Solving supply networks with discrete decisions

A method comprising:

• installing a dynamic library within a software application;
• calling the dynamic link library at runtime of the software application to execute a mixed-integer programming (MIP) model of a supply network with discrete constraints based on data stored in a storage device;
• executing a mixed integer programming (MIP) optimizer from within the installed dynamic link library on the model to generate a flow configuration within the supply network, by instructing the MIP optimizer to convert the discrete constraints within the supply network into linear constraints prior to generating the configuration, wherein the discrete constraints are converted to the linear constraints by removing specific discrete constraints corresponding to times necessary to find respective optimal solutions, locations within the supply network, and operations within the MIP model;
• segmenting the supply network into a plurality of sequences of the flow configuration, each of the plurality of sequences being segmented individually by the MIP optimizer and corresponding to a particular location of the locations within the supply network;
• identifying, at a beginning of the supply network, a first removed discrete constraint as having a highest priority by having most dependent operations in the flow configuration of the supply network as compared to other discrete constraints on the supply network that have less dependent operations than the highest priority identified first removed discrete constraint;
• incrementally removing the other discrete constraints on the supply network by relaxing binary variable values within the MIP model that correspond to the other discrete constraints, wherein the relaxed binary variable values become a linear requirement of greater than zero, resulting in shrinking the supply network and increasing flexibility for the MIP optimizer to find particular ones of the respective optimal solutions of the supply network in less time than the times necessary to find the respective optimal solutions;
• re-introducing a subset of the particular ones of the discrete constraints into the MIP model with relaxed binary variable requirements, the re-introduction starting, at each iteration, with the identified first removed discrete constraint having the highest priority at the beginning of the supply network being re-introduced before incremental re-introduction of the less dependent operations, the incremental re-introduction proceeding forward from the beginning of the supply network until end of the supply network;
• re-executing the MIP optimizer with the re-introduced subset of the particular ones of the discrete constraints to generate an updated flow configuration;
• outputting, by the MIP optimizer in a case the MIP optimizer cannot reach a complete solution by a stop condition defined by a timeout limit, results of how far the MIP optimizer has reached in finding the optimal solution;
• generating a graphical user interface that includes a plurality of graphical objects corresponding to a plurality of the locations included in the updated flow configuration and directional edges between the plurality of graphical objects corresponding to the updated flow configuration;
• and
• displaying the graphical user interface of the updated flow configuration within a page of the software application.

A computer-readable medium comprising instructions which when executed by a processor cause a computer to perform a method comprising:

• installing a dynamic library within a software application;
• calling the dynamic link library at runtime of the software application to execute a mixed-integer programming (MIP) model of a supply network with discrete constraints based on data stored in a storage device;
• executing a mixed integer programming (MIP) optimizer from within the installed dynamic link library on the model to generate a flow configuration within the supply network, by instructing the MIP optimizer to convert the discrete constraints within the supply network into linear constraints prior to generating the configuration, wherein the discrete constraints are converted to the linear constraints by removing specific discrete constraints corresponding to times necessary to find respective optimal solutions, locations within the supply network, and operations within the MIP model;
• segmenting the supply network into a plurality of sequences of the flow configuration, each of the plurality of sequences being segmented individually by the MIP optimizer and corresponding to a particular location of the locations within the supply network;
• identifying, at a beginning of the supply network, a first removed discrete constraint as having a highest priority by having most dependent operations in the flow configuration of the supply network as compared to other discrete constraints on the supply network that have less dependent operations than the highest priority identified first removed discrete constraint;
• incrementally remove the other discrete constraints on the supply network by relaxing binary variable values within the MIP model that correspond to the other discrete constraints, wherein the relaxed binary variable values become a linear requirement of greater than zero, resulting in shrinking the supply network and increasing flexibility for the MIP optimizer to find particular ones of the respective optimal solutions of the supply network in less time than the times necessary to find the respective optimal solutions;
• re-introducing a subset of the particular ones of the discrete constraints into the MIP model with relaxed binary variable requirements, the re-introduction starting, at each iteration, with the identified first removed discrete constraint having the highest priority at the beginning of the supply network being re-introduced before incremental re-introduction of the less dependent operations, the incremental re-introduction proceeding forward from the beginning of the supply network until end of the supply network;
• re-executing the MIP optimizer with the re-introduced subset of the particular ones of the discrete constraints to generate an updated flow configuration;
• outputting, by the MIP optimizer in a case the MIP optimizer cannot reach a complete solution by a stop condition defined by a timeout limit, results of how far the MIP optimizer has reached in finding the optimal solution;
• generating a graphical user interface that includes a plurality of graphical objects corresponding to a plurality of the locations included in the updated flow configuration and directional edges between the plurality of graphical objects corresponding to the updated flow configuration;
• and
• displaying the graphical user interface of the updated flow configuration within a page of the software application.