 'Curium', 'Berkelium', 'Californium', 'Einsteinium', 'Fermium', 'Mendelevium', 'Nobelium', 'Lawrencium', 'Rutherfordium', 'Dubnium', 'Seaborgium', 'Bohrium', 'Hassium', 'Meitnerium', 'Darmstadtium', 'Roentgenium', 'Copernicium', 'Nihonium', 'Flerovium', 'Moscovium', 'Livermorium', 'Tennessine', 'Oganesson']

with the respective element symbols:

['H', 'He', 'Li', 'Be', 'B', 'C', 'N', 'O', 'F', 'Ne', 'Na', 'Mg', 'Al', 'Si', 'P', 'S', 'Cl', 'Ar', 'K', 'Ca', 'Sc', 'Ti', 'V', 'Cr', 'Mn', 'Fe', 'Co', 'Ni', 'Cu', 'Zn', 'Ga', 'Ge', 'As', 'Se', 'Br', 'Kr', 'Rb', 'Sr', 'Y', 'Zr', 'Nb', 'Mo', 'Tc', 'Ru', 'Rh', 'Pd', 'Ag', 'Cd', 'In', 'Sn', 'Sb', 'Te', 'I', 'Xe', 'Cs', 'Ba', 'La', 'Ce', 'Pr', 'Nd', 'Pm', 'Sm', 'Eu', 'Gd', 'Tb', 'Dy', 'Ho', 'Er', 'Tm', 'Yb', 'Lu', 'Hf', 'Ta', 'W', 'Re', 'Os', 'Ir', 'Pt', 'Au', 'Hg', 'Tl', 'Pb', 'Bi', 'Po', 'At', 'Rn', 'Fr', 'Ra', 'Ac', 'Th', 'Pa', 'U', 'Np', 'Pu', 'Am', 'Cm', 'Bk', 'Cf', 'Es', 'Fm', 'Md', 'No', 'Lr', 'Rf', 'Db', 'Sg', 'Bh', 'Hs', 'Mt', 'Ds', 'Rg', 'Cn', 'Nh', 'Fl', 'Mc', 'Lv', 'Ts', 'Og']


The code you write should receive an element name in any sensible format, such as a string "helium". You may assume whatever capitalization that suits your needs.


The code you write should return the element's symbol as a string, with any capitalization that suits your needs. Bonus imaginary internet points if you return the symbol with the standard capitalization.


This is so your answer doesn't win by being the shortest! You will be provided 118 test cases. Your score will be your code's byte count divided by the percentage of test cases your code passes correctly. Lowest score wins!

E.g. my code has 1 byte and I get 1 test case correct. My score is \$1 / \frac{1}{118} = 118 \$. Someone else writes some code with 110 bytes but gets all the test cases correct. The other person scores \$ 110 / \frac{118}{118} = 110 \$, meaning the other person has a better score than me.

Test cases:

'Hydrogen' -> 'H'
'Helium' -> 'He'
'Lithium' -> 'Li'
'Beryllium' -> 'Be'
'Boron' -> 'B'
'Carbon' -> 'C'
'Nitrogen' -> 'N'
'Oxygen' -> 'O'
'Fluorine' -> 'F'
'Neon' -> 'Ne'
'Sodium' -> 'Na'
'Magnesium' -> 'Mg'
'Aluminium' -> 'Al'
'Silicon' -> 'Si'
'Phosphorus' -> 'P'
'Sulfur' -> 'S'
'Chlorine' -> 'Cl'
'Argon' -> 'Ar'
'Potassium' -> 'K'
'Calcium' -> 'Ca'
'Scandium' -> 'Sc'
'Titanium' -> 'Ti'
'Vanadium' -> 'V'
'Chromium' -> 'Cr'
'Manganese' -> 'Mn'
'Iron' -> 'Fe'
'Cobalt' -> 'Co'
'Nickel' -> 'Ni'
'Copper' -> 'Cu'
'Zinc' -> 'Zn'
'Gallium' -> 'Ga'
'Germanium' -> 'Ge'
'Arsenic' -> 'As'
'Selenium' -> 'Se'
'Bromine' -> 'Br'
'Krypton' -> 'Kr'
'Rubidium' -> 'Rb'
'Strontium' -> 'Sr'
'Yttrium' -> 'Y'
'Zirconium' -> 'Zr'
'Niobium' -> 'Nb'
'Molybdenum' -> 'Mo'
'Technetium' -> 'Tc'
'Ruthenium' -> 'Ru'
'Rhodium' -> 'Rh'
'Palladium' -> 'Pd'
'Silver' -> 'Ag'
'Cadmium' -> 'Cd'
'Indium' -> 'In'
'Tin' -> 'Sn'
'Antimony' -> 'Sb'
'Tellurium' -> 'Te'
'Iodine' -> 'I'
'Xenon' -> 'Xe'
'Cesium' -> 'Cs'
'Barium' -> 'Ba'
'Lanthanum' -> 'La'
'Cerium' -> 'Ce'
'Praseodymium' -> 'Pr'
'Neodymium' -> 'Nd'
'Promethium' -> 'Pm'
'Samarium' -> 'Sm'
'Europium' -> 'Eu'
'Gadolinium' -> 'Gd'
'Terbium' -> 'Tb'
'Dysprosium' -> 'Dy'
'Holmium' -> 'Ho'
'Erbium' -> 'Er'
'Thulium' -> 'Tm'
'Ytterbium' -> 'Yb'
'Lutetium' -> 'Lu'
'Hafnium' -> 'Hf'
'Tantalum' -> 'Ta'
'Tungsten' -> 'W'
'Rhenium' -> 'Re'
'Osmium' -> 'Os'
'Iridium' -> 'Ir'
'Platinum' -> 'Pt'
'Gold' -> 'Au'
'Mercury' -> 'Hg'
'Thallium' -> 'Tl'
'Lead' -> 'Pb'
'Bismuth' -> 'Bi'
'Polonium' -> 'Po'
'Astatine' -> 'At'
'Radon' -> 'Rn'
'Francium' -> 'Fr'
'Radium' -> 'Ra'
'Actinium' -> 'Ac'
'Thorium' -> 'Th'
'Protactinium' -> 'Pa'
'Uranium' -> 'U'
'Neptunium' -> 'Np'
'Plutonium' -> 'Pu'
'Americium' -> 'Am'
'Curium' -> 'Cm'
'Berkelium' -> 'Bk'
'Californium' -> 'Cf'
'Einsteinium' -> 'Es'
'Fermium' -> 'Fm'
'Mendelevium' -> 'Md'
'Nobelium' -> 'No'
'Lawrencium' -> 'Lr'
'Rutherfordium' -> 'Rf'
'Dubnium' -> 'Db'
'Seaborgium' -> 'Sg'
'Bohrium' -> 'Bh'
'Hassium' -> 'Hs'
'Meitnerium' -> 'Mt'
'Darmstadtium' -> 'Ds'
'Roentgenium' -> 'Rg'
'Copernicium' -> 'Cn'
'Nihonium' -> 'Nh'
'Flerovium' -> 'Fl'
'Moscovium' -> 'Mc'
'Livermorium' -> 'Lv'
'Tennessine' -> 'Ts'
'Oganesson' -> 'Og'

I used this code to shape the list into the test cases, might be useful to you.

Elements and symbols extracted from https://www.thoughtco.com/element-list-names-atomic-numbers-606529, visited at the 9th of February of 2020. At the time of writing, 118 elements were available. Source included "Aluminum" and "Aluminium" as alternatives, dropped "Aluminum" for the purposes of this challenge.

  • \$\begingroup\$ This is just a massive lookup table. I don't like it. \$\endgroup\$
    – S.S. Anne
    Feb 10 '20 at 0:50
  • \$\begingroup\$ "Aluminum" might be in the dictionaries of some languages. I suggest allowing either but and/or both in the solution and specifying which one was chosen. \$\endgroup\$
    – S.S. Anne
    Feb 10 '20 at 1:07
  • \$\begingroup\$ Dupe \$\endgroup\$
    – Jo King Mod
    Feb 10 '20 at 2:55
  • \$\begingroup\$ @JoKing this being a code challenge doesn't make it different from the dupe you linked? \$\endgroup\$
    – RGS
    Feb 10 '20 at 7:02
  • \$\begingroup\$ I think the problem you have is that your scoring strongly incentivises submitting H as the best answer. The other question had a shortest solution of 200 bytes - which can't get a better score than the 1 byte 1 answer program. \$\endgroup\$ Feb 10 '20 at 19:52
  • \$\begingroup\$ @FryAmTheEggman thanks for the feedback. Would you suggest tweaking the score or dropping this challenge? \$\endgroup\$
    – RGS
    Feb 10 '20 at 20:19
  • 2
    \$\begingroup\$ You might consider a similar idea but with something that hasn't been done as much. I think it depends on how you feel, it certainly isn't unsalvageable. \$\endgroup\$ Feb 10 '20 at 20:27
  • 1
    \$\begingroup\$ @FryAmTheEggman very nice idea. I am thinking airport codes, country codes or military rank abbreviations... \$\endgroup\$
    – RGS
    Feb 10 '20 at 21:21
  • \$\begingroup\$ I really like this idea, and I think allowing approximate answers makes it distinct enough from this that it's not a dupe. However, the current scoring encourages very inaccurate, trivial answers. The best I could find in 05AB1E is (05AB1E to output the first 2 characters of the input), which gets 45 test cases correct, scoring better than a 6 character answer that gets everything right. \$\endgroup\$
    – Grimmy
    Feb 14 '20 at 16:03
  • \$\begingroup\$ @Grimmy thanks for your very detailed feedback! Like I mentioned above, I'm exploring the possibility of using something other than chemical symbols. In your opinion, would you tweak the scoring or would you suggest something other than chemical symbols? \$\endgroup\$
    – RGS
    Feb 14 '20 at 16:20

Count switches in a smallest square root sequence mod \$2^n\$


Compress even permutations

Factorial number system

Every nonnegative integer can be encoded by factorial number system. Factorial number system doesn't have a fixed radix, but uses the factorial of nonnegative integers as radices. The place of \$0!\$ can have \$0\$ as the digit, the place of \$1!\$ can have \$0\$ or \$1\$ as the digit, the place of \$2!\$ can have \$0\$, \$1\$, or \$2\$ as the digit, and so on. Subscript exclamation mark denotes that the integer is encoded in factorial number system. For example:

$$ 41010_! = 4 \times 4! + 1 \times 3! + 0 \times 2! + 1 \times 1! + 0 \times 0!= 103_{10} $$


Numbers encoded by factorial number system has one-to-one correspondence to permutations:

$$ 41010_! \cong \begin{pmatrix} 0 \\ 0 \end{pmatrix} \begin{pmatrix} 1 & 0 \\ 0 & 1 \end{pmatrix} \begin{pmatrix} 2 & 1 & 0 \\ 2 & 1 & 0 \end{pmatrix} \begin{pmatrix} 3 & 2 & 1 & 0 \\ 0 & 3 & 2 & 1 \end{pmatrix} \begin{pmatrix} 4 & 3 & 2 & 1 & 0 \\ 3 & 2 & 1 & 0 & 4 \end{pmatrix} = \begin{pmatrix} 4 & 3 & 2 & 1 & 0 \\ 1 & 3 & 2 & 0 & 4 \end{pmatrix} $$

That is, each the place of \$n!\$ represents a right-direction rotation of \$n\$ through \$0\$, and its digit represents what \$0\$ becomes after the rotation.

It follows that the parity of the permutation is same as the parity of the sum of digits at the places of factorials of odd numbers. For \$41010_!\$, the permutation is even because \$1 + 1 = 2\$ is even.


An even permutation of 5 elements, or any equivalent object will be the input. That leaves \$5! \div 2 = 60\$ distinct permutations. Compress it to \$\lceil \log_2{60}\rceil=6\$ bits.


  • Input type and format doesn't matter. Possible choices of input format include (in C++):

    • std::map<int,int> scrambling the numbers from \$4\$ to \$0\$

    • int[5] containing the digits of the factorial number system

    • int[4] containing the digits of the factorial number system, except the place of \$0!\$ (it doesn't contribute to the permutation)

    • int encoded by the factorial number system

  • Output type and format doesn't matter either. It may be:

    • bool[6]

    • int8_t

    • a bit field

    • std::string containing ASCII digit 0s and 1s

  • Invalid inputs fall in don't care situation.


Simplify a directed graph


A connected directed graph, in any convenient format. A valid format (and probably the most convenient) would be a list of edges.


These two reductions are performed as often as possible. It does not suffice to apply one reduction first, then the other, as each of them can cause more of the other one.

Let \$E\$ be the set of all edges of the graph in the following explanations. For example \$E = \{(0, 1), (1, 2), (2, 0)\}\$ represents the following graph:

0 ----> 1
^       |
'-- 2 <-'

Let \$E_a\$, where \$a\$ is a vertex, indicate the set of edges involving \$a\$:

$$E_a := \{(u, v) \in E : u = a \vee v = a\}$$

Two graphs are isomorphic iff all vertices from one graph can be relabeled to make it equal to the other graph.


If there are disjoint sets of vertices \$M\$ and \$N\$ (\$M \cap N = \emptyset\$) not directly connected (\$((M, N) \cup (N, M)) \cap E = \emptyset\$) and the graphs with edges \$E \setminus E_M\$ and \$E \setminus E_N\$ are isomorphic, then the vertices \$M\$ (or \$N\$, but not both, are removed).

This means that of any two non-overlapping, non-directly-connected subgraphs that leave isomorphic graphs behind when removed, one is removed.

Simple example:

b --> 1
      ^  ==>  b --> 1

Relay removal

If there are vertices \$a\$ and \$b\$ and a non-empty set of vertices \$N\$, such that \$b \notin N\$ and \$E_b = \{(a, b)\} \cup (b, N)\$, then \$b\$ is removed from the graph and edges \$(a, N)\$ are added if they do not alreay exist.

This means that a vertex with exactly one edge ending there, at least one edge starting there and no self-loop can be removed, moving the starting edges to the start of the ending edge.

Simple example:

            1             1
            ^             ^
            |             |
O --> a --> b  ==>  O --> a ----.
      ^     |             ^     |
      |     v             |     v
      '---- 2             '---- 2


The output is the result of the reductions, in any convenient format. If identifiers are used for edges or vertices, as in a list of edges, these identifiers are not required to correspond to identifiers in the input.

A different format may be used for input and output.


This is : Lowest bytecount in each language wins. No answer will be accepted, as there is no overall winner.

Other rules

  • Standard loopholes apply
  • Functions or programs
  • Any input and output methods (STDIN, arguments, prompt(), ...)

Test cases

not yet


  • Everything clear?
  • Better explanations?
  • Better title?
  • I'm not sure whether output might depend on the order in which reductions are performed.

Let's Play Countdown!

(The Numbers round this time)

Countdown is a British TV game show composed of three different styles of rounds; the letters round, the conundrum, and the numbers round.

The conundrum could be solved with the same program you'd make for the Letters round, so let's tackle the third option that hasn't been done yet!


Take in a set of numbers. One of which is the "Target" number, and the rest are the building numbers.

The Countdown Number Round asks you to take the building numbers and to construct the Target number only using the four elementary operators. Every step must result in another strictly-positive integer (so non-perfect division is disallowed). Output the method to which you can construct the target number. If that's impossible, get as close as possible (above or below are scored the same). Numbers do not need to be used, but may NOT be reused.

Note - You will perform at most N-1 operations, where N is the number of building numbers. Every elementary operator takes in two inputs and provides one, so you "lose" one from your ranking every operation. That should give you an idea of the size of your output.

I/O is in any reasonable format, but target vs building numbers must be obviously distinct (either by the target being the first or the last number, or outside an array, or a different type, etc).

Output needs to explain exactly what operations are being performed on what numbers, and what the output for each operation would be, but can be done in whichever way seems reasonable.

Example I/O

In these examples, the first element is the target.

[888 100 2 75 3 1 10]

[766 22 10 8 3 1]
255x3=765 #You cannot get this one exactly, but one-off is close.

Sandbox Questions

I'm... like 85% sure this isn't anywhere here yet. I did a chunk of searching and couldn't find anything that fit the bill, so I think this is clear?

Any other neat examples you guys got?

  • 1
    \$\begingroup\$ As you say, I don't think we have exactly this challenge, but there are many similar ones. I doubt you will get anything besides brute forcing all possible arrangements then sorting them by nearness to the target. Separately, if there are multiple tying solutions you don't require that e.g. the shortest be output, correct? It is probably worth mentioning that. \$\endgroup\$ Feb 21 '20 at 21:24

Iterative Quadratics

Recently, in my algebra class, we proved that the following process always stops at some point, so I thought it would be a cool challenge!

Input: Two reals a,b.

Output Non-negative integer


Given two reals a,b, initialize a count variable c to 0, consider the quadratic equation


If this quadratic has real roots r,s (r<=s), increment the counter by 1, and replace a,b with r,s and repeat the process.

If the quadratic has complex roots, return c.

Test Cases

To be added.

  • \$\begingroup\$ you need to specify the precision required for stopping \$\endgroup\$
    – qwr
    Mar 11 '20 at 19:16
  • \$\begingroup\$ @qwr good catch. Any suggestions? \$\endgroup\$ Mar 11 '20 at 19:51

Doubly stochastic matrix

A doubly-stochastic matrix is a square matrix of non-negative real entries each of whose rows and columns sums to 1. Given a doubly-stochastic matrix, express it as a non-negative linear combination of permutation matrices, as is guaranteed to exist by the Birkhoff–von Neumann theorem.

TODO: Example, better explanation, test cases. If you want to develop and post this challenge, it's yours.

  • \$\begingroup\$ I liked this idea! May I work on this and give you credit? If I manage to do so before you, of course. \$\endgroup\$
    – RGS
    Feb 24 '20 at 19:33
  • \$\begingroup\$ @RGS You're very welcome to fully take it. \$\endgroup\$
    – xnor
    Feb 26 '20 at 2:01

Implement GF(2²)

Introduction to groups and fields

An additive group is a set with addition and negation defined. They must satisfy the following conditions:

  • \$0\$ is the additive identity.

  • Addition is associative.

  • For every \$x\$, the negation of \$x\$, \$-x\$ exists, and \$x + (-x) = (-x) + x = 0\$.

If addition is also commutative, the additive group is called abelian.

A field is an additive abelian group with multiplication and reciprocal defined. They must satisfy the following conditions:

  • \$1\$ is the multiplicative identity.

  • Multiplication is associative and commutative.

  • For every nonzero \$x\$, the reciprocal of \$x\$, \$x^{-1}\$ uniquely exists, and \$x\times x^{-1} = x^{-1}\times x = 1\$.

  • Multiplication distributes over addition.

Modular Arithmetic

For every positive integer \$n\$, you can define an additive abelian group as follows:

  • Define the set as integers from \$0\$ to \$n-1\$.

  • Define addition as usual addition with the result moduloed by \$n\$.

  • Define negation as usual negation with the result moduloed by \$n\$.

This group is denoted by \$ℤ_n\$. If \$n\$ is prime, multiplication can be analogously defined, making it a field. In particular, the operation tables of \$ℤ_2\$ are:

$$ \begin{array}{l|ll} + & 0 & 1 \\ \hline 0 & 0 & 1 \\ 1 & 1 & 0 \end{array} \begin{array}{l|ll} x & -x \\ \hline 0 & 0 \\ 1 & 1 \end{array} \begin{array}{l|ll} × & 0 & 1 \\ \hline 0 & 0 & 0 \\ 1 & 0 & 1 \end{array} \begin{array}{l|ll} x & x^{-1} \\ \hline 0 & \text{NaN} \\ 1 & 1 \end{array} $$

Galois Field GF(2²)

A Galois field \$\text{GF}(p^k)\$ emerges when one takes the set as polynomials over \$ℤ_p\$, and defines addition and multiplication as the usual operation with polynomial modulo, where the modding polynomial is irreducible and has degree of \$k\$. Since \$x^2+x+1\$ is an (in fact, the only) irreducible polynomial over \$ℤ_2\$ that has degree \$2\$, this results in \$\text{GF}(2^2)\$. Its operation tables are:

$$ \begin{array}{l|ll} + & 0 & 1 & x & x+1 \\ \hline 0 & 0 & 1 & x & x+1 \\ 1 & 1 & 0 & x+1 & x \\ x & x & x+1 & 0 & 1 \\ x+1 & x+1 & x & 1 & 0 \end{array} \begin{array}{l|ll} f(x) & -f(x) \\ \hline 0 & 0 \\ 1 & 1 \\ x & x \\ x+1 & x+1 \end{array} \\ \begin{array}{l|ll} × & 0 & 1 & x & x+1 \\ \hline 0 & 0 & 0 & 0 & 0 \\ 1 & 0 & 1 & x & x+1 \\ x & 0 & x & x+1 & 1 \\ x+1 & 0 & x+1 & 1 & x \end{array} \begin{array}{l|ll} f(x) & f(x)^{-1} \\ \hline 0 & \text{NaN} \\ 1 & 1 \\ x & x+1 \\ x+1 & x \end{array} $$

Your task is to implement the set and the operations. As a conseuqence, you must have:

  • The members of the set defined as constants (2-bit bitstring, an ASCII digit, or whatever). This won't contribute to the score.

  • Four codes that defines each operations, whose input(s) is/are as defined above.


  • Though defined as polynomials, the type and format of the inputs doesn't matter. You must have the same type for every input.

  • The type and format of the outputs doesn't matter either, but it must be the same as the input(s).

  • The reciprocal of \$0\$ must result in an "error" condition. This includes returning an errornous value, throwing an error, or terminating the program. It must halt.

  • Other invalid inputs fall in don't care situation.

  • Since there are multiple codes, the score for code golf is alloted by the sum of their lengths in bytes.

  • 1
    \$\begingroup\$ Related: GF(2^8) and GF(3^2), though neither asks for negation or multiplicative inverse. Also, A code that defines the set as a type can be problematic in many languages where the concept of "type" is not well-defined. I'd suggest to exclude it from the code size and just ask the answerer to provide the four values corresponding to \$0, 1, x, x+1\$ respectively. \$\endgroup\$
    – Bubbler
    Feb 24 '20 at 0:02
  • \$\begingroup\$ @DonThousand Requiring the uniqueness of a negation resolves the issue. Also, it would take too long to explain the "actual" definition of \$\mathbb{Z}_p\$. So I defined it with usual modulo. \$\endgroup\$ Feb 24 '20 at 1:25
  • \$\begingroup\$ @DonThousand \$(-a) + a = 0\$ implies \$(-(-a)) + (-a) = 0\$. So both \$a\$ and \$-(-a)\$ are additive inverses of \$-a\$. By uniqueness of additive inverse, it follows \$-(-a) = a\$. \$\endgroup\$ Feb 24 '20 at 1:57
  • \$\begingroup\$ @DannyuNDos Incorrect, you assume commutativity again. $a$ is a right inverse, and $-(-a)$ is a left inverse of $-a$ \$\endgroup\$ Feb 24 '20 at 2:03
  • \$\begingroup\$ @DonThousand The definition of group states that a left inverse must be the right inverse, and a right inverse must be the left inverse. Such inverse always uniquely exists. \$\endgroup\$ Feb 24 '20 at 3:21
  • \$\begingroup\$ @DannyuNDos That's my point. Your definition doesn't say that the right inverse = left inverse. \$\endgroup\$ Feb 24 '20 at 3:30
  • \$\begingroup\$ @DonThousand Oh my! So there was the flaw. I only remembered the definition, but not memorized it. Thanks anyways. \$\endgroup\$ Feb 24 '20 at 3:32
  • \$\begingroup\$ @DannyuNDos Yea, lol. Sheesh, that was a journey. \$\endgroup\$ Feb 24 '20 at 3:36
  • \$\begingroup\$ @DonThousand You only need a right zero and right inverses, then we get left zero and left inverses, and they are unique (we also don't have to demand that). Proof that right inverses are left inverses: \$(-a)+a=(-a)+a+0=(-a)+a+(((-a)+a)+(-((-a)+a)))=(-a)+(a+(-a))+a+(-((-a)+a))=((-a)+0)+a+(-((-a)+a))=((-a)+a)+(-((-a)+a))=0\$. Using this gives left zero: \$0+a=(a+(-a))+a=a+((-a)+a)=a+0=a\$. \$\endgroup\$ Feb 24 '20 at 15:07

Shared Letters in consecutive numbers

Inspired by this puzzling question.

It turns out that, in English, every pair of consecutive integers (e.g. 0,1, 1,2, etc.) shares at least one letter when spelled out (e.g. zErO, OnE (or NOught, ONe); One, twO, etc).


Any two non-negative integers (all natural numbers including 0) up to and including one googol. These can be input as any type you choose, but string representations must only use the characters 0123456789.,' (i.e. the numbers must not already be spelled out on input, but rather input as a numeral).

You can assume that the two numbers will be consecutive.

Some examples of valid inputs:

"123,245", "123,246"
"123.456", "123'457"

some examples of invalid inputs


The Challenge

Given the two inputs, output all shared characters when spelled (both numbers spelled in either lower or upper-case, the same case for both numbers).

A sample implementation for spelling numbers can be found here: https://stackoverflow.com/a/3911982/318414; but I'm sure other options exist; and there are certainly efficiencies to be found given that there are large amounts of shared strings, once you get into the higher numbers. See also https://simple.wikipedia.org/wiki/Names_for_large_numbers for the names of large numbers.

, usual exclusions apply

Output The shared letters, in any reasonable format. Any of the three numbering systems on the Wikipedia page are valid.


I will be assuming British English (long form) in my examples.

{6,7} -> "s" or "S" (six, seven)

{999,1000} -> {"n","e"," ","t","h","u","a","d"} (nine hundred and ninety nine, one thousand)

{88955,88956} -> `` (eighty eight thousand nine hundred and fifty five, eighty eight thousand nine hundred and fifty six)

1000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000,1000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001 -> "ONE THUSADQICL" (ONE THOUSAND QUINDECILLION, ONE THOUSAND QUINDECILLION AND ONE)

9999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999, 10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 -> el no (nine hundred and ninety nine thousand sexdecillion **nine hundred and ninety nine thousand quindecillion nine hundred and ninety nine thousand quatturodecillion nine hundred and ninety nine thousand tredecillion nine hundred and ninety nine thousand duodecillion nine hundred and ninety nine thousand undecillion nine hundred and ninety nine thousand decillion nine hundred and ninety nine thousand nonillion nine hundred and ninety nine thousand octillion nine hundred and ninety nine thousand septillion nine hundred and ninety nine thousand sextillion nine hundred and ninety nine thousand quintillion nine hundred and ninety nine thousand quadrillion nine hundred and ninety nine thousand trillion nine hundred and ninety nine thousand billion nine hundred and ninety nine thousand million nine hundred and ninety nine thousand nine hundred and ninety nine; one googol)

  • \$\begingroup\$ "Any two positive integers (all natural numbers including 0)" You should reword this because 0 is not a positive integer. \$\endgroup\$
    – 79037662
    Feb 22 '20 at 4:34
  • \$\begingroup\$ Possibly a chameleon challenge because answerers are required to convert an integer to its English form first, which could take much more effort than finding the shared letters. \$\endgroup\$ Feb 29 '20 at 3:33
  • \$\begingroup\$ @ShieruAsakoto I'm envisaging that the intermediate step won't necessarily need to be performed in code - perhaps there's a way to just look at the numbers? If not, does a challenge exist for converting numbers to words yet? \$\endgroup\$ Mar 2 '20 at 8:17

Find spies in a multilingual csv


You are an NSA undercover agent in a Middle-Eastern HR company, you just received a list of people with their jobs in many different languages. Some are spies and you need to know who. Your mission, if you accept it, is to get what are people working in, and relate it to a list of job categories. Most dangerous elements are those working in Law enforcement and security. However as your code will be part of a bigger file it needs to be as short in lines as possible for stealthness. This comes from expectations I encountered in the administration to keep some mystery behind code, if ever it were to be stolen.

  • This challenges your way to handle loops, map, reduce and filter, destructuring or unpacking an array/csv/df. Last but not least it allows you to get into the fascinating world of cross-language nlp.

I provide an example in python (64 lines of code)


The challenge is to get, in the shortest amount of bytes (libraries not included) the most similar element in an array of string with another string, and this for each line of a csv taken as input.

  • Inputs:

    1. X.csv a csv/dataframe of actual jobs that look like this one:
5,موجه تربوي 
6,Réalisateur film cartoon
7,إإطار مالي
8,موضف إطار
9,مهندس بمكتب دراسات
10,باحثة  _ كاتبة _ 
12,Chef de projet
15,Professeur Universitaire
16,cadre supérieur
20,Chef de projet
21,مدير  شركة
23,cadre à Maroc Telecom
25,Consultant en Immobilier
27,اجير أو عامل
  1. df.csv job categories that must include all the following categories:

    ['Agriculture, farming and environment',
       'Accountancy, banking and finance',
       'Teacher training and education', 'Leisure, sport and tourism',
       'Transport and logistics', 'Information technology',
       'Hospitality and events management',
       'Business, consulting and management', 'Creative arts and design',
       'Trade', 'Law enforcement and security',
       'Property and construction', 'Law',
       'Engineering and manufacturing', 'Social care',
       'Charity and voluntary work', 'Sales',
       'Public services and administration', 'Other. Please specify:',
       'Healthcare', 'Energy and utilities',
       'Marketing, advertising and PR', 'Media and internet',
       'Recruitment and HR', 'Science and pharmaceuticals']
    • Output would be the column in X.csv plus a new column, the most similar job. The most accurate results are:
0,Entrepreneur,Public services and administration
1,طالبة,Teacher training and education
2,ETUDIANT,Teacher training and education
3,ETUDIANT ,Teacher training and education
4,موظف,Recruitment and HR
5,موجه تربوي ,Teacher training and education
6,Réalisateur film cartoon,Creative arts and design
7,إإطار مالي,"Accountancy, banking and finance"
8,موضف إطار,Trade
9,مهندس بمكتب دراسات,Engineering and manufacturing
10,باحثة  _ كاتبة _ ,Recruitment and HR
11,طالب,Teacher training and education
12,Chef de projet,Creative arts and design
13,ASSUREUR,Business, consulting and management
14,FONCTIONNAIRE D'ÉTAT,Public services and administration
15,Professeur Universitaire,Teacher training and education
16,cadre supérieur,Business, consulting and management
17,fonctionnaire,Public services and administration
18,CDB Retraite,Recruitment and HR
19,professeur,Teacher training and education
20,Chef de projet,Public services and administration
21,مدير  شركة,Recruitment and HR
23,cadre à Maroc Telecom,Media and internet
24,Employé ,Sales
25,Consultant en Immobilier,Property and construction
26,fonctionnaire,Public services and administration
27,اجير أو عامل,Recruitment and HR
  • Inputs should be tested against all categories.

The winner of the challenge will be the one with the most accurate results. If on the test set. If several are as accurate, the shortest amount of bytes will be the winner. You can use any methods to get the most similar item. The state of the art method seems to be according to Google Multilingual Universal Sentence encoder. I provide an attempt with the code below but you will see it is not quite acccurate.

Example in Python

Double agent: A spy who works for two countries, and sometimes even three, in which case he is definitely a trouble. - Mots et Grumots (2003), Marc Escayrol

#@title Setup common imports and functions
import numpy as np
import os
import pandas as pd
import tensorflow.compat.v2 as tf
import tensorflow_hub as hub
from tensorflow_text import SentencepieceTokenizer
import sklearn.metrics.pairwise

from simpleneighbors import SimpleNeighbors
from tqdm import tqdm
from tqdm import trange

import json

def most_similar(embeddings_1, embeddings_2, labels_1, labels_2):

  assert (len(embeddings_1) == len(labels_1) and len(embeddings_2) == len(labels_2))

  # arccos based text similarity (Yang et al. 2019; Cer et al. 2019)
  sim = 1 - np.arccos(sklearn.metrics.pairwise.cosine_similarity(embeddings_1, embeddings_2))/np.pi

  embeddings_1_col, embeddings_2_col, sim_col = [], [], []
  for i in range(len(embeddings_1)):
    for j in range(len(embeddings_2)):
  df = pd.DataFrame(zip(embeddings_1_col, embeddings_2_col, sim_col),
                    columns=['embeddings_1', 'embeddings_2', 'sim'])

  # return the higest similarity one
  category = df['embeddings_1'].iloc[df['sim'].argmax()]
  return category

def main():

    X = pd.read_csv('X.csv')
    y = pd.read_csv('y.csv')
    df_rni = pd.read_csv('df.csv')

    # The 16-language multilingual module is the default but feel free
    # to pick others from the list and compare the results.
    module_url = 'https://tfhub.dev/google/universal-sentence-encoder-multilingual/3' #@param ['https://tfhub.dev/google/universal-sentence-encoder-multilingual/3', 'https://tfhub.dev/google/universal-sentence-encoder-multilingual-large/3']

    model = hub.load(module_url)

    def embed_text(input):
        return model(input)

    def compute_similarity(references, target):
        # I want to create as many rows as there are references and fill them with the results
        # arccos based text similarity (Yang et al. 2019; Cer et al. 2019)
        for row in target.iterrows():
            for reference in references:
                sim = 1 - np.arccos(
                result = sklearn.metrics.pairwise.cosine_similarity(row,

            # place the result in the column "reference"

    # get unique job categories and job of people
    job_categories = X.S02Q11_Professional_field.unique()
    # turn them to list
    job_categories = job_categories.tolist()
    # emebedding job categories 
    references_result = embed_text(job_categories[1:])

    for _, row in df_rni.iterrows():
        actual_job = row['new_professionactuelle']
        # check for nan that can't be embedded
        if str(actual_job) != 'nan':
            # embedding actual job
            target_result = embed_text(actual_job)
            # visualize similarity
            category = most_similar(references_result, target_result, job_categories[1:], [actual_job])
        else: category = None

if __name__ == "__main__":
  • \$\begingroup\$ @Arnauld Indeed, fixed, thanks! \$\endgroup\$ Feb 24 '20 at 15:19
  • \$\begingroup\$ It looks like your reference code fetches output from an external source, which is forbidden by default. Please detail within the question how one would implement the Google Universal Sentence Encoder. (If the challenge actually requires querying tfhub.dev/google/universal-sentence-encoder-multilingual/3, I don't think that's gonna be appropriate for CGCC). \$\endgroup\$
    – Grimmy
    Feb 24 '20 at 15:55
  • \$\begingroup\$ @Grimmy Google Universal Encoder: that's not compulsory, but I find it very handy and a good tip to start. External source: yes, that's the way I provided example data like the inputs above. Should I hard code the sources? \$\endgroup\$ Feb 24 '20 at 16:05
  • \$\begingroup\$ Wait, Google Universal Encoder isn't compulsory? That seems to contradict the most similar element in an array of string with another string according to Google Multilingual Universal Sentence encoder. If GMUSE isn't required, this sentence should be replaced by a proper definition of "most similar element", and GMUSE should only be mentioned in the footnotes. \$\endgroup\$
    – Grimmy
    Feb 24 '20 at 16:10
  • \$\begingroup\$ External source: yes, that's the way I provided example data like the inputs above. I'm confused. Do you use tfhub.dev/google/universal-sentence-encoder-multilingual/3 only to get example data? It sure doesn't look like that in your code. \$\endgroup\$
    – Grimmy
    Feb 24 '20 at 16:11
  • \$\begingroup\$ @Grimmy Porbably I misunderstood "It looks like your reference code fetches output from an external source" then, my apologizes. \$\endgroup\$ Feb 24 '20 at 16:19
  • \$\begingroup\$ This needs an explanation of how "accuracy" is computed. (And you still need to define "most similar"). \$\endgroup\$
    – Grimmy
    Feb 24 '20 at 18:45

Auto Tic Tac Toe

Okay, so after thinking about the comment I think I thought of a way to make it more interesting.


Given no input, write a program or function which outputs an entire game of Tic-Tac-Toe where X always wins, or the game ends in a tie.


  • X goes first
  • O must make moves at random
  • X must make smart moves such that it always wins the game, or the game ends in a tie


Here's what I would expect a game to look like:







  • X does not need to win in the fewest moves, it is enough to just make it always block O from winning
  • You can output the game in whatever form you like, as long as it is easily conveys every turn of the game. For example you could output a string like above, or a list of lists of ints like below, where 0 is an empty space and 1, 2 are X, O respectively:
  [1,0,0, 0,0,0, 0,0,0], 
  [1,2,0, 0,0,0, 0,0,0], 
  [1,2,0, 0,1,0, 0,0,0],
  [1,2,2, 0,1,0, 0,0,0], 
  [1,2,2, 0,1,0, 0,0,1]

This is code golf, answer in the fewest bytes wins. Standard rules apply.

Is this a better challenge? I'd love to know what people think

Working example of ungolfed code: Try it online!

  • 1
    \$\begingroup\$ I suspect this hasn't been asked because the huge majority of the work is spent on IO and not on an interesting problem. If you want to ask a tic-tac-toe challenge, you might be better served asking something like "can the next move win" which might still be a lot of parsing, but prevents excessive output at least. That said, you did do a good job of alleviating these problems in your notes, so you may be fine - most of what I've written here is my opinion, not precise advice. \$\endgroup\$ Jan 26 '20 at 19:31
  • \$\begingroup\$ @FryAmTheEggman Made a pretty big edit, do you think this would be less boring? \$\endgroup\$
    – Quinn
    Feb 1 '20 at 1:15
  • \$\begingroup\$ I think the problem I have with this version is that it is probably more work to make a tic-tac-toe AI than it is to encode all the possible games and just pick one at random to output. It is possible that that isn't true, I haven't tried yet, but it still feels a bit tedious. But this might be the right direction - perhaps instead just ask for one random valid final tic-tac-toe board? Then it will likely be an encoding problem, but perhaps one with interesting strategies. Again, all opinion here, the challenge is written well, etc. \$\endgroup\$ Feb 1 '20 at 5:13
  • \$\begingroup\$ Do you mean "uniformly distributed" when you write "random"? \$\endgroup\$ Feb 4 '20 at 9:54
  • \$\begingroup\$ Is the challenge in its current form not a kolmogorov-complexity challenge with choice? \$\endgroup\$ Feb 25 '20 at 1:25
  • \$\begingroup\$ @FryAmTheEggman I don't think making the AI is actually that hard for tic-tac-toe, though if im underestimating it, i could make a change that X always go first, which I think would make the ai pretty trivial \$\endgroup\$
    – Quinn
    Feb 25 '20 at 14:14
  • \$\begingroup\$ @JonathanFrech You tell me i have no clue, is that meaning the amount of code to output all possible tictactoe outcomes? \$\endgroup\$
    – Quinn
    Feb 25 '20 at 14:15
  • \$\begingroup\$ I am just saying that if you take no input and require a (semi-)static output, a part of the challenge is to find out which tic-tac-toe game requires the least bytes to represent and the rest is a kolmogorov-complexity task, which in my opinion is s slightly over-used challenge format. \$\endgroup\$ Feb 25 '20 at 15:59
  • \$\begingroup\$ kolmogorov-complexity challenges are code golf challenges with no input and a static output. \$\endgroup\$ Feb 25 '20 at 15:59
  • \$\begingroup\$ @JonathanFrech Is that a bad thing then? Do you think this challenge wouldn't be fun? \$\endgroup\$
    – Quinn
    Feb 25 '20 at 20:18
  • \$\begingroup\$ @FryAmTheEggman okay fixed my example, hopefully that illustrates how simple the ai could be \$\endgroup\$
    – Quinn
    Feb 25 '20 at 20:32
  • 1
    \$\begingroup\$ I feel as though your example does more to demonstrate that nobody would approach this by writing an AI - they would just encode each possible game as Jonathan is suggesting. That doesn't mean it isn't a good challenge, the problem I am trying to get at is that the phrasing of the challenge implies that writing a "player" is required - which I think is a bad requirement. If the challenge was just "output a random, valid, tic-tac-toe board" you could still maybe get an AI solution, if it happened to be shorter, but wouldn't come with a lot of needless baggage. Sorry if this sounds a bit rambly. \$\endgroup\$ Feb 25 '20 at 20:39
  • \$\begingroup\$ @FryAmTheEggman well my example has tons of unnecessary baggage and is not doing anything in an efficient way in terms of golfing, i just wanted the general algorithm to be shown, id imagine it could easily be shortened to a couple hundred bytes in most languages. That said if they can find a way to output winning board states so long as they show each turn that was played than that would be valid, I don't intend to require anyone to write an AI, just achieve the desired output. Not sure how to rephrase the question to make sure its clear that that is a valid answer. \$\endgroup\$
    – Quinn
    Feb 25 '20 at 21:15
  • \$\begingroup\$ @Quinn However, finding such a game would not be done in the submission but to be able to write the submission, leaving the actual challenge itself to be a bit of a boilerplate. \$\endgroup\$ Feb 25 '20 at 21:19
  • \$\begingroup\$ I guess I've posted one, exactly same but ask for O's input \$\endgroup\$
    – l4m2
    May 14 '20 at 7:30

Compute the Pareto frontier

Given a set of triplets, output its Pareto frontier.


A triplet is a list of 3 positive integers, for example [120, 15, 21] (order matters).

A triplet [A, B, C] is objectively worse than [a, b, c] when A >= a, B >= b, and C >= c (lower is better).

A triplet is on the Pareto frontier when it's not objectively worse than any other triplet in the input.

I/O rules

Input and output are both a set of triplets. Each triplet must be represented as either a list of integers ([1, 2, 3]) or a /-delimited string ("1/2/3"). The format of the outer sets is flexible (built-in set type, list, or delimited string are all okay).

Test cases

[[1, 1, 1], [2, 2, 2]] => [[1, 1, 1]]
[[3, 3, 1], [3, 1, 3], [1, 3, 3], [2, 2, 2]] => [[3, 3, 1], [3, 1, 3], [1, 3, 3], [2, 2, 2]]
... (more to come) ...
  • \$\begingroup\$ I had sandboxed a similar challenge, but it didn't seem to be going anywhere and I wasn't going to post it. Maybe the discussion on it about filtering is somewhat relevant to this. \$\endgroup\$
    – xnor
    Mar 8 '20 at 18:42
  • \$\begingroup\$ The IO rule for the triplets seems odd, why not any kind of ordered list? \$\endgroup\$ Mar 8 '20 at 19:35

A similar challenge was posted here but that's outdated and I have a few twists.


Write a program, expression or subroutine which, given an arithmetical expression in infix notation, like 1 + 2, outputs the same expression in postfix notation except the numbers are now float representative in the string, i.e. 1.0 2.0 +.

The input can include parentheses (()), exponents (^), division (/) and multiplication (*), addition (+) and subtraction (-) (in that order of operation), such as

4 ^ (2 / 3) * 9 * 3 - - 4 * 6

output the same expression in prefix notation.

4.0 2.0 3.0 / ^ 9.0 * 3.0 * -4.0 6.0 * -

Spaces are optional in the input as well as the output.


  • Must support using floats too, instead of just integers. so 4 ^ (2.0/3.0) * 9.0 * 3.0 - - 4 *6 output's is the same as the ones above.
  • Must return a string with all numbers as its float representative instead of as its original form, so 9 -> 9.0 and if it was 9.0 it stays 9.0.
  • Must support negation (see example above "3.0 - - 4" ends up with "-4.0", negative sign stays in place).
  • Cannot use exec() or eval() functions.


  • You can assume that there will not be any special numbers such as those in scientific notation or those with hanging zeroes, e.g. 000004 or 0005.000000.
  • You should also not assume the commutative or associative properties. This means that, while the operators will move around, the numbers will always remain in the same order.
  • You can always assume a valid infix input.


  • You should not evaluate any expression.
  • The output should not contain any unneeded parentheses. ((2+1))-1 should reduce to 2+1-1.

Game Winning Criteria

Fewest amount of characters wins. Bytes are not of a matter here.

  • 2
    \$\begingroup\$ To reiterate my comment on the original question: 1. Using the word "twist" is not recommended. 2. Don't ban exec/eval just because you don't like it. The task isn't about evaluating the value, so it is unnecessary and pretty much arbitrary. 3. You say "prefix notation", but 4.0 2.0 3.0 / ^ 9.0 * 3.0 * -4.0 6.0 * - is in postfix, not prefix. You should fix either the description or the example result. \$\endgroup\$
    – Bubbler
    Mar 12 '20 at 1:27
  • 1
    \$\begingroup\$ And you didn't explain why you want to handle unary negation in a different way (which is essentially changing the underlying expression, which seems against "You should not evaluate any expression"). By using a different symbol for unary minus (say ~), it is possible to translate the example to postfix as 4.0 2.0 3.0 / ^ 9.0 * 3.0 * 4.0 ~ 6.0 * -. \$\endgroup\$
    – Bubbler
    Mar 12 '20 at 1:34
  • 2
    \$\begingroup\$ For code-golf scoring, scoring by bytes is preferred over characters. (I believe there must be a more persuasive argument somewhere...) \$\endgroup\$
    – Bubbler
    Mar 12 '20 at 1:47
  • 1
    \$\begingroup\$ The second line in "clarifications" is unnecessary because "postfix notation" implies no parentheses. For the floating-point output, do you have a reason to demand adding .0 to integer values? A floating-point number 4 can be printed as 4, 4., or 4.0, and all of them represent the same value anyway. \$\endgroup\$
    – Bubbler
    Mar 12 '20 at 2:13
  • \$\begingroup\$ The linked challenge explicitly assumes left-associativity for all operators ^ * / + - (that is, 1 - 2 + 3 = (1 - 2) + 3, 4 / 5 * 6 = (4 / 5) * 6, and 7 ^ 8 ^ 9 = (7 ^ 8) ^ 9), but ^ is mathematically right-associative (7 ^ 8 ^ 9 = 7 ^ (8 ^ 9) != (7 ^ 8) ^ 9). Which one should we use? \$\endgroup\$
    – Bubbler
    Mar 12 '20 at 2:17
  • \$\begingroup\$ Can negation be stacked? i.e. is 1 - - - 1 a valid input? Doesn't negation in the output kinda negate (heh) the point of it being postfix, because it it prefix? \$\endgroup\$
    – Jo King Mod
    Mar 12 '20 at 2:33

Dilapidated art gallery problem

In a typical art gallery problem, the objective is to place as few guards as necessary inside an arbitrary polygon so that all of it is visible by some guard. This time, we'll make some changes to simplify the task:

  • The gallery can only afford to pay one guard. No more. This means that it won't be possible to keep everything in sight, so the objective becomes to maximize the amount of art visible.
  • Not even a dilapidated art gallery is barbaric enough to put art on the floor, and they don't have enough money to afford pedestals, so we only care about how much wall area is visible.
  • Since the art gallery doesn't even have a roof, the whole floor is visible anyways. If you are wondering how the guard manages to hover high above the gallery - he doesn't. He has a telescopic camera tied to a weather balloon. There. Problem solved. Therefore, what counts for visibility isn't occlusion. The only thing that matters is that the walls are facing the right way.
  • You can assume the floor plan is a (not necessarily connected) union of rectangles. I was thinking about including diagonal walls, but this version fits better with the theme. Any internal walls have a definite thickness. If they didn't, you could just ignore them, and that's no good.
  • The guard doesn't have to be inside the gallery. They can sit outside, or even lean against the walls. You, however, cannot place the guard with infinite precision. If you try to align the guard with one of the walls, they will be displaced infinitesimally in one direction chosen with uniform probability. But yes, you can try.

The input will consists of the following characters:

  • The space character represents 1m x 1m of empty space. It's up to you to decide if it's inside the gallery, or outside. You may assume that the input encodes a rectangular area that includes all of the gallery floor and that all of the walls are depicted.
  • - and | represent 1m x 1m squares with walls (east-west and north-south respectively) passing through their centers. Each wall will be adjoined by either the same type of wall or a corner in its lengthwise direction, and empty space in its transverse direction.
  • + Represents a corner. Each corner will be adjoined by an east-west wall either on its east side or on its west side, but not both, and by a north-south wall either on its north side or its south side, but not both, and by empty space on the remaining two sides, and in all four diagonally neighboring tiles.

Your objective is to determine and mark all guard locations - tile centers - that maximize the number of wall segments viewed from the correct side - empty space should be marked with . and wall tiles should be marked with #.
In the event that none of the optimal spots align with tile centers, do not mark any tile as optimal. It is the user's responsibility to provide a more detailed floor plan. You may optionally display an error message in that case. If you choose so, the error message must be: displayed in every situation in which no optimal tile would have been marked; same for every input that causes it to be shown; includes at one character not allowed in any valid output.

Example cases:

+----+    +----+
|    |    |....|
|    | => |....|
|    |    |....|
+----+    +----+

The guard can be anywhere inside the gallery, but they can't lean against the wall because they might suddenly find themself on the other side of that wall unexpectedly.

+---+        +---+    
|   |        |   |  . 
+---+ +-+    +---+ +-+
      | | =>       | |
      | |          | |
      | |          | |
      +-+          +-+

In this case, the gallery consists of two separate buildings, and the guard's best spot lies outside either of them.

|   |
|   +-+ => error
+-+   |
  |   |

In this case, there is a 1m x 2m area in which the guard can see all of the walls, but there's no way to depict that in the output. You may pass the input unmodified, or you may output an error message.

+-----+    +-----+
|     |    |.   .|
| +-+ |    | +-+ |
| | | | => | | | |
| +-+ |    | +-+ |
|     |    |.   .|
+-----+    +-----+

This art gallery has a courtyard. Two inner walls must be left unprotected, but it doesn't matter which ones.

  +-+          +-+    
  | |         .#.#... 
  +-+ +-+     .###.##+
      | |     .....#.|
+-+   +-+ => +##...##+
| |          |.#..... 
+-+ +-+      +##.###. 
    | |       ...#.#. 
    +-+          +-+  

For every wall but four, there is another wall such that exactly one of the two can be seen at any given time. As long as the other four walls are guarded, the number of walls guarded is maximized. This is also one of the rare cases where the guard can lean against a wall - if they fall through, they'll wind up guarding another wall instead.


This is not an art gallery. It's a fence with no inside or outside. Invalid input.

+---+-+---+   +
|   | |   |   |
|   | |   |   +

You might think this depicts two rooms with an internal wall, or three rooms with two internal walls - either way, internal walls are banned. Also, fences are banned. Invalid input.

+---+ +--+ +-+
|   | |  ++| |
|   +-+   |+-+

This building has a clear interior and exterior, but the southern corridor is too narrow, the dent in the north-east corner is too jagged to leave enough room for art, and the nearby closet is too close to the main building. Each of these reasons suffices to make this an invalid input.

| +-+     |
|    *a k |
| ei 32A  |

There's a gap in the northwest corner, debris all over the floor, and the southern wall has exposed scaffolding. All wrong.

  • \$\begingroup\$ I'm not sure if I'm misreading, but I feel as though there is a lot of excess information provided, while certain more basic concepts are left unexplained. For example, the rule about guard displacement being uniformly random and the size of the tiles being 1m2 seem unnecessary for computing the output, whereas the rule for what the guard sees was difficult for me to interpret. My understanding is that the guard sees infinitely far in every direction, but not through walls? \$\endgroup\$ Mar 20 '20 at 16:00
  • \$\begingroup\$ @FryAmTheEggman from the third point: "Therefore, what counts for visibility isn't occlusion. The only thing that matters is that the walls are facing the right way". Are you suggesting I should reformulate that? \$\endgroup\$ Mar 20 '20 at 16:09
  • \$\begingroup\$ Yes, that part wasn't totally clear to me, and it feels like a very important detail that is somewhat buried amongst much less relevant information. \$\endgroup\$ Mar 20 '20 at 16:24

For How Long am I Alone?


You are a factory worker, whose shift is from time X to time Y. It's a very boring job, and you want to know if any other workers are working during your shift. Given a list of start and end times for the other workers and your own shift time, output the longest amount of time that you are the only one working in the factory.


List of start and end times. Any reasonable format is allowed, such as a list of tuples representing (startHour, startMinutes, endHour, endMinutes) or a list of date objects.

A pair of times, which represent your own start and end times. These may be received as a tuple/list or as separate arguments. Again, the times can be passed as a tuple, date objects, or two object array representing (hour, minutes), or you can pass the hours and minutes as separate arguments.

Each person starts working precisely at their start time and gets off work right when the end time starts. For example, if someone is working from 8:00 to 17:00, at 17:00 they are not considered to be at work anymore.

Each person does his shift 7 days a week.

If you choose to use date objects, the "Year" field of the date objects must always be the same across all inputs.

Note that the end time of your shift can look like it's earlier than your start time, e.g. 21:30 - 5:30. This means that your shift starts at 21:30 at the first day and ends at 5:30 on the next day.


The longest interval in minutes in which you are the only one working in the factory.

Test Cases

In the form of [(hh:mm,hh:mm)...], hh:mm, hh:mm

[(3:30, 12:00), (13:00, 21:40)], (8:30), (16:30) -> 60
[(1:01, 1:03), (1:04, 1:06), (1:07, 1:10)], (1:00), (1:10) -> 1
[(21:00, 5:00), (22:30, 7:00)], (0:00), (4:00) -> 0


Should I keep the part about the shift being able to stretch across midnight?

Is the input specification clear enough?

Any suggestions welcome.

  • \$\begingroup\$ Say my shift is overnight and someone has a shift that isn't overnight. How do I know what day said shift belongs to? e.g. if I'm working from 21:30 to 5:30 and I get another input as 1:00 to 4:00 how do I know if I haven't even started? \$\endgroup\$
    – RGS
    Mar 20 '20 at 13:26
  • \$\begingroup\$ @RGS Good question. Added a part about the shifts being 7 days a week, so there is no confusion. I feel like the part about having a overnight shift might make this challenge unnecessarily complicated; What do you think? \$\endgroup\$
    – Gymhgy
    Mar 20 '20 at 18:13
  • \$\begingroup\$ I am not a sandbox veteran, but from my POV this challenge will have us handling intervals and do arithmetics with the interval endpoints and that is probably the main core of the challenge. But adding the overnight shifts means we are trying to intersect segments of a circumference, instead of regular intervals, which is also interesting, I think! (do you understand what I mean with this?) So maybe either remove overnight shifts or rephrase the challenge as intersecting segments of a circumference? So that it becomes more clear that it isn't just an edge case, but the core challenge itself \$\endgroup\$
    – RGS
    Mar 20 '20 at 19:00
  • \$\begingroup\$ As for the title, I would have "For how long am I alone" because "How long am I alone" looks like you are asking for your length when you are alone, instead of the amount of time during which you will be alone. \$\endgroup\$
    – RGS
    Mar 20 '20 at 19:01
  • \$\begingroup\$ Instead of "must always be the same across all inputs", I suggest "will always be the same across all inputs". This makes it more clear that you don't have to deal with the year. \$\endgroup\$
    – S.S. Anne
    Mar 20 '20 at 23:56

Decimal to String with Mandatory Length



Inspired by this SO question, which asks for the most accurate precision of decimal values using either rounding or scientific notation as string, with at most 15 characters long. This would include the -, ., and E in the output-string.



  • A decimal value \$s\$
  • An integer output-length \$n\$


  • A string of the most accurate representation of the given decimal value, with a length exactly equal to the output-length \$n\$. NOTE: the linked SO question ask for at most 15 characters long, but this challenge asks for exactly \$n\$ characters long instead.

Challenge rules:

  • The output-length input is guaranteed to be \$n\geq7\$
  • The decimal string input is guaranteed to be valid and non-empty
  • You are allowed to take the input-decimal \$s\$ as string
  • The input-decimal is guaranteed to only contain the characters 0123456789-., and will not start with an . (but 0. instead), nor start with unnecessary leading 0s like 001.23 instead of 1.23 (except for 0.).
  • The output-string is guaranteed to only contain the characters 0123456789-.E (or e instead of E). The restrictions mentioned one bullet-point above doesn't apply here however!
  • If the length of the integer part of a number (including - for negative values) is larger than the given output-length: use a scientific notation (with either e or E) and rounded precision. I.e. with inputs s = "-987654321987654321.987654321"; n = 15, the output is supposed to be one of these: ["-987654321.99E9", "-9876543219.9E8", "-98765432199.E7"]
  • It is allowed to add leading 0s to get to length \$n\$ without changing its decimal (base-10) value. I.e. with inputs s = "-123.00"; n = 15, the output "-00000000000123" would be valid.
  • It is allowed to add trailing 0s to the decimal parts to get to length \$n\$ without changing its decimal (base-10) value. I.e. with inputs s = "-1.23"; n = 15, the output "-1.230000000000" or "-1.2300000000E0" are both valid. With inputs s = "-0.123456789123456789"; n = 15, the only possible output is "-.1234567891235".
  • Rounding can be one of: HALF_UP, HALF_DOWN, HALF_EVEN (please specify which one your answer is using). (My test cases where this applies use HALF_UP.)
  • If multiple outputs are possible, just output one, multiple, or all of them.

General rules:

  • This is , so shortest answer in bytes wins.
    Don't let code-golf languages discourage you from posting answers with non-codegolfing languages. Try to come up with an as short as possible answer for 'any' programming language.
  • Standard rules apply for your answer with default I/O rules, so you are allowed to use STDIN/STDOUT, functions/method with the proper parameters and return-type, full programs. Your call.
  • Default Loopholes are forbidden.
  • If possible, please add a link with a test for your code (i.e. TIO).
  • Also, adding an explanation for your answer is highly recommended.

Test cases:

All these test cases will use the output-length as 15 (and rounding mode HALF_UP):

Input:                           Possible Outputs:
"987654321987654321.987654321"   "987654321.988E9","9876543219.88E8","98765432198.8E7","987654321988.E6"
"-987654321987654321.987654321"  "-987654321.99E9","-9876543219.9E8","-98765432199.E7"
"1234567891234567.123456789"     "1234567.89123E9","12345678.9123E8","123456789.123E7","1234567891.23E6","12345678912.3E5","123456789123.E4"
"-1234567891234567.123456789"    "-1234567.8912E9","-12345678.912E8","-123456789.12E7","-1234567891.2E6","-12345678912.E5"
"0.123456789123456789"           ".12345678912346"
"-0.123456789123456789"          "-.1234567891235"
"5.5555555555555555555555555"    "5.5555555555556"
"-5.5555555555555555555555555"   "-5.555555555556"
"123456789123456"                "123456789123456"
"-123456789123456"               "-123456.78912E9","-1234567.8912E8","-12345678.912E7","-123456789.12E6","-1234567891.2E5","-12345678912.E4"
"123.00"                         "123.00000000000","0123.0000000000","00123.000000000","000123.00000000","0000123.0000000","00000123.000000","000000123.00000","0000000123.0000","00000000123.000","000000000123.00","0000000000123.0","00000000000123.","000000000000123","123.000000000E0","0123.00000000E0","00123.0000000E0","000123.000000E0","0000123.00000E0","00000123.0000E0","000000123.000E0","0000000123.00E0","00000000123.0E0","000000000123.E0","12.3000000000E1","012.300000000E1","0012.30000000E1","00012.3000000E1","000012.300000E1","0000012.30000E1","00000012.3000E1","000000012.300E1","0000000012.30E1","00000000012.3E1","1.23000000000E2","01.2300000000E2","001.230000000E2","0001.23000000E2","00001.2300000E2","000001.230000E2","0000001.23000E2","00000001.2300E2","000000001.230E2","0000000001.23E2",".123000000000E3","0.12300000000E3","00.1230000000E3","000.123000000E3","0000.12300000E3","00000.1230000E3","000000.123000E3","0000000.12300E3","00000000.1230E3","000000000.123E3","1230.0000000E-1","01230.000000E-1","001230.00000E-1","0001230.0000E-1","00001230.000E-1","000001230.00E-1","0000001230.0E-1","00000001230.E-1","000000001230E-1","12300.000000E-2","012300.00000E-2","0012300.0000E-2","00012300.000E-2","000012300.00E-2","0000012300.0E-2","00000012300.E-2","000000012300E-2","123000.00000E-3","0123000.0000E-3","00123000.000E-3","000123000.00E-3","0000123000.0E-3","00000123000.E-3","000000123000E-3","1230000.0000E-4","01230000.000E-4","001230000.00E-4","0001230000.0E-4","00001230000.E-4","000001230000E-4","12300000.000E-5","123000000.00E-5","1230000000.0E-5","12300000000.E-5","123000000000E-5","123000000.00E-6","1230000000.0E-6","12300000000.E-6","123000000000E-6","1230000000.0E-7","12300000000.E-7","123000000000E-7","12300000000.E-8","123000000000E-8","123000000000E-9"
"-123.00"                        "-123.0000000000","-0123.000000000","-00123.00000000","-000123.0000000","-0000123.000000","-00000123.00000","-000000123.0000","-0000000123.000","-00000000123.00","-000000000123.0","-0000000000123.","-00000000000123","-123.00000000E0","-0123.0000000E0","-00123.000000E0","-000123.00000E0","-0000123.0000E0","-00000123.000E0","-000000123.00E0","-0000000123.0E0","-00000000123.E0","-000000000123E0","-12.300000000E1","-012.30000000E1","-0012.3000000E1","-00012.300000E1","-000012.30000E1","-0000012.3000E1","-00000012.300E1","-000000012.30E1","-0000000012.3E1","-1.2300000000E2","-01.230000000E2","-001.23000000E2","-0001.2300000E2","-00001.230000E2","-000001.23000E2","-0000001.2300E2","-00000001.230E2","-000000001.23E2","-.12300000000E3","-0.1230000000E3","-00.123000000E3","-000.12300000E3","-0000.1230000E3","-00000.123000E3","-000000.12300E3","-0000000.1230E3","-00000000.123E3","-1230.000000E-1","-01230.00000E-1","-001230.0000E-1","-0001230.000E-1","-00001230.00E-1","-000001230.0E-1","-0000001230.E-1","-00000001230E-1","-12300.00000E-2","-012300.0000E-2","-0012300.000E-2","-00012300.00E-2","-000012300.0E-2","-0000012300.E-2","-00000012300E-2","-123000.0000E-3","-0123000.000E-3","-00123000.00E-3","-000123000.0E-3","-0000123000.E-3","-00000123000E-3","-1230000.000E-4","-01230000.00E-4","-001230000.0E-4","-0001230000.E-4","-00001230000E-4","-12300000.00E-5","-123000000.0E-5","-1230000000.E-5","-12300000000E-5","-123000000.0E-6","-1230000000.E-6","-12300000000E-6","-1230000000.E-7","-12300000000E-7","-12300000000E-8"

TODO: Fix the possible outputs of the other test cases as well:

"0.123"                          ".12300000000000"
"-0.123"                         "-.1230000000000"
"1.23"                           "1.2300000000000"
"-1.23"                          "-1.230000000000"
"0.000000000000001"              "10000000000E-25" // Currently incorrect in my reference implementation
"0"                              "000000000000000"

All these test cases will use the output-length as 7 (and rounding mode HALF_UP):

Input:                           Output:
"987654321987654321.987654321"   "9.88E17"
"-987654321987654321.987654321"  "-9.9E17"
"1234567891234567.123456789"     "1.23E15"
"-1234567891234567.123456789"    "-1.2E15"
"0.123456789123456789"           ".123457"
"-0.123456789123456789"          "-.12346"
"5.5555555555555555555555555"    "5.55556"
"-5.5555555555555555555555555"   "-5.5556"
"123456789123456"                "1.23E14"
"-123456789123456"               "-1.2E14"
"123.00"                         "0000123" or "01.23E2" or "1.230E2"
"-123.00"                        "-000123" or "-12.3E1"
"0.123"                          ".123000"
"-0.123"                         "-.12300"
"1.23"                           "1.23000"
"-1.23"                          "-1.2300"
"0.000000000000001"              "100E-17" or "1.0E-15" // Currently incorrect in my reference implementation
"0"                              "0000000"
  • 1
    \$\begingroup\$ Suggest testcase: "0.0000000001", 10 -> ".000000000", "0", 10, "0000000000" \$\endgroup\$
    – tsh
    Jan 13 '20 at 1:50
  • \$\begingroup\$ @tsh Added (although I've used n=15 instead of 10 so I could add it to the other list instead of creating two separated test cases for the n=10. Principle remains the same for your test cases, so thanks for the suggestion! \$\endgroup\$ Jan 13 '20 at 7:44
  • \$\begingroup\$ @tsh Your test case actually made me realize that s="0.0000000001", n=10 should be "100000E-15" instead for the most accurate result. Will have to do some fixes to my reference implementation. \$\endgroup\$ Jan 13 '20 at 7:54
  • \$\begingroup\$ Why should it be 100E-17, not 1.0E-15? The rule makes me confusing. \$\endgroup\$
    – tsh
    Jan 13 '20 at 9:14
  • \$\begingroup\$ @tsh 1.0E-15 would be allowed as well. I still have to change the rules after realizing that 0.00000 would be an incorrect result for s="0.0000000001", n=7, but both 100E-17 and 1.0E-15 are allowed, since they are of length 7 and retain the same exact value as 0.0000000001. I currently don't have the time to revise the rules, test cases, and reference implementation unfortunately (and if I delete the Sandbox post temporarily I can't search back for it). \$\endgroup\$ Jan 13 '20 at 9:17
  • \$\begingroup\$ You say this challenge asks for exactly 15 characters long but you then have outputs of varying length. I assume you want the latter. I'm being pedantic, I know, but it's the only thing I can find to fix at the moment ;) \$\endgroup\$
    – lyxal
    Jan 13 '20 at 9:28
  • \$\begingroup\$ @Jono2906 I've changed the part at the output-section. I hope it's a bit clearer now? \$\endgroup\$ Jan 13 '20 at 9:33
  • \$\begingroup\$ @KevinCruijssen it is clearer now. I mean, as I said, I was just being pedantic about things. \$\endgroup\$
    – lyxal
    Jan 13 '20 at 9:34
  • \$\begingroup\$ @Jono2906 Well, it was still a valid remark that I agree with, so thanks. :) \$\endgroup\$ Jan 13 '20 at 9:36
  • \$\begingroup\$ Just another thing: Perhaps The decimal string input is guaranteed to be valid and non-empty should be The decimal string input is guaranteed to be a valid float and non-empty \$\endgroup\$
    – lyxal
    Jan 13 '20 at 10:41
  • \$\begingroup\$ @Jono2906 Textual there isn't a difference between decimal/double/float, though. I had the sentence in my head as "the decimal string input is guaranteed to be (a) valid (decimal) and non-empty". I could change it to that if it makes it clearer, but talking about decimal first and float after that is more confusing than clarifying imho. But if you indeed meant "the decimal string input is guaranteed to be a valid decimal and non-empty" I'll change it. \$\endgroup\$ Jan 13 '20 at 11:49
  • \$\begingroup\$ @KevinCruijssen Yep, I indeed meant decimal. I've no clue why I said float. \$\endgroup\$
    – lyxal
    Jan 13 '20 at 20:10

Ordinal to Cardinal

Given a positive integer represented as the English spelling of an ordinal number, return the equivalent cardinal number.


  • Where an integer requires multiple words to spell, only the last word changes.

  • The following integers are strongly irregular:

    • "one" becomes "first"
    • "two" becomes "second"
    • "three" becomes "third"
  • Other integers take a suffix of "th", however there are a few integers that are weakly irregular:

    • "five" becomes "fif(th)"
    • "eight" becomes "eigh(th)"
    • "nine" becomes "nin(th)"
    • "twelve" becomes "twelf(th)"
    • "twenty" to "ninety" become "twentie(th)" to "ninetie(th)".
  • The input can be assumed to be the English spelling of an ordinal number that follows the above rules to transform it into the equivalent cardinal number.


  • "one hundred and nineteen" becomes "one hundred and nineteenth"
  • "one hundred and twenty" becomes "one hundred and twentieth"
  • "one hundred and twenty one" becomes "one hundred and twenty first"

This is , so the shortest program or function that breaks no standard loopholes wins!

  • \$\begingroup\$ There are a lot of loosely related challenges, with this one being the closest. I don't think this is a dupe at all, though, since the amount to change is much more significant. Is there an upper limit on the input? If not, you definitely need to specify how the larger numbers might appear i.e. do we need to handle "milliard" as well as "million"? \$\endgroup\$ Mar 26 '20 at 15:47
  • \$\begingroup\$ @FryAmTheEggman That and the other challenge takes the numbers as digits rather than words, which IMHO is a significant difference already. As for large numbers, you can assume for the purposes of the question that any number I forgot about takes a "th" suffix. \$\endgroup\$
    – Neil
    Mar 26 '20 at 22:30
  • \$\begingroup\$ Which integers do we need to handle? I'd suggest limiting it to, say, numbers from 1 to 99. Or if you do want all positive integers, could you please clarify how these are written out? \$\endgroup\$
    – xnor
    Mar 27 '20 at 20:57
  • \$\begingroup\$ @xnor a) this challenge is about words, not numbers b) the rules are there, I don't understand what you're missing \$\endgroup\$
    – Neil
    Mar 27 '20 at 21:51
  • \$\begingroup\$ @Neil Like, is "one billion, two hundred and thirty four million, five hundred and sixty seven thousand, eight hundred and ninety" a possible input, for which the output would be "one billion, two hundred and thirty four million, five hundred and sixty seven thousand, eight hundred and ninetieth"? If so, what is the exact format for such numbers? I understand that really only the last word matters for the conversion in the challenge, but it might make a difference for, say, a regex that does a replacement that might falsely trigger on something like "Duotrigintillion". \$\endgroup\$
    – xnor
    Mar 27 '20 at 22:01
  • \$\begingroup\$ @xnor Why would it falsely trigger on duotrigintillion? Is there no duotrigintillionth? \$\endgroup\$
    – Neil
    Mar 27 '20 at 22:39
  • \$\begingroup\$ @Neil I mean if it's part of a longer number and the regex does a replacement that doesn't check for the end of the string, but simply replaces certain sequences of characters. Duotrigintillion is an arbitrary example; I don't expect it specifically to actually "collide" with anything useful. \$\endgroup\$
    – xnor
    Mar 27 '20 at 22:42
  • \$\begingroup\$ @xnor Well, surely if it collides as the last word, then it will collide as an earlier word, which would be an error, according to the first rule? \$\endgroup\$
    – Neil
    Mar 27 '20 at 23:41
  • \$\begingroup\$ @Neil Oh, you're right, that would catch it. Maybe a more useful example is "one hundred and one" wrongly being made into 'first hundred and first". In any case, I think it would be useful to either add large-valued test cases or put an upper bound. \$\endgroup\$
    – xnor
    Mar 28 '20 at 0:54
  • \$\begingroup\$ @xnor I still don't see that it needs an upper bound. You can just assume that the rules I've given apply, even if they don't in real life for some reason. \$\endgroup\$
    – Neil
    Mar 28 '20 at 1:07

Mom-rounding the time


Growing up with my mother, whenever she looked at a clock to check the time, she would always say "shoot, it's already X!" and then I would look at the clock and realize she was just rounding the time in a really weird way.


Given a time with hours and minutes, round it like my mom would. Rounding always occurs upwards. Say it is currently H hours and M minutes.

  • if M is 0, no rounding occurs; my mom isn't that crazy;
  • if M is 9 or less, my mom rounds to H:15;
  • if M is 19 or less, my mom rounds to H:30;
  • if M is 34 or less, my mom rounds to H:45;
  • for any other value of M, my mom rounds to H+1:00.


You will take a time that needs rounding, in any sensible format. ISO strings for date/time, two integers representing hours/numbers, a string with two integers and a separator; these are all fair game.


The string "shoot, it's already X" with X replaced with the mom rounding time.


This is so shortest answer wins. However, if your source code contains the substring shoot, it's already then you may subtract 19 from your score.

Test cases

Here is the program I use to generate the test cases.

11:00 -> shoot, it's already 11:00!
 3:08 -> shoot, it's already 03:15!
 1:09 -> shoot, it's already 01:15!
13:13 -> shoot, it's already 13:30!
 2:35 -> shoot, it's already 03:00!

  • 1
    \$\begingroup\$ +1 because my mom's rounding is similarly weird too. One thing though: shoot, it's already has a ' which must be escaped in many languages, and the substring condition is unfairly penalizing langauges without string literal support. \$\endgroup\$
    – Bubbler
    Apr 1 '20 at 0:16
  • 5
    \$\begingroup\$ I suggest removing the unnecessary fluff about adding text and only keep the conversion. Bonuses in code golf are bad in general. Here it seems you try to even the playing field by explicitly disadvantaging languages with string compression, but end up making the false assumption that the substring will occur if used literally, even though some languages will need to escape the quote. \$\endgroup\$
    – Adám
    Apr 1 '20 at 6:17
  • 2
    \$\begingroup\$ Assuming hours wrap around in 24-hour time, some test cases showing this would be good. \$\endgroup\$
    – xnor
    Apr 1 '20 at 9:25
  • \$\begingroup\$ Those bullet-points should either be else-ifs instead of ifs or you should just define ranges. Currently a minute of 3 would first be rounded to 15 for being <=9, then 15 is rounded again to 30 for being <=19, and then again to 45 for being <=34. So basically: 0 remains 0; <=34 becomes 45, and >34 becomes 0 with the hour increasing, and the other bullet-points could be ignored. I think something like M=0→H:00; M=[1,9]→H:15; M=[10,19]→H:30; M=[20,34]→H:45; M=[35,59]→H+1:00 (perhaps in text form) would be clearer imo. \$\endgroup\$ Apr 1 '20 at 10:10
  • \$\begingroup\$ @petStorm thanks for your edit but I would prefer if you did not edit any reference programs into my sandboxed posts (you may comment with a TIO link) nor edited the challenge to cope with the feedback I get from commenters, nor to include a whole "test cases" section (but you can include it in your TIO link). In particular, usually when I don't include test cases right from the start is because I want to polish the spec a bit before trying to understand what test cases are really relevant and needed. \$\endgroup\$
    – RGS
    Apr 1 '20 at 11:50
  • 1
    \$\begingroup\$ And this is especially true because you create a whole section and then write "Here is the program I used to generate the test cases" as if you were me, which is not ok. \$\endgroup\$
    – RGS
    Apr 1 '20 at 11:52

  • \$\begingroup\$ I like this idea. Do we have a default for taking "infinite lists" as input? I'd mildly suggest limiting the numbers to positive integers on general principle. \$\endgroup\$
    – xnor
    Apr 2 '20 at 4:26
  • \$\begingroup\$ How do you propose taking an infinite list as input? I think the natural approach would use as "input" a program (or function with no arguments) that runs forever; the list would consist of the numbers that it outputs when run. (This would be sort of like a plug-in that your answer could use.) The problem is that that might make the challenge trivial. But I'm not sure how else to take the input list. \$\endgroup\$ Apr 2 '20 at 6:14
  • \$\begingroup\$ In some languages the input could be a stream or iterator \$\endgroup\$ Apr 2 '20 at 8:36
  • \$\begingroup\$ I don't know what you want to allow, but some other possible implementations of infinite lists could be as a generator function that produces a new value on each call, or a black-box function taking a natural n and giving the n'th value. \$\endgroup\$
    – xnor
    Apr 2 '20 at 12:35
  • \$\begingroup\$ @xnor I don't think I will limit it to positive integers because positive integers are not bounded below, meaning sequences could only increase. This pretty radically changes the content of the challenge. \$\endgroup\$
    – Wheat Wizard Mod
    Apr 2 '20 at 13:14
  • \$\begingroup\$ Let the first and second items of the list be x and y respectively. This checks whether input-x is divisible by y-x. (Doesn't work sometimes, I'll take a look.) \$\endgroup\$
    – user92069
    Apr 3 '20 at 7:27

How many Temtem can I breed?

Temtem is a monster-catching MMORPG. Within the game you have the ability to breed two Temtem to create an egg which then hatches into the baby of its mother. The ability to do this depends on several properties of the parent Temtem:

  • A Temtem has a gender, which is either male or female. To breed, you need one Temtem of each gender.
  • A Temtem has one or two types. A pair of Temtem can only breed if they have a type in common.
  • A Temtem has a fertility which ranges from 0 to 8. Temtem with a fertitilty of 0 can no longer breed in captivity. The fertility of each parent decreases by 1 when they breed.

The resulting Temtem inherits some of its properties from its parents.

  • The baby Temtem's gender is random. For the purposes of this question, this means that you can choose the gender, but you cannot change it later.
  • The baby Temtem inherits its mother's type. (This is not strictly true but it is always possible to evolve the baby to give the mother's type if necessary.)
  • The baby Temtem inherits the lesser of its parents' fertility.

Simple example:

  • One female Temtem with fertility 3 and one male Temtem with fertility 2 of the same type.
  • Breeding reduces their fertility to 2 and 1 and we choose the baby, which also has 1 fertility, to be male.
  • The female can breed again with both males, at which point all the Temtem now have 0 fertility.
  • This gives you a total of five Temtem.

You challenge is to write a program or function which accepts a list of Temtem and outputs the maximum number of Temtem it is possible to breed, assuming luck is on your side.

You can use any convenient input method for the Temtem, as long as it breaks no standard loopholes. For instance, you could use a string of three or four characters encoding the gender, fertility and type(s). (Your input method must be able to support 12 different types and 132 different pairs of types.)

You can output either the total number or the number of new Temtem, or you can also output the resulting list of Temtem in the order they were born.

This is , so the shortest program or function wins!

  • \$\begingroup\$ Without the types this makes an interesting "limited Fibonacci-like" growth - I'd be curious to see if it had already been studied. I think adding in the types makes it more likely that brute forcing will be the best approach. I haven't done much work on this yet, so of course I could be wrong. Separately, this definitely needs a test case where inter-type breeding gives a larger result than handling each separately, though I'm pretty sure you knew that. \$\endgroup\$ Apr 6 '20 at 16:47

Word Grid Puzzle Iterator

I commonly see an advert for a word-based game, where, by removing a word from a grid of letters, the remaining letters "collapse", horizontally and vertically, leading to further words being findable.

The challenge is to take in a grid of characters and a word as an input, and output the collapsed grid after that word has been removed.

Words in the grid could be oriented in any direction.

This is code-golf, usual exclusions apply.


The following set of examples follow on from each other - so the output from the first example is the input to the second.



N   F   A   D   S   T
I   O   I   E   N   T
A   G   R   W   O   H
R   L   I   A   H   A
S   L   E   E   W   W

Out: If the word "AIR" is removed, none of the rows or columns are empty and so the rest of the grid remains as-is:

N   F       D   S   T
I   O       E   N   T
A   G       W   O   H
R   L   I   A   H   A
S   L   E   E   W   W


If the word "HAIL" is removed, the letters in the 2nd-5th columns drop down one, making the word "SNOW" accessible:

N                   T
I   F       D   S   T
A   O       E   N   H
R   G       W   O   A
S   L   E   E   W   W


If SNOW is removed, the columns collapse horizontally:

N               T   
I   F       D   T   
A   O       E   H   
R   G       W   A   
S   L   E   E   W   


If THAW is removed, the T drops down so that SLEET is now accessible:

I   F       D       
A   O       E       
R   G       W       
S   L   E   E   T   


Removing SLEET clears a row and a column, and so the grid collapses in both directions:

I   F   D           
A   O   E           
R   G   W           


The remaining three vertical words (RAIN, DEW, FOG) can then be removed individually:

F   D               
O   E               
G   W               


In the above examples, I have not resized the arrays as the outside rows/columns are made empty; instead just leaving them blank (i.e. the final array is 6x5, as is the starting array). Your program may also do this, or it may resize the array to remove empty rows and columns if you prefer.

For example, assuming the input is a 1x5 array:

H A T C E removing HAT could become _ _ _ C E or CE, both are valid. (underscores represent spaces for formatting purposes)

Inputs and Outputs:

Any reasonable format is acceptable - arrays, strings, etc. But:

  • you cannot assume that the orientation of the word to remove is the same as that in the input. e.g you won't get the input "LIAH" because the word HAIL is written right-to-left on the grid.
  • the output must be the same format as the input (i.e. the program must accept the output from the previous iteration as the grid input to the next)
  • You can assume the word will be in the grid, horizontally or vertically (not diagonally)
  • Words will never contain spaces, only the letters A-Z; and if the word is found but with a space in it, then that doesn't count as matching the word
    • You can assume the case of the word will match the grid, in whatever case suits your language (UPPER, lower, camelCase, whatever)
  • You can assume that the Input word will only appear once on the grid

Sandbox Questions

  • Does this feel too much like multiple challenges (find the word, collapse the array)?
  • \$\begingroup\$ I think the idea stands on its own logically, though perhaps there is a dupe out there somewhere. Separately, I think you should explain the situation where a word appears more than once. Your resizing comment also seems a bit odd: I would expect that if instead of HAIL the input was AGWOH the word wouldn't be able to be removed. \$\endgroup\$ Apr 3 '20 at 20:20
  • \$\begingroup\$ Thanks for your comments as always. I have tried to clarify. I'm not sure what you mean about the resizing comment - only empty rows and columns can be removed; and words will always be A-Z (or whatever case you prefer) and must match exactly the input, in any orientation. \$\endgroup\$ Apr 6 '20 at 7:43
  • \$\begingroup\$ I see, that is what I expected, but when I read the challenge I was unsure. I think your edit makes it much clearer. (What I was trying to get at was that I expected gaps in unempty rows/columns to be "unmatchable", which I think before was made somewhat ambiguous by the wording of the purely aesthetic output) \$\endgroup\$ Apr 6 '20 at 16:30

chaining couples with parity


Take the \$n\$ first integers (with 0 included or not) with \$n\$ an even number except 0.

The goal is to produce a (not so) random chain of couples with these numbers, for example with \$n=6\$ : (5, 4), (1, 6), (3, 2)

But you have to respect a bit of parity and randomness :

  1. Each second number in a couple must have the same parity than the first number of the next couple. No rule for the first number of the first couple and the second number of the last couple. So the example above is not a correct answer.

(5, 4), (6, 1), (3, 2) is a correct answer for \$n=6\$.

So this is a sort of parity chain.

  1. First number (of the first couple) has to be chosen randomly (uniform) in the \$n\$ first integers.


An even number \$n\$ greater or equal than 2.

Valid output examples

  • Input: \$n=2\$ Outputs (1, 2) and (2, 1) are valid. (0, 1) and (1,0) are also.

  • Input: \$n=4\$ Output: (0, 1), (3, 2) (if start with 0) because 1 and 3 are odd

  • Input: \$n=4\$ Output: (1, 4), (2, 3) (if start with 0)

  • Input: \$n=6\$ Output: (0, 5), (1, 3), (4, 2)

  • Input: \$n=8\$ Output: (6, 7), (1, 0), (2, 5), (3, 4)

Invalid output examples

  • Input: \$n=4\$ Output:(1, 2), (3, 4) because 2 is even and 3 is odd.

  • Input: \$n=6\$ Output:(5, 4), (1, 6), (3, 2) because 4 is even and 1 is odd and also because 6 is even and 3 is odd.

What if \$n\$ is odd?

No rule for \$n\$ if it's odd. All outputs accepted!

What about output's format?

No special formatting is expected. You just have to separate the couples such that one can correctly see them.

Sandbox Questions


  • 1
    \$\begingroup\$ Use \$ instead of $ to use MathJax. Also, what do you exactly want by "random"? Is it acceptable to pick from two answers (e.g. pick between (1 2)(4 3)(5 6) and (5 6)(4 3)(1 2))? Check this and this. \$\endgroup\$
    – Bubbler
    Apr 8 '20 at 1:43
  • \$\begingroup\$ @Bubbler Thx. I tried to be clearer about randomness expectations. \$\endgroup\$
    – david
    Apr 11 '20 at 12:58
  • \$\begingroup\$ "First number (of the first couple) has to be chosen randomly (uniform) in the n first integers." It is still easier than having to pick from all possible answers. No problem if you intended it though. \$\endgroup\$
    – Bubbler
    Apr 12 '20 at 23:46


Write a program or function such that given a string \$ S \$, encode it with Rot1Rot3. To encode it, do the following:

  1. Take \$ S\$ and partition it into blocks with \$ 3 \$ characters each. If it does not partition equally (i.e. \$ |S| \bmod 3 \neq 0 \$), take away some characters from the end of the string until it does, and form a smaller partition from the 'taken-away' characters.

  2. For each individual partition, rotate it to the right by \$ 1 \$ step.

  3. Finally, rotate all of the partitions together to the right by \$ 3 \$ steps, as if each partition were one character.


Input: Hello, code golf!

  1. Partition it like so: |Hel|lo,| co|de |gol|f!| (pipes to show separation). Notice that f! does not form a full partition, and is therefore left as a partition with \$ 2 \$ characters instead.

  2. Rotate each individual partition to the right by \$ 1 \$ step: |lHe|,lo|o c| de|lgo|!f|.

  3. Rotate the partitions as a whole to the right by \$ 3 \$ steps: | de|lgo|!f|lHe|,lo|o c|.

Final Output: delgo!flHe,loo c

Additional Info

  • You can expect any sequence of any characters, as long as they are printable.
  • This is , so the shortest code in bytes wins.

Test Cases


Hello, code golf!
flog yhcrana
import this
All animals are equal, but some animals are more equal than others.


 delgo!flHe,loo c
yg rhcaanofl
torh tsipim
hotser.lAln aaim lsearq eluab,  utmsoae mnisalr ame eorq eluah t an

Check my Knight's Tour

Given as input an n-by-n matrix of integers, check that all integers from 1 to are present, and that all pairs of consecutive integers are exactly a Knight's move apart. (For instance, a 4-by-4 board with the values

1 2 1 2
2 1 2 1
1 2 1 2
2 1 2 1

would meet the condition that all 1s would have a 2 that was exactly a Knight's move away, but this would of course not constitute a Knight's Tour.) The integers 1 and do NOT need to be a Knight's move apart. You must be able to support at least 7 different values for n including 10.

A Knight's move is possible between any two squares that are a Euclidean distance of √5 apart.

Output follows normal rules.

This is , so the shortest program or function that breaks no standard loopholes wins!

  • \$\begingroup\$ I think your example would be clearer if you showed the matrix with the columns vertically stacked. It took me a while to figure out what the example was showing as is. \$\endgroup\$ Apr 12 '20 at 5:03
  • \$\begingroup\$ @FryAmTheEggman My MathJax isn't good enough for that, sadly. Feel free to change all of my backquotes to backslashed dollar signs. \$\endgroup\$
    – Neil
    Apr 12 '20 at 9:25
  • \$\begingroup\$ Maybe you can add more test-cases \$\endgroup\$ Apr 12 '20 at 9:54

How could these rep and scores have happened?

Challenge: Given a reputation score, a count of questions with accepted answers, and a list of net votes on questions, calculate a sequence of actions that could have produced them. To keep this simple, here's a few simplifications from how Stack Exchange actually works:

  • The only things that change reputation are upvotes on questions (+10), downvotes on questions (-2), and accepting an answer to a question (+2).
  • The daily cap of 200 never comes into play.
  • Votes are never retracted and accepts are never rescinded.

The minimum possible reputation of 1 still applies. If there are multiple sequences of actions that can produce the given state, then your program/function should return the shortest possible sequence (or one of the shortest, if there's multiple that are tied for the shortest). If there are no sequences of actions that can produce the given state, consider that Undefined Behavior, so it doesn't matter what your program does.

Examples/test cases:

  • A user has 1 reputation, hasn't accepted any answers, and has no questions. The shortest possible sequence is "".
  • A user has 33 reputation, has accepted 1 answer, and has questions with scores 2 and 1. A shortest possible sequence is "upvote, upvote, upvote, accept".
  • A user has 3 reputation, hasn't accepted any answers, and has a question with score -5. The shortest possible sequence is "downvote, downvote, upvote, downvote, downvote, downvote, downvote".

The standard restrictions against loopholes apply. I/O may be in any convenient format. Examples: For input, [101,0,4,3,2,1] or (101,0,[4,3,2,1]) could mean a user with 101 reputation, 0 questions with accepted answers, and questions with scores 4, 3, 2, and 1. For output, the string UDA could mean "upvote, downvote, accept".

This is code golf, so the shortest program/function wins.


Draw a Bracket

A bracket is a way of organizing a tournament as a tree structure. It follows several rounds of matches with half of the competitors being eliminated each round, until one competitor remains.

In this challenge you will be given as input a list of one digit \$(0-9)\$ integers as input representing competitors and you should output an ASCII representation of a bracket. The exact specifications are in the Output section of this challenge.

For example if the input is 1 2 3 4 5 6 7 8 then the output would be:


In this challenge the lower number is always the winner in a matchup.


The input will always be of length \$2^n\$ for some positive integer \$n\$, and will consist entirely of one digit numbers. You may take the integers as their representative characters instead if this pleases you.


I will explain precisely how the output is created momentarily, it is a bit hard.


This is the shortest answer in bytes wins.

  • \$\begingroup\$ I fail to see how this is a bracket. \$\endgroup\$ Apr 13 '20 at 6:39
  • \$\begingroup\$ I am worried because this says "you have a string of digits of uncertain length, and I'll explain the rest of the challenge later, it's a bit hard" and that's it. \$\endgroup\$ Apr 13 '20 at 15:47
  • \$\begingroup\$ @mypronounismonicareinstate The length is explicitly \$2^n\$. I do have an example which should help get the idea across. Even though it is a simple idea it is just hard to put the idea in a very concrete specification in a ways that is not incredibly roundabout or esoteric. \$\endgroup\$
    – Wheat Wizard Mod
    Apr 13 '20 at 15:55
  • \$\begingroup\$ @infinitezero I added an explanation of what a bracket is if that helps, I don't know if that resolves your issue. It is hard for me to know since this is very clearly a bracket to me. \$\endgroup\$
    – Wheat Wizard Mod
    Apr 13 '20 at 16:04
  • \$\begingroup\$ The word "bracket" is ambiguous, but I could easily understand from the word "tournament" that the figure is the result of a tournament with \$2^n\$ players where the winner is deterministic. I guess you could draw the same figure rotated 90 degrees to help understanding the task. (And I once misunderstood it as "winner is always the left one" before I saw the line under the figure. It'd be better to randomize the example input a bit to make it clearer.) \$\endgroup\$
    – Bubbler
    Apr 14 '20 at 0:08

The assignment logic may be used in a planning KoTH, in which a student bot chooses a full permutation from the classes and is awarded points from what preference they get.

Assign classes fairly and satisfyingly


In a school, students are signing up for after-school classes. However, the capacity of each class is limited. In order to facilitate class assignment, each student is required to fill in a questionnaire to show his/her preference to the classes by listing all classes in decreasing order of his/her interests.

You are assigned to help assigning classes to them. You want to maximally satisfy their preferences while being fair with the assignment. A student can be assigned with multiple classes.


Write a program or function that receives the following input:

  • a list of classes with the corresponding capacities; and
  • a list of students with their corresponding preference lists,

and output or return either:

  • a list of class with the list of students being assigned to that class; or
  • a list of students with their assigned classes.

You may use any reasonable alternative format for both input and output, for example, apart from receiving two arrays, you may choose to receive two strings, and especially for the second input (which requires a 2-D array), you may even have an input like this (first level delimiter \n, second level delimiter space):

1 2 3 4
4 2 3 1
1 3 2 4
2 3 4 1
3 4 1 2
4 1 2 3

To simplify the challenge, you may assume both classes and students are 0-indexed or 1-indexed. For the use of illustrating the requirements and samples, 1-indexing is used. You may also assume that each preference list is a full permutation of all classes.

The assignment requirements are as follows:

  • Fairness: All students must have roughly the same amount of classes assigned to them, that is, for every \$1\le i\le\text{[Number of students]}\$, $$\left|{\text{[Number of classes assigned]}_i-\frac{\sum\text{[Class capacities]}}{\text{[Number of students]}}}\right|<1.$$
  • Satisfaction: You should fulfill the preferences as well as possible. Specifically, you should fulfill as much first preferences as possible, then as much second preferences as possible, and so on. In case of having the same preference order, the classes should be assigned on first-come-first-served basis.

The fairness rule should be taken first if it conflicts with the satisfaction rule. Test case 3 is an example of handling such conflicts.

The program should terminate in finite time for all practical sizes of inputs.

Test cases

Test case 1


classes = [2, 2, 2, 2],
students = [
 [1, 2, 4, 3],
 [2, 4, 1, 3],
 [3, 4, 2, 1],
 [4, 3, 2, 1]


classes = [
 [1, 3],
 [2, 1],
 [3, 4],
 [4, 2]
students = [
 [1, 2],
 [2, 4],
 [3, 1],
 [4, 3]


  1. It is clear that all 1st priorities can be fulfilled because all of them are different. So each student gets his 1st priority.
  2. Student 1 wants Class 2 as his 2nd priority, and Class 2 still has place for him. So he gets Class 2.
  3. Student 2 wants Class 4 as his 2nd priority, and Class 4 still has place for him. So he gets Class 4.
  4. Student 3 wants Class 4 as his 2nd priority, but Class 4 is already full. No place for him.
  5. Student 4 wants Class 3 as his 2nd priority, and Class 3 still has place for him. So he gets Class 3.
  6. Now each of the students except Student 3 has 2 classes already, so by the rule of fairness they are not considered in the subsequent assignments.
  7. Only Class 1 has place for Student 3, so he gets his 4th priority.

Test case 2


classes = [2, 2, 2],
students = [
 [3, 1, 2],
 [2, 3, 1],
 [2, 3, 1],
 [3, 2, 1]


classes = [
 [1, 2],
 [2, 3],
 [1, 4]
students = [
 [3, 1],
 [2, 1],


  1. It is clear that all 1st priorities can be fulfilled because none of the classes was chosen by 3 or more students as their 1st priorities.
  2. Student 1 wants Class 1 as his 2nd priority, and Class 1 still has place for him. So he gets Class 1.
  3. Student 2 wants Class 3 as his 2nd priority, but Class 3 is already full. No place for him.
  4. Student 3 wants Class 3 as his 2nd priority, but Class 3 is already full. No place for him.
  5. Student 4 wants Class 2 as his 2nd priority, but Class 2 is already full. No place for him.
  6. Now Student 1 has 2 classes already, so by the rule of fairness he is not considered in the subsequent assignments.
  7. Student 2 wants Class 1 as his 3rd priority, and Class 1 still has place for him. So he gets Class 1.
  8. All classes are already full, so no more seats can be assigned. Students 3 and 4 will only get 1 class each.

Test case 3


classes = [1, 1, 1, 2],
students = [
 [1, 2, 4, 3],
 [3, 4, 2, 1],
 [2, 4, 3, 1],
 [2, 4, 1, 3]


classes = [
 [4, 2]
students = [
 [3, 4],


  1. It is clear that all 1st priorities except for Student 4 can be fulfilled.
  2. If we ignore Student 4 and proceed to the second round, Student 2 and 3 will occupy the remaining seats and Student 4 will not get a place (which is disallowed by the fairness rule), so the 2nd priority of Student 4 will be considered first. Since Class 4 still has place for him, he gets Class 4.
  3. Student 1 wants Class 2 as his 2nd priority, but Class 2 is already full. No place for him.
  4. Student 2 wants Class 4 as his 2rd priority, and Class 4 still has place for him. So he gets Class 4.
  5. All classes are already full, so no more seats can be assigned. All students get 1 class each, except Student 2, who gets 2 classes.

Test case 4


classes = [1, 1, 1, 2],
students = [
 [1, 2, 4, 3],
 [3, 4, 2, 1],
 [2, 4, 3, 1],
 [2, 1, 3, 4]


classes = [
 [2, 3]
students = [
 [3, 4],


  1. It is clear that all 1st priorities except for Student 4 can be fulfilled.
  2. If we ignore Student 4 and proceed to the second round, Student 2 and 3 will occupy the remaining seats and Student 4 will not get a place (which is disallowed by the fairness rule), so the 2nd priority of Student 4 will be considered first. However both Classes 1 and 3 are full, he can only get Class 4.
  3. The remaining place for Class 4 goes to Student 2, and we have 3 1st priorities, 1 2nd priority and 1 4th priority fulfilled.
  4. But this is not the best. By breaking the first-come-first-served rule for 1st priority, we can get the best - 3 1st priorities and 2 2nd priorities fulfilled.

Winning Condition

This is a code-golf challenge, so the shortest submission for each language wins. Standard loopholes are forbidden.


Euler's Geometry Puzzle


  • 1
    \$\begingroup\$ This probably needs a description of an incircle and circumcircle. As for what you ask, personally I believe the best thing to do with challenges like this is to say "you must meet <arbitrary precision> for the test cases, work in general, but you need not handle floating point errors" or something similar. \$\endgroup\$ Apr 12 '20 at 5:00
  • \$\begingroup\$ Thanks for the suggestion. @FryAmTheEggman I'm not sure if I can give a good definition, so I linked the pages in wolfram mathworld. \$\endgroup\$
    – newbie
    Apr 12 '20 at 5:36
  • 1
    \$\begingroup\$ Since it is for triangles, the definitions can be fairly simple (even if finding them is still cryptic). In the interest of allowing people to understand what the challenge asks without requiring them to go to other websites, I think you can include a brief summary of the two definitions: incircle - the largest circle that fits inside the triangle, circumcircle - the circle that passes through each of the traingle's vertices. \$\endgroup\$ Apr 12 '20 at 15:37
  • \$\begingroup\$ Done. Thanks! @FryAmTheEggman \$\endgroup\$
    – newbie
    Apr 12 '20 at 15:46
  • \$\begingroup\$ I don't think this is a very interesting challenge, as there is very likely no other beautiful formula for such a thing as the distance between the incenter and the circumcenter, so this is 2.5 challenges in one: "find the incircle radius", "find the circumcircle radius" and "substitute both into this formula". \$\endgroup\$ Apr 13 '20 at 15:44
  • \$\begingroup\$ Hmm, but if you need to compute both the incircle radius and circumcircle radius, the formula can be simplified. (so the bytecount should be fewer than the sum of these two individual challenges, at least to me it's true) the last 0.5 is... for the context. @mypronounismonicareinstate \$\endgroup\$
    – newbie
    Apr 13 '20 at 15:46
  • 2
    \$\begingroup\$ I think something that might help with the challenge feeling like a few stapled together would be dropping the requirement that the two radii be included in the output. That way, golfing benefits from coupling are less hampered by needing to remember/store/output intermediate values. It is definitely possible that this won't fix the problem totally, but I think it should help. \$\endgroup\$ Apr 13 '20 at 16:01
  • \$\begingroup\$ Sounds like a good idea to me. I'll look into it tomorrow. (I'm a bit afraid if there're some much easier way to calculate that value alone) \$\endgroup\$
    – newbie
    Apr 13 '20 at 16:12
  • \$\begingroup\$ @newbie For what it's worth, here's what a mildly golfed formula for d alone looks like: Try it online! \$\endgroup\$
    – xnor
    Apr 13 '20 at 21:55
  • \$\begingroup\$ Thanks! I did some more golf and it's 66 bytes now. Would it be a better idea to output say \$R+r+d\$? \$\endgroup\$
    – newbie
    Apr 14 '20 at 1:29
  • \$\begingroup\$ Alright, I would go with output \$d\$. \$\endgroup\$
    – newbie
    Apr 14 '20 at 4:08
97 98
100 101

You must log in to answer this question.

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