2.1. Description

MindOpt is an efficient large-scale mathematical optimization problem solver software, which covers a variety of efficient optimization algorithms for solving multiple types of optimization models, and can help quickly solve the best solution for business problems. In this chapter, we briefly introduce MindOpt’s solving capabilities and interaction methods.

2.1.1. Overview of Solving Capabilities

MindOpt currently supports solving the following types of optimization problems.

Problem type

Algorithms

Linear Programming (LP)

Primal/Dual simplex method; Interior-point method

Mixed Integer Programming (MILP)

Branch and bound method

(Convex) Quadratic Programming (QP)

Interior-point method

Semidefinite Programming (SDP)

Interior-point method

The ability to solve more optimization problems is under development, please pay attention to our update notice.

2.1.2. Platform Usage and Acquisition of the Solver

  • Users can obtain a free local version (standalone version) from Alibaba Cloud’s cloud products and download it to their personal machines. For specific download, installation, and authorization application methods, please refer to the solver cloud products documentation.

  • Users can also try MindOpt on our cloud platform MindOpt Studio. We provide computing resources and a notebook environment on this cloud platform for users to use directly. At the same time, we offer several modeling and solving examples to help users quickly master the usage techniques of MindOpt.

  • For enterprise users, we can also provide customized versions of the standalone version and Compute/Server version. You can consult and obtain them through the methods in Contact Us.

2.1.3. Interaction and Invocation of the Solver

Users can call the solver or write their optimization programs through the command line or APIs in the following languages.

We also provide some modeling and optimization examples to explain how to establish, modify optimization models and solve, or use the initial basis to warm start optimization algorithms, etc., to help users master the skills of using the API for modeling and solving.

MindOpt also supports direct reading of optimization problems in the following standard formats, as well as their corresponding GZIP and BZIP2 compressed format files.

  • .mps format

  • .lp format

  • .dat-s format

  • .nl format

2.1.4. Modeling Tools

Considering users’ modeling needs, we also provide several modeling language docking methods and examples. Among them, the Mindopt APL modeling language was independently developed by the MindOpt team.

2.1.5. Advanced Modeling Techniques

We provide the following advanced modeling techniques and examples to help users better solve optimization models.