Group Therapy: Identify Groups

Question

Write a program, that determines whether the multiplication table of the given finite magma represents a group. A magma is a set with a binary operation that is closed, that means

• for all a,b in G, a*b is again in G (Closedness)

Let (G, * ) be a magma. (G, * ) is a group if

• for all a,b,c in G, (a*b) * c = a * (b*c) (Associativity)
• there exists an element e in G such that e * a = a * e = a for all a in G (Existence of neutral Element)
• for all a in G there is a b in G such that a * b = b * a = e where e is the neutral element (Existence of Inverse)

Specs

The input is of a string of n^2-1 characters (one character for each element of the magma, allowed are 0-9,a-z) and just represents the table read row by row, ommiting the operator name. You can assume that the input represents a valid magma (that means each of the elements appers exactly once in the header row/colum).

Example: Here we have the table of Z_4

+ | 0 1 2 3
-----------
0 | 0 1 2 3
1 | 1 2 3 0
2 | 2 3 0 1
3 | 3 0 1 2

The input string will be 012300123112302230133012. (Or if we use symbols it could also be nezdnnezdeezdnzzdneddnez). Be aware that the sequence of the elements in the row and in the column do not have to be the same, so the table of Z_4 could also look like so:

+ | 1 3 2 0
-----------
1 | 2 0 3 1
0 | 1 3 2 0
2 | 3 1 0 2
3 | 0 2 1 3

This also means that the neutral element is not necessarily in the first column or first row.

If it is a group, the program has to return the character representing the neutral element. If not it has to return a falsy (distinct from the values 0-9 a-z) value

Test cases

Non-groups can easily be constructed by just altering one digit of the string or by artificially altering the tables defining a operation that contradicts one of the group axioms.

Groups

Trivial

* | x
-----
x | x

xxx

Neutral Element: x

H (quaternion group)

* | p t d k g b n m 
-------------------
m | b d t g k p m n 
p | m k g d t n p b 
n | p t d k g b n m 
b | n g k t d m b p 
t | g m n p b k t d 
d | k n m b p g d t 
k | t b p m n d k g 
g | d p b n m t g k 

ptdkgbnmmbdtgkpmnpmkgdtnpbnptdkgbnmbngktdmbptgmnpbktddknmbpgdtktbpmndkggdpbnmtgk

Neutral Element: n

D_4

* | y r s t u v w x
-------------------
u | u x w v y t s r
v | v u x w r y t s
w | w v u x s r y t
x | x w v u t s r y
y | y r s t u v w x
r | r s t y v w x u
s | s t y r w x u v
t | t y r s x u v w


yrstuvwxuuxwvytsrvvuxwrytswwvuxsrytxxwvutsryyyrstuvwxrrstyvwxusstyrwxuvttyrsxuvw

Neutral Element: y

Z_6 x Z_2

x | 0 1 2 3 5 7 8 9 a b 4 6
---------------------------
0 | 0 1 2 3 5 7 8 9 a b 4 6 
1 | 1 2 3 4 0 8 9 a b 6 5 7 
2 | 2 3 4 5 1 9 a b 6 7 0 8 
7 | 7 8 9 a 6 2 3 4 5 0 b 1 
8 | 8 9 a b 7 3 4 5 0 1 6 2 
9 | 9 a b 6 8 4 5 0 1 2 7 3 
a | a b 6 7 9 5 0 1 2 3 8 4 
b | b 6 7 8 a 0 1 2 3 4 9 5 
3 | 3 4 5 0 2 a b 6 7 8 1 9 
4 | 4 5 0 1 3 b 6 7 8 9 2 a 
5 | 5 0 1 2 4 6 7 8 9 a 3 b 
6 | 6 7 8 9 b 1 2 3 4 5 a 0 

01235789ab46001235789ab4611234089ab6572234519ab67087789a623450b1889ab7345016299ab684501273aab6795012384bb678a0123495334502ab67819445013b67892a5501246789a3b66789b12345a0

Neutral Element: 0

A_4

* | i a b c d e f g h j k l
---------------------------
i | i a b c d e f g h j k l
a | a b i e c d g h f l j k
b | b i a d e c h f g k l j
c | c f j i g k a d l b e h
d | d h k b f l i e j a c g
e | e g l a h j b c k i d f
f | f j c k i g d l a h b e
g | g l e j a h c k b f i d
h | h k d l b f e j i g a c
j | j c f g k i l a d e h b
k | k d h f l b j i e c g a
l | l e g h j a k b c d f i

iabcdefghjkliiabcdefghjklaabiecdghfljkbbiadechfgkljccfjigkadlbehddhkbfliejacgeeglahjbckidfffjckigdlahbegglejahckbfidhhkdlbfejigacjjcfgkiladehbkkdhflbjiecgalleghjakbcdfi

Neutral Element: i

Non-Groups

A loop (Group missing associativity, or a Quasi-Group with neutral element)

* | 1 2 3 4 5
-------------
1 | 1 2 3 4 5 
2 | 2 4 1 5 3 
3 | 3 5 4 2 1 
4 | 4 1 5 3 2 
5 | 5 3 2 1 4

12345112345224153335421441532553214

Neutral Element: 1
(2*2)*3 = 4*3 = 5 != 2 = 2*1 = 2*(2*3)

An IP-loop (from http://www.quasigroups.eu/contents/download/2008/16_2.pdf)

* | 1 2 3 4 5 6 7
-----------------
1 | 1 2 3 4 5 6 7
2 | 2 3 1 6 7 5 4
3 | 3 1 2 7 6 4 5
4 | 4 7 6 5 1 2 3
5 | 5 6 7 1 4 3 2
6 | 6 4 5 3 2 7 1
7 | 7 5 4 2 3 1 6

123456711234567223167543312764544765123556714326645327177542316

Neutral Element: 1
2*(2*4) = 2*6 = 5 != 7 = 3*4 = (2*2)*4

Monoid (by Quincunx, thanks!)

Monoids are Magmas with associativity and a neutral element.

* | 0 1 2 3
-----------
0 | 0 1 2 3
1 | 1 3 1 3
2 | 2 1 0 3
3 | 3 3 3 3

012300123113132210333333

Neutral Element: 0

Another Monoid

(Multiplication mod 10, without the 5) We obviously do not have inverses, and the associativity is given by the multiplication modulo 10.

* | 1 2 3 4 6 7 8 9
-------------------
1 | 1 2 3 4 6 7 8 9
2 | 2 4 6 8 2 4 6 8
3 | 3 6 9 2 8 1 4 7
4 | 4 8 2 6 4 8 2 6
6 | 6 2 8 4 6 2 8 4
7 | 7 4 1 8 2 9 6 3
8 | 8 6 4 2 8 6 4 2
9 | 9 8 7 6 4 3 2 1

Neutral Element: 1   12346789112346789224682468336928147448264826662846284774182963886428642998764321

Answer 1

Octave, 298 290 270 265 chars

function e=g(s)
c=@sortrows;d=a=c(c(reshape(a=[0 s],b=numel(a)^.5,b)')');
for i=2:b a(a==a(i))=i-1;end;
a=a(2:b,2:b--);u=1:b;
e=(isscalar(e=find(all(a==u')))&&a(e,:)==u&&sum(t=a==e)==1&&t==t')*e;
for x=u for y=u for z=u e*=a(a(x,y),z)==a(x,a(y,z));end;end;end;e=d(e+1);

265: Removed unnecessary function handle.

270: After all, the check that e==h for e always satisfying e·a=a and h always satisfying a·h=a was not necessary. This is not possible for them to be different (e·h=?).

The details from the explanation for solution below are still relevant.

---

290:

function e=g(s)
c=@sortrows;d=a=c(c(reshape(a=[0 s],b=numel(a)^.5,b)')');
for i=2:b a(a==a(i))=i-1;end;
a=a(2:b,2:b--);u=1:b;
s=@isscalar;e=(s(e=find(all(a==u')))&&s(h=find(all(a'==u')'))&&sum(t=a==e)==1&&t==t')*e;
for x=u for y=u for z=u e*=a(a(x,y),z)==a(x,a(y,z));end;end;end;e=d(e+1);

The first line

c=@sortrows;d=a=c(c(reshape(a=[0 s],b=numel(a)^.5,b)')'); just stores the input into n x n table (with zero character at the place of operation mark), and then lexicographically sorts columns and rows, so that rows and columns receive the same order:

+ | z a t b                        + | a b t z
-----------                        -----------
z | t b a z         becomes        a | t a z b
b | z a t b      ============>     b | a b t z
t | a z b t                        t | z t b a
a | b t z a                        z | b z a t

Now, I remap "a","b","t","z"to standard 1, 2, 3, 4 , so that I can index the table efficiently. This is done by the line for i=2:b a(a==a(i))=i-1;end;. It yields table like

0   1   2   3   4
1   3   1   4   2
2   1   2   3   4
3   4   3   2   1
4   2   4   1   3

, where we can get rid of the first row and column with a=a(2:b,2:b--);u=1:b;:

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

This table has the given properties:

• if the neutral element e exists, exactly one (isscalar) row and one column have the value of row vector u=[1 2 3 ... number-of-elements]:

s=@isscalar;e=(s(e=find(all(a==u')))&&s(h=find(all(a'==u')'))&&...

• if each element a has a reverse element a', two things hold: the neutral element e occurs only once each column and only once each row (sum(t=a==e)==1) and, to satisfy a'·a = a·a', the occurences of e are symmetrical in respect to translation t==t'

• a·b can be retrieved by simple t(a,b) indexing. Then we check the associativity in the boring loop:

for x=u for y=u for z=u e*=a(a(x,y),z)==a(x,a(y,z));end;end;end;

The function returns the neutral element the way it appeared in original table (e=d(e+1)) or nil character if the table does not describe a group.

Answer 2

Ruby, 401 ... 272

f=->s{n=(s.size+1)**0.5
w=n.to_i-1
e=s[0,w].split''
s=s[w,n*n]
m={}
w.times{(1..w).each{|i|m[s[0]+e[i-1]]=s[i]}
s=s[n,n*n]}
s=e.find{|a|e.all?{|b|x=m[a+b]
x==m[b+a]&&x==b}}
e.all?{|a|t=!0
e.all?{|b|x=m[a+b]
t||=x==m[b+a]&&x==s
e.all?{|c|m[m[a+b]+c]==m[a+m[b+c]]}}&&t}&&s}

This is my first ruby program! This defines a lambda function that we can test by doing puts f[gets.chomp]. I return false for my false value. The first half of the function is simply parsing the input into a map, then the second half checks the possibilities.

f=->s{
    n=((s.size+1)**0.5).to_i
    w=n-1
    e=s[0,w].split'' # create an array of elements of the potential group
    s=s[w,n*n]
    m={} # this map is what defines our operation
    w.times{
        (1..w).each{               # for each element in the row of the table
            |i|m[s[0]+e[i-1]]=s[i] # put the value into the map
        }
        s=s[n,n*n]
    }
    s=e.find{|a| # s is the identity
        e.all?{|b|
            x=m[a+b]
            x==m[b+a]&&x==b # is a the identity?
        }
    }
    e.all?{|a| # implicit return statement
        t = !0 # t = false
        e.all?{|b| # check for inverses
            x=m[a+b]
            t ||= x==m[b+a]&&x==s # t is now true if b was a's inverse
            e.all?{|c|
                m[m[a+b]+c]==m[a+m[b+c]] # check associativity
            }
        } && t
    }&&s
}

Answer 3

JavaScript (ES6) 285 243 278

Run snippet to test (being ES6 it works only on Firefox)

Edit 2 Bug fix. I was wrong in finding the neutral element, checking just one way. (Need better test cases!!!)

Edit Using simpler string concatenation instead of double index (like @Quincunx), I don't know what I was thinking. Also, simplified inverse check, it should still work.

F=t=>(
  e=t.slice(0,d=Math.sqrt(t.length)|0),
  t=t.slice(d).match('.'.repeat(d+1),'g'),
  t.map(r=>{
    for(v=r[i=0],
        j=e.search(v)+1, // column for current row  element
        r!=v+e|t.some(r=>r[j]!=r[0])?0:n=v; // find neutral
        c=r[++i];
       )h[v+e[i-1]]=c
  },h={},n=''),
  e=[...e],!e.some(a=>e.some(b=>(
    h[a+b]==n&&--d, // inverse
    e.some(c=>h[h[a+b]+c]!=h[a+h[b+c]]) // associativity
  )
  ))&&!d&&n
)

input { width: 400px; font-size:10px }

Click on textbox to test - Result : <span id=O></span><br>
<input value='...' onclick='O.innerHTML=F(this.value)'> (?)
<br>Groups<br>
<input value='nezdnnezdeezdnzzdneddnez' onclick='O.innerHTML=F(this.value)'> (n)<br>
<input value='ptdkgbnmmbdtgkpmnpmkgdtnpbnptdkgbnmbngktdmbptgmnpbktddknmbpgdtktbpmndkggdpbnmtgk' onclick='O.innerHTML=F(this.value)'> (n)<br>
<input value='yrstuvwxuuxwvytsrvvuxwrytswwvuxsrytxxwvutsryyyrstuvwxrrstyvwxusstyrwxuvttyrsxuvw' onclick='O.innerHTML=F(this.value)'> (y)<br>
<input value='01235789ab46001235789ab4611234089ab6572234519ab67087789a623450b1889ab7345016299ab684501273aab6795012384bb678a0123495334502ab67819445013b67892a5501246789a3b66789b12345a0'onclick='O.innerHTML=F(this.value)'> (0)<br>
Non groups <br>
<input value='12345112345224153335421441532553214' onclick='O.innerHTML=F(this.value)'> (FAIL)<br>
<input value='123456711234567223167543312764544765123556714326645327177542316' onclick='O.innerHTML=F(this.value)'> (FAIL)<br>
<input value='012300123113132210333333' onclick='O.innerHTML=F(this.value)'> (FAIL)<br>

Answer 4

Haskell 391B

import Data.Maybe
import Data.List
o a b=elemIndex b a
l£a=fromJust.o a$l
a§b=[a!!i|i<-b]
f s|isJust j&&and(map(isJust.o h)s)&&and[or[p%q==e|q<-h]&&and[p%(q%r)==(p%q)%r|q<-h,r<-h]|p<-h]=[e]|True="!"where n=floor$(sqrt(fromIntegral$length s+1))-1;h=take n s;g=[s§[a..b]|(a,b)<-zip[1+n,2+n+n..][n+n,3*n+1..(n+1)^2]];v=s§[n,1+2*n..n+n*n];a%b=g!!(b£v)!!(a£h);j=o g h;e=v!!fromJust j

Curse those imports!

import Data.Maybe
import Data.List

{- rename elemIndex to save characters -}
o a b=elemIndex b a

{- get the index of l in a -}
l£a=fromJust.o a$l

{- extract a sublist of a with indices b -}
a§b=[a!!i|i<-b]

f s |isJust j {-Identity-}
     &&and (map (isJust.o h) s) {-Closure-}
     &&and[
        or [p%q==e|q<-h] {-Inverse-}
        && and [ p%(q%r)==(p%q)%r | q<-h,r<-h ] {-Associativity-}
     |
        p<-h
     ]=[e]
    |True="!"
    where
    {-size-}    n=floor$(sqrt(fromIntegral$length s+1))-1
    {-horiz-}   h=take n s
    {-table-}   g=[s§[a..b]|(a,b)<-zip[1+n,2+n+n..][n+n,3*n+1..(n+1)^2]]
    {-vert-}    v=s§[n,1+2*n..n+n*n]
    {-operate-} a%b=g!!(b£v)!!(a£h)
                j=o g h {-index of the first row identical to the top-}
    {-ident-}   e=v!!fromJust j

Explanation

f::String->String maps the string to either e::Char, the identity element, or !.

The where clause creates a bunch of variables and functions, which I've commented; v::[Int] is the vertical list of elements, h::[Int] the horizontal one.

%::Char->Char->Char applies the group operation to its arguments.

g::[[Int]] is the group table (for dereferencing using %)

j::Maybe Int contains the index of the identity in v if it exists, otherwise Nothing, which is why isJust j is the condition in f for identity.