Instances for the dynamic joint replenishment problem (DJRP)

This repository contains the randomly generated benchmark instances used in the computational experiments of the article “An uncapacitated dynamic joint replenishment problem with concave costs”, submitted to _____. The instances are provided in plain-text `.djrp` format and are intended for reproducible testing and for the comparison of exact, heuristic, and matheuristic solution approaches.

Each instance file specifies:
(i) the number of items $K$ and planning periods $T$,
(ii) the type of holding-cost and replenishment-cost functions (LINEAR, PIECEWISE, or POWER),
(iii) the joint setup cost for each period, $S_t$,
(iv) the item-by-period demand matrix, $d_t^k$,
(v) the item-specific setup costs, and
(vi) the coefficients defining replenishment and holding costs.

For example, the repository includes two instances, `F1K10T26A5B6S1.djrp` and `F6K10T26A5B6S1.djrp`, which illustrate two cost structures over the same 10-item, 26-period demand scenario. The first instance is a linear-cost instance, with linear holding and replenishment costs. The second instance is a piecewise-cost instance with discount-based concave costs, using replenishment and holding breakpoints at 0, 50, and 100 units, and discount factors of 1.0, 0.9, and 0.8. In both cases, the files include a common joint setup cost per period and item-specific setup-cost parameters.

These instances were randomly generated following the benchmark style commonly used in the dynamic joint replenishment literature, as in Boctor et al. (2004), and are intended to support transparency, replicability, and future comparative studies.