Optimization Solver:History versions of Solver SDK

Download Optimization Solver SDK V1.1.1

Release date: March 06, 2024

Release notes:

V1.1.1 Release date: March 06, 2024

• Fixed a bug

V1.1.0 Release date: January 30, 2024

• MILP and QP algorithm performance improvements.

• Added 5 new tunable parameters for MILP.

• Added new parameter SolutionNumber to retrieve multiple suboptimal solutions of MIP solving process.

• Introduced C# API along with documentation and examples.

• Support for invoking MindOpt in Julia using JuMP and AmplNLWriter.jl packages.

• Refined the license verification scheme.

• Bug fixes

  • Fix a precision bug in writing Lpf.

  • Fix a bug in reporting solution time for concurrent optimization.

  • Fix an iteration limit bug for concurrent optimization.

  • Fix a bug in hot-start for simplex method.

  • Fix a bug in counting dense columns for QP problems.

  • Fix a bug in preprocessing for QP problem.

Download Optimization Solver SDK V1.0.1

Release date: November 14, 2023

Release notes:

• Algorithm performance improvements and bug fixes

Download Optimization Solver SDK V1.0.0

Release date: October 12, 2023

Release notes:

• Introduce a new MindOpt SDK API and corresponding code examples

• Add features like MILP’s warm start, SOS constraints and Indicator constraints

• Add IIS solving functionality, and upgrade LP’s IIS solving module

• Add Callback functionality

• Add support for qps format for input files

• Improve pipeline of concurrent optimization methods

• Providing primal-dual feasible solution for LP

• Algorithm performance improvements and bug fixes

Download Optimization Solver SDK V0.25.1

Release date: August 16, 2023.

V0.25.1 Release date: August 16, 2023

Release notes:

• Fix a bug

V0.25.0 Release date: August 16, 2023

Release notes:

• Improved the performance and the numerical stability of the interior point solver.

• Improved the performance of Python API.

• Improved the stability of the IIS algorithm.

• Intorudced new IIS algorithm for MILP.

• API changes

  • New error codes:

    • MDO_INVALID_ARGUMENT(-3),MDO_IIS_NUMERIC(-30000),MDO_IIS_NO_SOLN(-3004),MDO_IIS_FEASIBLE(-3005).

  • New attributes:

    • C API:MDO_INT_ATTR_ROW_IIS,MDO_INT_ATTR_COL_IIS.

    • C++ API:MDO_INT_ATTR::ROW_IIS,MDO_INT_ATTR::COL_IIS.

    • Python API:MDO_INT_PARAM.ROW_IIS,MDO_INT_ATTR.COL_IIS.

    • JAVA API:INT_ATTR::ROW_IIS,INT_ATTR::COL_IIS.

  • Updated APIs:

    • C API:Mdo_computeIIS.

    • C++ API:MdoModel::computeIIS.

    • Python API:MdoModel.compute_iis.

    • JAVA API:MdoModel::computeIIS.

  • New integer parameters:

    • C API:MDO_INT_PARAM_MIP_MAX_STALLING_NODES,MDO_INT_PARAM_MIP_MAX_SOLS,MDO_INT_PARAM_MIP_DETECT_DISCONNECTED_COMPONENTS,MDO_INT_PARAM_MIP_SOLUTION_POOL_SIZE,MDO_INT_PARAM_MIP_ALLOW_DUAL_PRESOLVE.

    • C++ API:MDO_INT_PARAM::MIP_MAX_STALLING_NODES,MDO_INT_PARAM_MIP_MAX_SOLS,MDO_INT_PARAM::MIP_DETECT_DISCONNECTED_COMPONENTS,MDO_INT_PARAM::MIP_SOLUTION_POOL_SIZE,MDO_INT_PARAM::MIP_ALLOW_DUAL_PRESOLVE.

    • Python API:MDO_INT_PARAM.MIP_MAX_STALLING_NODES,MDO_INT_PARAM.MIP_MAX_SOLS,MDO_INT_PARAM.MIP_DETECT_DISCONNECTED_COMPONENTS,MDO_INT_PARAM.MIP_SOLUTION_POOL_SIZE,MDO_INT_PARAM.MIP_ALLOW_DUAL_PRESOLVE.

    • JAVA API:INT_PARAM::MIP_MAX_STALLING_NODES,INT_PARAM_MIP_MAX_SOLS,INT_PARAM::MIP_DETECT_DISCONNECTED_COMPONENTS,INT_PARAM::MIP_SOLUTION_POOL_SIZE,INT_PARAM::MIP_ALLOW_DUAL_PRESOLVE.

  • New real parameters:

    • C API:MDO_REAL_PARAM_MIP_LINEARIZATION_BIG_M,MDO_REAL_PARAM_MIP_CUTOFF.

    • C++ API:MDO_REAL_PARAM::MIP_LINEARIZATION_BIG_M,MDO_REAL_PARAM::MIP_CUTOFF.

    • Python API:MDO_REAL_PARAM.MIP_LINEARIZATION_BIG_M,MDO_REAL_PARAM.MIP_CUTOFF.

    • JAVA API:REAL_PARAM::MIP_LINEARIZATION_BIG_M,REAL_PARAM::MIP_CUTOFF.

• Bug fixes

  • Fixed an API bug related to SDP.

  • Fixed a performance issue related to row deletion.

  • Fixed a few bugs in MPS reader and LPF reader.

Download Optimization Solver SDK V0.24.1

Release date: April 30, 2023

Release notes:

• New installation package for Linux ARM and Apple Silicon.

• Add pip install MindOpt Python package for Python only users. pip install mindoptpy.

• Improved the performance of MILP solver.

• Change name of comand-line function: --sanitize and --undo_sanitize. Refer to Sanitize Model Files for usage.

Download Optimization Solver SDK V0.24.0

Release date: March 30, 2023

Release notes:

• Improved the performance of MILP solver. A formal release version. Open to use.

• Improved the performance of the concurrent optimization method.

• Redesigned the Cholesky factorization code in IPM.

• API changes

  • New comand-line function, see Sanitize Model Files for details:

    • --desensitize to desensitize optimize problem model files, by removing the names of variables and constraints so that only leave the numerical values of the model.

    • --sensitize to recover the model which has been desensitized.

  • New integer parameters:

    • C API: MDO_INT_PARAM_MIP_AUTO_CONFIGURATION.

    • C++ API: MDO_INT_PARAM::MIP_AUTO_CONFIGURATION.

    • Python API: MDO_INT_PARAM.MIP_AUTO_CONFIGURATION.

    • JAVA API: MDO_INT_PARAM::MIP_AUTO_CONFIGURATION.

  • New error codes:

    • MDO_MODEL_INVALID_CHAR_ATTR(-2014), MDO_MODEL_INVALID_ATTR_NAME(-2019).

• Bug fixes

  • Fixed a bug in API Mdo_loadProb.

  • Fixed a minor bug in MPS writer when GZ/BZ2 compression format is required.

  • Fixed a bug in LPF writer and MPS writer.

  • Fixed a bug related to Mdo_copyMdl.

  • Fixed a minor bug in the AMPL application (mindoptampl).

Download Optimization Solver SDK V0.23.1

Release date: December 21, 2022

Release notes:

• An installation package for the macOS operating system that runs on an Apple M1 chip is added based on Optimization Solver SDK V0.23.0. This installation package supports LP, convex QP, and SDP solving capabilities, but it does not support the MILP solving capability.

Download Optimization Solver SDK V0.23.0

Release date: November 29, 2022

Modification records:

• APIs related to SDP are added, and input files in dat-s format are supported.

• The descriptions of .nl files are added. The ampl subfolder of the examples folder contains .nl file examples. You can use the mindoptampl filename.nl command to call the .nl files.

• A minor bug in the LPF reader is fixed.

• Bugs that affect MILP solving are fixed, and MILP performance is optimized. Note: Only few test licenses for solving MILP problems are available. If you need a test license, send us an email to apply for it.

• C, C++, Python, and Java APIs are added. Example: MdoModel.add_sym_mat, MdoModel::add_sym_mats, MdoModel::replace_sym_mat_objs, MdoModel::replace_sym_mat_elements, MdoModel::get_real_attr_sym_mat, and MdoModel::copy_model Python APIs.

• The following error codes are added:

  • MDO_MODEL_INVALID_SYM_MAT_IDX (-2006), MDO_MODEL_INVALID_SYM_MAT_ROW_IDX (-2007), MDO_MODEL_INVALID_SYM_MAT_COL_IDX (-2008), and MDO_MODEL_INVALID_REAL_ATTR_SYM_MAT (-2013).

• The following error code is modified.

  • MDO_LIB_FOLDER_NOT_FOUND (-12) is replaced with MDO_DLL_ERROR (-12).

• C, C++, Python, and Java attributes are added. Example: MDO_REAL_ATTR.SYM_MAT_PRIMAL_SOLN, MDO_REAL_ATTR.MIP_GAP_ABS, and MDO_REAL_ATTR.MIP_GAP_REL Python attributes.

Download Optimization Solver SDK V0.20.2

Release date: October 24, 2022

Release notes:

• Bugs such as file read failures and hardware environment issues are fixed.

Download Optimization Solver SDK V0.20.1

Release date: September 23, 2022

Release notes:

• MILP performance is optimized, and the value range of the MIP/rootParallelism parameter is changed.

• Python SDK installation scripts are optimized. If you call Python APIs in Window operating systems, an error is reported. The script optimization can resolve this issue.

• Software output is optimized.

Download Optimization Solver SDK V0.20.0

Release date: August 31, 2022

Release notes:

• A branch-and-cut solver is added to solve MILP problems. An API for A Mathematical Programming Language (AMPL) software is added to solve MILP problems.

• License verification is optimized. Authentication logic modification:

  • The following authentication-free logic is supported: You do not need to use the mindopt.lic or fl_client.ini license file to solve an optimization problem that involves less than 300 variables and less than 300 constraints.

  • You need to use one of the preceding license files to solve an optimization problem that involves more than 300 variables and more than 300 constraints. For more information how to obtain a license file, see License settings.

  • Note: Only few test licenses for solving MILP problems are available. If you need a test license, send us an email to apply for it. Then, we manually grant the required permissions. You can obtain the email address from the Contact us topic.

• Installation packages are optimized to simplify installation steps.

  • SDK installation scripts for Linux and macOS operating systems are optimized. If a script is run, no additional parameters need to be added, and the installation path is interactively confirmed. A custom path is supported. After the SDK is installed, environment variables are automatically configured. A prompt message is displayed if a Windows operating system detects that one or more SDK versions are installed.

• Multiple APIs related to the MILP feature are added. For more information, see MindOpt User Manual.

Download Optimization Solver SDK V0.19.0

Release date: May 11, 2022

Release notes:

• The input and solving of convex QP problems in nonlinear programming (NLP) are supported. For example, MdoModel.set_quadratic_elements, MdoModel.get_quadratic_elements, MdoModel.delete_quadratic_elements, MdoModel.delete_all_elements, and MdoModel.delete_all_quadratic_elements APIs can be used to create and modify convex QP models.

• Performance of the simplex algorithm and the interior point method (IPM) is improved, and features and stability of AMPL and the concurrent optimization method are optimized.

• License management is designed, and APIs such as the MdoEnv API related to the createEnv function are added. In this case, Optimization Solver can be called in a large number of high concurrency and low latency scenarios.

• Installation methods of installation packages and documentation instructions are optimized.

Download Optimization Solver SDK V0.15.1

Release date: November 8, 2021

Release notes:

• The network simplex algorithm and the multi-threading simplex algorithm are added.

  • The network simplex algorithm performs well in Large Network-LP Benchmark. We recommend that you use this method to resolve network structure issues. The added EnableNetworkFlow parameter is set to 1.

• Stability of the simplex algorithm and the IPM is improved.

• The data read module is updated.

• APIs are modified.

  • The following parameters are added:EnableNetworkFlow, SPX/PrimalPricing, and SPX/DualPricing.

  • The following parameters are updated: Presolve and Method.

  • The following error codes are added: MDO_HOME_ENV_NOT_FOUND and MDO_LIB_FOLDER_NOT_FOUND.

  • The following APIs are added:

    • C APIs: Mdo_setStrAttr and Mdo_getStrAttr

    • C++ APIs: MdoModel::setStrAttr and MdoModel::getStrAttr

    • Python APIs: MdoModel.set_str_attr and MdoModel.get_str_attr

  • The ProbName model attribute is added.

Download Optimization Solver SDK V0.14.0

Release date: September 26, 2021

Release notes:

• Java APIs and floating licenses are supported.

• Performance of the simplex algorithm and the IPM is improved.

• C, CPP, and Python APIs for computing an irreducible infeasible system (IIS) are added.

• The quicksum Python API is added, and Python API design is optimized.

• AMPL or Pyomo APIs and applications are added, and related examples are provided.