LINDO Systems is proud to introduce LINGO 12. The new release includes a powerful new feature to allow users to incorporate uncertainty into their optimization models. In addition, the new release has a number of solver performance enhancements.

All New Stochastic Programming (SP) Solver

The SP solver supports decision making under uncertainty through multistage stochastic models with recourse. The user expresses the uncertainty via distribution functions, either built-in or user-defined, and the stochastic solver will optimize the model to minimize the cost of the initial stage plus the expected value of recourse decisions over the planning horizon. Advanced sampling modes are also available to approximate stochastic parameters from parametric distributions.

Other features include:

• Available for modeling linear, nonlinear and integer stochastic programs (SP).

• Supports most standard distributions, e.g., Normal, Poisson, as well as user supplied.

• Full solutions for each of the possible scenarios are available at the scripting level, (calc sections) allowing for the creation of custom reports on variable values over the full range of scenarios.

• Ability to generate and display the underlying deterministic equivalent used to optimize SP models.

• Variance reduction with Latin-Hyper-Square sampling.

• Ability to generate statistically dependent samples based on Pearson, Spearman or Kendalls correlation measures.

• Pseudorandom number generator with long cycle length and excellent high dimensional uniformity.

Global Solver Improvements

• Significant improvements in exploiting quadratic expressions, making the global solver more efficient on non-convex quadratic models, as well as general nonlinear models with quadratic terms.

• Automatic recognition of second-order cone quadratic problems, such as Value-at-Risk models, allowing for dramatically faster solution times via the barrier solver.

• Reformulation capabilities that improve performance for a wide range of composite functions.
  

Integer Solver Improvements

Enhancements in the feasibility-pump heuristic to help find improved feasible solutions on many difficult problems.

• Enhancements in the rounding techniques exploit an even wider range of constraint structures.

• Standard heuristics have been improved.

Improved Performance on Models with Nested Loops

Loop optimization reformulates expressions containing nested set looping functions in order to make them more efficient, while maintaining mathematical equivalency. The end goal of loop optimization is to minimize the number of passes through the inner loop of any nested loops in an expression. Improvements in model generation times for some models can be dramatic.

Simplex Solver Improvements

Large linear models solve an average of 20% faster with the enhanced dual and primal simplex solvers.

More Flexibility in Solution Report Precision

LINGO's solution reports are no longer restricted to 7 significant digits when reporting numeric results. The user may now control the degree of precision, with anywhere from 1 to 17 significant digits.

New Scripting Function Capabilities

A number of calc section scripting functions were added or improved:
@GENDUAL generates the dual formulation of a linear program.
@FORMAT can now format output of strings, as well as numeric values.
@SMPS generates MPS format model files.

Variable Name Lengths Extended

Prior releases of LINGO had a limit of 32 characters on variable name lengths. This limit has been increased to 64 characters.