How much to charge for a software? The higher the price, the larger the profit from selling a copy. On the other hand, different consumers assign different values to this software, so, the higher the price, the fewer copies will be sold. This is true for all products, not only for software, and the problem is usually solved as follows: initially, the new product is sold at a higher price $p_1$; in course of time $t$, the price $p_t$ goes down: $p_1\ge p_2\ge ...\ge p_T$. The problem is to find the values $p_1,...,p_T$ that maximize the expected profit. This problem has been analyzed for general products. The situation with software products is more complicated because software is easily {\it copyable}: An average programmer often {\it copies} software from a friend; so, if a user has bought this software, the probability of his friends buying it decreases drastically. This additional feature makes a formula for the expected profit very complicated, and the resulting non-convex optimization problem becomes difficult to solve.

The authors successfully apply a constraints version of interval branch and bound optimization techniques to find global maxima of the profit function and thus, to compute the best prices $p_1,...,p_T$.