# Code Golf: Mix the nuts so that none of the same kind are touching

Input:

Input is a randomized array of nuts (in your language), the possible nuts follow. Your program must have a way of representing each kind of nut, such as an integer code. Program must be able to handle any size array of any configuration of nuts.

Possible Nuts:

Kola nut
Mamoncillo
Maya nut
Mongongo
Oak acorns
Ogbono nut
Pili nut
Pistachio
Walnut


Output:

Output must be the array sorted in such a fashion that there are no adjacent nuts of the same kind. If this is impossible, the output should be an empty array.

Example Input (simplified):

["walnut", "walnut", "pistachio"]


Example Output:

["walnut", "pistachio", "walnut"]


Solutions may not simply shuffle the array until it becomes unique by chance. The sort employed must be a deterministic one

• "Your program must have a way of representing each kind of nut, such as an integer code" why is that? — "may not simply shuffle the array until it becomes unique by chance. The sort employed must be a deterministic one" a shuffle can still be deterministic. Do you just mean to impose a limit on the program's time complexity? – ceased to turn counterclockwis Jun 28 '12 at 7:50
• I have to agree with @leftaroundabout forbidding a particular algorithm is silly without a very good reason. One of the most rewarding things about code games like this is exactly the variety of methods that get employed. – dmckee --- ex-moderator kitten Jun 28 '12 at 14:37
• @dmckee, I think the requirement that the algorithm be deterministic is reasonable -- if the RNG is faulty or the input fairly long, a nondeterministic solution may fail to terminate. – boothby Jun 28 '12 at 20:27
• @boothby. Meh. I'm a particle physicist. Monte Carlo is a important tool in its own right. Moreover, If I choose a fixed PRNG and a fixed seed it is deterministic. – dmckee --- ex-moderator kitten Jun 28 '12 at 20:32
• I think I found an example that has several solutions, but may cause some answers to fail to find any of them. Can I add it? (5,4,4,3,3,2) perl6 -e 'my @a="aaaaabbbbccccdddee".comb;my @b = @a.pick(*) while @b.squish !== @a;say [~] @b' baedcbdacdecbabaca (3,3,2) may cause them to fail also. – Brad Gilbert b2gills Nov 19 '15 at 22:18

### GolfScript, 424137 38 characters

~.{\{=}+%1-,}+$.,)2//zip[]*.2<..&=*p  The code expects input on STDIN and prints result to STDOUT, e.g.: > ["walnut" "walnut" "walnut" "macadamia" "pistachio"] ["walnut" "macadamia" "walnut" "pistachio" "walnut"] > ["walnut" "walnut" "walnut" "macadamia" "walnut"] []  The script became longer than expected but I suppose there is room for improvement. Edit: The case of a list with a single item costs me 1 character (the best comparison I could come up with is the same as Peter's). • I hadn't sat down to implement this yet, but $.,)2//zip is exactly what I had in mind. My interpretation of the spec was that it could take input on the stack and leave it on the stack, so maybe we should push for clarification. – Peter Taylor Jun 28 '12 at 18:03
• @PeterTaylor, cool. Works for me. – boothby Jun 29 '12 at 7:39
• This crashes on input ["walnut"] in the compare-the-first-two section. – Peter Taylor Jun 29 '12 at 21:21
• @PeterTaylor You're right. I'll have to work on that corner case. – Howard Jun 30 '12 at 5:04

~:x{]x\-,}$.,)2//zip[]*.2<..&=*  Same input and output format as Howard's solution. • I had the same idea on the sort part but didn't code it up yet :-) Good work! – Howard Jun 30 '12 at 5:00 # Brachylog v2, 10 bytes p.¬{s₂=}∨Ė  Try it online! Brute-force solution. (This is a function, allowed because the challenge does not say "full program".) It's also mostly a direct translation of the spec (the only real subtlety is that I managed to arrange things so that all the implicit constraints arrived in exactly the right places, thus not needing any extra characters to disambiguate them). Note that this is a generic algorithm for rearranging any sort of list so that it does not have two touching elements; it can handle string representations of the elements, and it can handle integer codes just as well. So it doesn't really matter how the "Your program must have a way of representing each kind of nut, such as an integer code." requirement from the question is interpreted. ## Explanation p.¬{s₂=}∨Ė p Find a permutation of {the input} ¬{ } which does not have the following property: s₂ it contains a pair of adjacent elements = that are equal ∨ {no constraint on what value the equal elements can have} . If you find such a permutation, output it. ∨ If no permutation is found, ignore the input and Ė {output} an empty list  ## J, 80 characters ]_:@.(0<2&([:+/=/\))({.~-:@#),((],.|.)~>.@-:@#)<"1;(\:#&.>)(</.])[;.1' ',1!:1[1  Not really in the same league as Golfscript on this one. I suspect there are gains to be made, but the 14 characters needed just to get the list into the program [;.1' ',1!:1[1 is a major handicap. Basically the program takes in the list, groups similar items together, sorts by number of items in each group descending, and alternates the output between the first half and the second half of the list. The rest if the code gets rid of extraneous items and decides if the list is valid output (outputting infinity _ if it isn't). Example: macadamia walnut walnut pistachio walnut  group (</.]): macadamia walnut walnut walnut pistachio  sort (\:#&.>): walnut walnut walnut macadamia pistachio  ravel ((],.|.)~>.@-:@#): walnut macadamia walnut pistachio walnut  # Jelly, 14 bytes Ġz0UẎḟ0ịµẋ⁻ƝẠ$


Try it online!

The last 6 bytes can be removed if we can have undefined behavior for invalid inputs.

# Stax, 10 bytes

│éÿ∞å[zàL⌂


Run and debug it

Here's the same program unpacked, ungolfed, and commented.

|T      get all permutations
{       block to filter by
:g_=  after dropping repeated elements, it's still equal
f       execute filter
|c      terminate and pop if falsy (no match)
hJ      take the first permutation, and join with spaces


Run this one