An upper bound or a lower bound of the Multiple-Choice Knapsack Problem can
be calculated by solving LP relaxation. In 1979, Sinha and Zoltners propos
ed a branchand-bound algorithm for solving the Multiple-Choice Knapsack Pro
blem, and provided a method to obtain the strict upper bound. In this paper
, we propose a new calculation method to obtain a stricter upper bound than
the upper bound of Sinha-Zoltners and compare the results of the two metho
ds. (C) 2001 Scripta Technica.