AMO - Advanced Modeling and Optimization, Volume 19, Number 2, 2017

Mathematical Approach for N-Queens Problem with

Isomorphism

Sanjay Jain
Department of Mathematical Sciences, SPC Government College, Ajmer
Affiliated to M. D. S. University Ajmer, Ajmer-305001, India
E-mail drjainsanjay@[Link]

Chander Shakher
Research Scholar, Mahatma Gandhi University, Meghalaya
E-mail csghorela200921@[Link]

Abstract
The N-Queens problem is commonly used to teach the programming technique of
backtrack search. The N-Queens problem may also be used to illustrate the important
concept of isomorphism. Here, we study how the N-Queens problem can be used as a
vehicle to teach the concepts of isomorphism, transformation groups or generators, and
equivalence classes. We also indicate how these ideas can be used in a programming
exercise and compute non-isomorphic solutions.

Keyword: N-Queens, Isomorphism, Transformation, Non-isomorphic solutions.

1. Introduction
The 8-Queens problem or more general N-Queens problem is often used to explicate
backtracking in computing courses. In this paper we want to point out that N-Queens can also
be used as a vehicle for teaching the ideas of isomorphism and transformation groups. First
we provide the brief introduction of terminology used in this work.

 Backtracking
Bitner & Reingold (1975) studied backtrack programming problem. “Backtracking is a
general algorithm for finding all (or some) solutions to some computational problem that
incrementally builds candidates to the solutions, and abandons each partial candidate c
("backtracks") as soon as it determines that c cannot possibly be completed to a valid
solution.”

AMO - Advanced Modeling and Optimization. ISSN: 1841-4311

 Sanjay Jain and Chander Shakher

 Eight queens puzzle

The eight queens puzzle is the problem of placing eight chess queens on an 8×8 chessboard
so that no two queens attack each other. Thus, a solution requires that no two queens share
the same row, column, or diagonal. The eight queens puzzle is an example of the more
general n-queens problem of placing n queens on an n × n chessboard, where solutions exist
for all natural numbers n with the exception of n = 2 and n = 3. This puzzle provides a
classical example in the field of computation for studying recursion, depth-first search and
backtracking.
Clay (1986) and Goldsby (1987) provides solutions in different manner for 8 queens
problem. The eight queens puzzle has 92 distinct solutions. If solutions that differ only by
symmetry operations (rotations and reflections) of the board are counted as one, the puzzle
has 12 unique (or fundamental) solutions.

Figure 1(1.1 & 1.2): Two solutions to the 8-Queens problem.

 The N-Queens Problem

The objective in the N-Queens problem is to place N queens on an N X N chessboard such

that no two queens can attack one another under the normal rules of chess. A queen may
attack any other piece lying along the row, column, or diagonal containing the queen. Bruen
& Diyon (1975) also studied N-Queens problem. Demirors & Tanik (1991) works on queens
with magic squares.
The N-Queens problem is often stated as: Can N queens are placed on an N X N chessboard
so that no queen can attack another queen. In this form the answer is easy: if N not belongs to
310
 Mathematical Approach for N-Queens Problem with Isomorphism

{2,3} say Yes. Further, it is easy to find a non-attacking placement for the queens. Abramson
& Yung (1989) and Hoffman, Loessi & Moore (1969) also provides solutions in different
manner for N-Queens problem.
Two solutions to the 8-Queens problem are shown in Figure 1. Since any solution to N-
Queens must have exactly one queen per row (and column) of the chessboard, a solution such
as Figure 1.1 can be expressed as the vector (2,4,6,8,3,1,5,7) i.e. the row which the queen in
each column occupies, Similarly, solution 1.2 can be notated as (1,7,4,6,8,2,5,3).
The construction of an enumerating backtracking program seems to be necessary to find all
the non-attacking placements. Even finding the number of such placements seems to be
difficult.
The point here is that the N-Queens problem gives us a chance to discuss isomorphic and
non-isomorphic solutions. Further, by a small addition to the standard N-Queens program, we
can produce a program which outputs only the non-isomorphic solutions. In addition, this
exercise will also introduce the use of transformation groups.

Figure 2: Two Isomorphic Graphs.

 Isomorphism

Two objects are isomorphic when they have equal forms, that is, in some sense the objects
are identical. The sense is, of course, important. For example, a metal paper clip and a plastic
paper clip are isomorphic when I want to clip papers together, but they are not isomorphic
when I want to play with my magnetic paper clip holder. In more complicated cases, the
objects are made up of parts and the sense of sameness includes some of the
interrelationships between the parts. For example, I might consider bees and mosquitoes as
isomorphic in producing red welts on humans, but when I consider the body parts, the bee

311
 Sanjay Jain and Chander Shakher

and the mosquito are not isomorphic because their business ends are different.

In mathematical study, the idea of isomorphism is usually introduced in algebra, For

example, the field of real numbers is isomorphic to the field of complex numbers with zero
imaginary part, or two semi groups (S, +) and (G,*) are isomorphic when there is an
invertible function

h: S G so that h(s1+s2) = h(s1)*(s2) for all s1 and s2.

Computer scientists are usually introduced to isomorphism in the context of graphs. Two
graphs G1 and G2 are isomorphic if there is an invertible function h which maps each vertex
of G1 to a vertex of G2 so that adjacent vertices of G1 map to adjacent vertices of G2 and
vice versa, as in Figure 2.

 Isomorphism and the N-Queens Problem

For solutions to the N-Queens, we want the renaming to preserve the relationships between
the queens. What transformations should be allowed? Clearly mirror image, mentioned
above, should be included, but it seems that several mirror images are possible. The obvious
ones place a mirror parallel to one side of the board, but what about reflecting across one of
the diagonals of the board? Other allowed transformations should include rotations by
multiples of 90º (π/2 radians). Obviously one could combine rotations and mirror images to
get other transformations. This combining is function composition since each transformation
takes a solution as input and gives a solution as output. Function composition is associative.
Associativity means that if T1, T2, and T3 are three transformations and T1.T2 is the
composite transformation obtained by applying T2 and then T1, then the following equation
holds (T1.T2).T3 = T1.(T2.T3).

Further, the composition of any allowed transformations always gives an allowed

transformation, so the set of transformations is closed under composition. There is a special
transformation, 'do nothing to the board', which is the identity transformation. Finally, each
transformation can be undone, that is, each transformation has an inverse. Each mirror image
is its own inverse. For any rotation by a multiple of 90 degree, applying the rotation 3 times
will give the inverse of the rotation. So for any sequence of rotations and mirror images, there
is an inverse which can be formed by taking the inverses of the transformations in reverse
order.
312
 Mathematical Approach for N-Queens Problem with Isomorphism

A set with an operation that is associative, closed, and has an identity and an inverse is, of
course, a group. In fact, the allowed transformation for the N-Queens is the group of
transformations which transform a square into itself. Consider a board with corners labelled
A, B, C, and D in clockwise order. When a transformation is applied, the corner labelled A
can be mapped to any of 4 positions, and the next corner clockwise after A must be either B
or D. So there are 4*2 = 8 different transformations. Chandra (1974) studied for independent
transformations.
The set of transformations for N-Queens is a dihedral group. A dihedral group ∆n is defined
as the group of symmetries of a regular polygon Pn of n sides. Elements of ∆n can be
obtained by the operations of rotation R through 360º /n, and the operation reflection M about
some side:
2 3 n
I R R R …R

Using the operations of reflection (M)

2 3 n
M RM R M R M…R M

For N-Queens the appropriate group is ∆4. Using the operations of reflection (M), and 90º
rotation ( R ), one solution can be transformed into seven other solutions.

 Generating Non-isomorphic N-Queens Solutions

An interesting problem is finding the number of non-isomorphic solutions for each value of
N, as well as a method of quickly generating these non-isomorphic solutions. It might seem
easy to find the number of non-isomorphic solutions to N-Queens by finding the total number
of solutions and dividing by 8. Unfortunately this doesn't work because there are solutions
which can be mapped to themselves by some of the transformations.

For example, the solution (2,4,6,8,3,1,5,7) to 8-Queens (Figure 1.1) is transformed to itself
after two rotations:

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

Using the operations of reflection (M)

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

Below is a table showing the total number of solutions to N-Queens, as well as the number of
non-isomorphic solutions to the N-Queens problem for various values of N:

313
 Sanjay Jain and Chander Shakher

Number of non isomorphic

N Total number of solution
solution

0 0 0

1 1 1

2 0 0

3 0 0

4 2 1

5 10 2

6 4 1

7 40 6

8 92 12

9 352 46

10 724 92

11 2680 341

12 14200 1787

13 73712 9233

14 365596 45752

15 2279184 285053

2. Mathematical Approach

While we want to be able to count the number of non-isomorphic solutions, we also want to
find one solution for each isomorphism class. Two solutions are in the same isomorphism
class exactly when some allowed transformation transforms one solution to the other. It is
probably worth mentioning at this point that being isomorphic is an equivalence relation, and
reminding students that an equivalence relation is reflexive, symmetric, and transitive. As an

314
 Mathematical Approach for N-Queens Problem with Isomorphism

easy exercise you can ask the students to find which group property implies reflexive, which
group property implies symmetric, and which group property implies transitive.

Since there are only 8 transformations to consider, a simple way to find non-isomorphic
solutions is to maintain a set of the non-isomorphic solutions found so far, generate all
solutions, and then determine if applying any of the 8 transformations gives a solution
already in the set. If each of the 8 transformations gives a solution not in the set, then the new
solution should be added to the set of non-isomorphic solutions.

Unfortunately this method has the drawback that there may be a very large (more than
exponential) number of non-isomorphic solutions in the set. So comparing a new solution
with the non-isomorphic solutions can take a very long time. Luckily, there is a shortcut. In
the backtracking algorithm, the first queen is placed in the first allowed square in the first
column, then the second queen is placed in the first allowed position in the second column,
and so forth. This means that the solutions will be generated in order if we consider each
solution of N-Queens as a base N+1 number. Hence a solution is isomorphic to a previously
found solution if and only if one of the 8 transformations produces a solution which is less
than the present solution.

3. Mathematical Formulation of equation

The less than relation may be stated as:

Let S1 = (a1,a2,a3…,an) and S2 = (b1,b2,b3…,bn)

which are two solutions to the N-Queens, so that each ai,bi ϵ {1,2,…,n}.

Then S1 < S2 iff (a1a2a3…an) < (b1b2b3…bn),

where: (a1a2a3…an) < (b1b2b3…bn) iff a1<b1, in the usual ordering 1< 2 <…< n

or, a1 = b1 and (a2a3…an) < (b2b3…bn)

…
…
to bottom out the recursion, we have (an) < (bn) iff an < bn in the usual ordering.

So given a solution S, one should add S to the set of non-isomorphic solutions when for each
of the seven non-identity transformations T1,T2,…,T7,

S ≤ Ti(S), for i= 1,2,…,7.

315
 Sanjay Jain and Chander Shakher

4. Programming Algorithm

Biernat (1993), Clapp, Trevor & Volz (1986), Cockayne & Hedetniemi (1986), Erbas &
Tanik (1992), Sosic & Gu (1990) and many researchers gives different algorithms for
different queens problem. Here we prepared an algorithm to check isomorphic and find
generator from a queens problem and non-isomorphic solutions.

/*
 check_if_isomorphic: If the solution is an isomorphic to a previously generated solution,

return TRUE. Return FALSE if solution is new.

*/

BOOLEAN check_if_isomorph(int original[MAX])

int i;
 int transformed[MAX];

BOOLEAN iso_flag;

for (i = 0; i < size; i++) //make a copy of the solution vector

transformed[i] = original[i];

for (i=0;i<7;i++) //generated the 7 transformation

if(i!=3)

Rotate(transformed);
else
Mirror_image(transformed);
 iso_flag = compare_vector(original, transformed);

if (! iso_flag) return TRUE;

}
 return FALSE;

316
 Mathematical Approach for N-Queens Problem with Isomorphism

The generators R and M are not the only possible generators. Let F be the transformation
which flips the square across its counter-diagonal. Then F and M are a set of generators for
∆4, and FMFMFMF is a sequence of these generators which will generate the whole
transformation group. Showing that the previous statement is true would b e a reasonable
exercise to see if your students have followed the development. You also might want them to
decide which set of generators is easier to program.

In several texts, the fact that the queen in the first column never has to be placed in the
second half of the column if one is only interested in non-isomorphic solutions is mentioned,
as Horowitz and Sahni (1979).

"Observe that for finding in equivalent solutions the algorithm need only set
X (I) = 2,3,…,[n/2]."

Unfortunately this has been widely misinterpreted by users to mean that this restriction alone
is sufficient to generate only non-isomorphic solutions. We hope that the above description
has been sufficient to explain the actual scenario.

5. Conclusion

The major point of this note is that N-Queens is a good example which is typical of
combinatorial enumeration problems. The typical features are:

1 solutions are generated by backtracking algorithms

2 ordering on solutions is imposed by the generating algorithm

3 desire for non-isomorphic solutions

4 group of transformations indicating which solutions are isomorphic

5 group expressible by simple generators

6 "on-the-fly" non-isomorphism test by applying a sequence of generators to

a solution and checking that all the transformed solutions are ≥ the solution in
question.

Cull & De Curtins (1978) studied Knight's Tour Revisited, which is another problem with
similar characteristics of N queens problem. Knight's Tour Problem (where the
transformation group consists of rotation, reflection, and "take a path backwards") is open for
further research in isomorphism approach.
317
 Sanjay Jain and Chander Shakher

References

[1] Abramson, Bruce and Yung, Moti, (1989) "Divide and Conquer under Global
Constraints: A Solution to the N-Queens Problem", Journal of Parallel and Distributed
Computing, 6, 649-662.

[2] Bell, Jordan, and Brett Stevens. (2009) "A Survey of Known Results and Research
Areas For N-Queens." Discrete Mathematics 309.1: 1-31.

[3] Bessière, Christian, Et Al. (2005) "Asynchronous Backtracking Without Adding

Links: A New Member In The Abt Family." Artificial Intelligence 161.1: 7-24.

[4] Biernat, Martin J., (1993) "Teaching Tools for Data Structures and Algorithms",
ACM SIGCSE Bulletin, 25, 9-12.

[5] Bitner, James R., and Reingold, Edward M., (1975) "Backtrack Programming
Techniques", Communications of the ACM, 18, 651-656.

[6] Bruen, A., and Dixon, R., (1975) "The n-Queens Problem" Discrete Mathematics,
ACM SIGCSE Bulletin, 12:393-395.
[7] Chandra, A. K., (1974) "Independent Permutations as Related to a Problem of Moser
and a Theorem of Polya." Journal of Combinatorial Theory, Series A, 16, 111–120.

[8] Clapp, Russell M., Mudge, Trevor N., and Volz, Richard A., (1986) "Solutions to the
n Queens Problem Using Tasking in Ada."., ACM SIGPLAN Notices, 21, 99-110.

[9] Clay, C., (1986) "A New Solution to the N≤8 Queens Problem", ACM SIGPLAN
Notices, 21: 28-30.

[10] Cockayne, E. J., and Hedetniemi, S. T., (1986) "On the Diagonal Queens
Domination Problem" Journal of Combinatorial Theory, 42, 137-139.

[11] Cull, Paul, and De Curtins, J., (1978) "Knight's Tour Revisited," Fibonacci
Quarterly 16: 276–285.

[12] Demirors, O. and Tanik, M. M., (1991) "Peaceful Queens and Magic Squares"
Technical Report 91-CSE-7, Dept. of Comp. Sci. and Eng., Southern Methodist
University.

[13] Erbas, Cengiz, and Tanik, Murat M., (1992) "N-Queens Problem and its
Algorithms", Technical Report 91-CSE-8, Dept. of Comp. Sci. and Eng., Southern
Methodist University.

[14] Goldsby, Michael E., (1987) "Solving the 'N≤8 Queens' Problem with CSP and
Modula", ACM SIGPLAN Notices, 43-52.

318
 Mathematical Approach for N-Queens Problem with Isomorphism

[15] Gutiérrez-Naranjo, Miguel A., Et Al. (2009) "Solving The N-Queens Puzzle With P
Systems." Seventh Brainstorming Week On Membrane Computing 1: 199-210.

[16] Hoffman, E., J., Loessi, J. C., and Moore, R. C., (1969) "Constructions for the
Solution of the m Queens Problem", Mathematics Magazine, XX, 66-72.

[17] Horowitz, E., and Sahni, S., (1979) “Fundamentals of Computer Algorithms”,
Computer Science Press, Rockville, Maryland.

[18] Lijo, V. P. and Jasmin T. Jose (2015) “ Solving N-Queen Problem by Prediction”
International Journal of Computer Science and Information Technologies, Vol. 6 (4) ,
3844-3848

[19] Sosic, Rok, and Gu, Jun, (1990) "A Polynomial Time Algorithm for the N-Queens
Problem", SIGART Bulletin 1(3), 7-11.

319