JOURNAL OF COMBINATORIAL THEORY

9, 324--326 (1970)

On Nonreconstructable Tournaments*
L. W. BEINEKE AND E. T. PARKER
Purdue University at Fort Wayne, Indiana 46805, and University of lllinois, Urbana, Illinois 61801 Communicated by Frank Harary

Received December 7, 1968

ABSTRACT Pairs of non-isomorphic strong tournaments of orders 5 and 6 are given for which the subtournaments of orders 4 and 5, respectively, are pairwise isomorphic. Heretofore, only pairs of orders 3 and 4 were known.

A tournament o f o r d e r n has n nodes with exactly one directed arc j o i n i n g each pair o f distinct nodes; in o t h e r words, it is a complete, irreflexive, a s y m m e t r i c relation on a set o f n elements. A t o u r n a m e n t T o f o r d e r n clearly determines n s u b t o u r n a m e n t s o f o r d e r n - - 1, each o b t a i n e d f r o m T b y deleting one n o d e a n d the incident arcs. A p r o b l e m in the study o f t o u r n a m e n t s , sometimes called U l a m ' s p r o b l e m in a n a l o g y with one in u n d i r e c t e d graphs, h a s been considered b y H a r a r y a n d P a l m e r [1] a n d can be stated as follows: F o r w h a t pairs o f n o n - i s o m o r p h i c t o u r n a m e n t s o f o r d e r n are the two sets o f n s u b t o u r n a m e n t s o f o r d e r n - - 1 pairwisei s o m o r p h i c ? Alternatively, given n t o u r n a m e n t s o f o r d e r n - - 1, when can n o n - i s o m o r p h i c t o u r n a m e n t s o f o r d e r n be r e c o n s t r u c t e d f r o m t h e m ? A t o u r n a m e n t is called strong if there is no p a r t i t i o n o f its nodes into n o n - e m p t y subsets U a n d V such t h a t all arcs go f r o m U to V. A w e l l - k n o w n t h e o r e m states that a t o u r n a m e n t is s t r o n g if a n d only if it has a s p a n n i n g directed cycle. H a r a r y a n d P a l m e r s h o w e d that, if two n o n - i s o m o r p h i c t o u r n a m e n t s o f o r d e r n > 4 are n o t strong, then their s u b t o u r n a m e n t s o f o r d e r n - - 1 c a n n o t be pairwise-isomorphic. It s h o u l d be n o t e d (see M o o n [2, p. 3]) that " a l m o s t all" t o u r n a m e n t s are strong, however.

* Research of first author supported by Air Force Office of Scientific Research grant AFOSR-68-1515. Research of second author supported by Office of Naval Research grant N000 14-67-A-0305-0008. 324

ON NONRECONSTRUCTABLE TOURNAMENTS

325

The phenomenon of pairs of tournaments of order n having their subtournaments of order n -- 1 pairwise-isomorphic is apparently quite rare. The example of order 3, shown in Figure 1, must be regarded as a degenerate case, since there is only one tournament of order 2. However, the example of order 4, given in [1] and shown in Figure 2, cannot be considered degenerate.

FIG. 1.

Order 3.

FIG. 2. Order 4. Harary and Palmer conjectured the non-existence of examples of higher order, but we give examples of orders 5 and 6 here. Schematically, these are shown in Figures 3 and 4, in which the arcs not shown are understood to be directed from left to right. (This type of diagram was used by Moon [2] in his book on tournaments.)

FIG. 3. Order 5.

FIG. 4. Order 6.

326

BE1NEKE AND PARKER

T o see that the t o u r n a m e n t s of order 5 are n o n - i s o m o r p h i c , we note that, a l t h o u g h the list of scores ( n u m b e r s of out-going arcs from the nodes) is (3, 2, 2, 2, 1) in both, in one the arc f r o m the node of score 1 goes to the one o f score 3, while in the other the c o r r e s p o n d i n g arc goes to a n o d e o f score 2. Since each of the two t o u r n a m e n t s of order 6 has a u n i q u e transitive s u b t o u r n a m e n t of order 5, they are clearly non-isomorphic. There are n o k n o w n larger examples; perhaps there are none. O r ' i s there an example for every order ? Such conjectures a n d questions a b o u n d . By considering the converse t o u r n a m e n t s (obtained by reversing the direction of every arc), we observe that the pairs of t o u r n a m e n t s of orders 4 a n d 6 are converses of each other, while the t o u r n a m e n t s of orders 3 a n d 5 are self-converse. Is this a pattern which might prove helpful ? I n conclusion we m e n t i o n the analogous p r o b l e m for directed graphs in general: D o examples exist which are n o t t o u r n a m e n t s ?

Added in proof Exhaustive examination of all 56 tournaments with six nodes has yielded two additional pairs whose subtournaments of order 5 are pairwise-isomorphic. These are shown in Figs. 5 and 6. In Fig. 5, the arc from the node with score 1 goes to a node with score 2 in the first case, and the corresponding arc goes to a node with score 3 in the second. In Fig. 6, the three nodes having score 2 form a transitive triple in one case, and a cyclic triple in the other. Thus, the two tournaments in each figure are non-isomorphic. That the proper subtournaments are pairwise-isomorphic in each case is a matter of routine verification. We note that the tournaments in each pair are converses of each other.

FIG. 5. Order 6.

FIG. 6. Order 6.

REFERENCES 1. F. HARARYAND E. PALMER,On the problem of reconstructing a tournament from subtournaments, Monatsh. Math. 71 (1967), 14-23. 2. J. W. MOON, Topics on Tournaments, Holt, Rinehart, & Winston, New York, 1968.