Topic: Suitable software for optimization problem

Hi guys,  I am part of an operations research team and we are working in an optimization project in the transportation industry. Our goal is to maximize the total revenue from moving loads minus the cost of moving trucks to pick up and deliver loads. We have a considerable amount of resources to allocate: around 60 trucks, and 80 drivers, five central spots, and we would like to elaborate a dynamic model that takes into account the evolution of a network through space and time. 
We are currently deciding whether to model the situation as a linear programming problem or like we have seen in some papers, as a dynamic programming problem. According to our research, the dynamic allocation problem grows exponentially with the number of resources, so the number of decision variables can be in the thousands order.
To model/solve the problem, we are testing Python+scipy (or another module), python+mpl, or matlab+ Optimization toolbox. In particular we are concerned with the arrays' management in those programs. Knowing that maybe we will be working with thousands (maybe millions?) of coefficients, which software package in your opinion (by your experience) is the best to solve a problem of these dimensions?.

Re: Suitable software for optimization problem

Your info is too general to give an advice. You haven't described your objective and constraints precisely (and even if you would, here hardly are any people here to dig into it; maybe someone from would).

First of all you should decide how you'll model the problem - as an ordinary MILP or somehow else. If former, you could use any modeling language (including Python, MATLAB) + any MILP solver.

scipy has neither MILP solver nor other tools for graph modeling and solving.

BTW, there are some people in our optimization dept with strong graph probs experience, although they haven't deal with neither Python nor MATLAB.

Re: Suitable software for optimization problem

Thank you very much Dmitrey!. I did not describe the objective and constraints precisely because there are lots of ways to solve a problem of these characteristics (each one with its advantages and disadvantages) and we have not decided how to focus the problem. However, for us it is clear that we need a software able to manage big amounts of  coefficients or data.  Your info was very useful, I described my problem in OR-exchange and got an answer. Nice info about scipy. Thanks again!