Symmetries of plane partitions and the permanent-determinant method (Q1336443)

A partition may be viewed by its Ferrers diagram---a finite collection of unit squares in the first quadrant that is stable under gravitational attraction towards the origin. Plane partitions are the three-dimensional analogue. A plane partition in an \(a\times b\times c\) box is a collection of unit cubes in the rectangular solid \([0,a]\times [0,b]\times [0,c]\) which is stable under gravitational attraction towards the origin. The analogue of a self-conjugate partition is the symmetric plane partition. \textit{R. P. Stanley} [J. Comb. Theory, Ser. A 43, 103-113 (1986; Zbl 0602.05007)] and Robbins observed that there are ten symmetry classes of plane partitions in a box and conjectured that the number of plane partitions in each symmetry class is given by a certain product of factorials. These conjectures are considered to be among the most difficult problems in enumerative combinatorics. This paper together with the papers of \textit{G. E. Andrews} [J. Comb. Theory, Ser. A 66, No. 1, 28- 39 (1994; Zbl 0797.05003)] and \textit{J. R. Stembridge} [Adv. Math. 111, No. 2, 227-243 (1995)] completes the program of proving all ten formulas. The author proves the remaining case. He proves that the number of cyclically symmetric, self-complementary plane partitions in \([0, 2a]\times [0, 2a]\times[0,2a]\) is \[ \prod^{a- 1}_{i= 0} \Biggl({(3i+ 1)!\over (a+ i)!}\Biggr)^ 2. \] Many techniques have been developed to tackle these ten conjectures. However, so far there has been no uniform proof although the formulas look similar. The proof usually involves using the permanent-determinant method or the method of non-intersecting paths to write the number of plane partitions in a given symmetry class as a determinant or a Pfaffian. Then the difficult problem of evaluating the determinants remains. The permanent-determinant method was originally used by Kasteleyn to solve the dimer problem. The author uses the permanent-determinant method and a generalization called the Hafnian- Pfaffian method to write the answer as a determinant or a Pfaffian in each of the ten cases. The author is then able to row-reduce the relevant matrix and thus to prove the corresponding formula in four of the ten cases. Thus this presents a first step toward a more unified treatment. It should be mentioned that there are \(q\)-analogues of these conjectures. The totally symmetric plane partition \(q\)-analogue still remains open. Recently \textit{J. R. Stembridge} [J. Comb. Theory, Ser. A 68, No. 2, 372- 409 (1994; Zbl 0809.05007)] has observed an amazing ``\(q= -1\) phenomenon''. There is also the related problem of the alternating-sign matrix conjectures. The first of these conjectures has recently be proved by Zeilberger (preprint).