In using principles of combinatorial game theory it has been discussed in the literature how in the game of Go one can assign values to eyes in order to decide whether a position lives unconditionally, simply by adding these values and checking whether or not their sum reaches the value of two (see Landman [1]). These concepts are applied in computer Go programs (as in [3]) and a computer generated library of eye shapes is available from Dave Dyer [2].

In this contribution we describe the generation of a library of eyes surrounded by only one chain which we call monolithic eyes. Compared to Dyer’s library our database has a number of extensions: an evaluation of the number of ko threats needed to live or to kill the eye, the consideration of a larger number of external liberties and of an extra attached eye, larger eye sizes, the determination of all winning moves and others.

In the following section we describe a procedure to bring any set of empty or occupied points into a unique position by using shifts and symmetries of the board. The purpose is to avoid the generation of equivalent eyes as described in Section 3. The evaluation of eyes is done with the program GoTools [4] as outlined in Section 4. Computational aspects including a listing of optimizations, comments about performed consistency tests, usefulness and availability follow in Section 5.

The database of monolithic eyes is a rich source of strange positions. In Section 6 we give two examples of how to inspect it: by looking for boolean-calm eyes, i.e., eyes which are not settled but where passing belongs to the boolean-best moves1 of both sides; and by checking for eyes where the boolean-best attacking move depends in a nonmonotonic way on the number of external liberties.