The recently released
What’sBest! 7.0 solves broad classes of problems faster and
includes new tools to tackle tough nonlinear models.
New Global Solver
What’sBest! 7.0 can
now find the mathematically proven global optimum on non-convex nonlinear
models. Rather than stopping after the first local optimum is found,
the global solver will search until the global optimum is confirmed.
The nonlinear and global license options are required in order to
utilize the global optimization capabilities with What’sBest!.
Multistart Capability
The new Multistart feature
can be a powerful tool for finding good solutions to nonlinear models
more quickly. This feature intelligently generates a set of candidate
starting points in the solution space of nonlinear models and mixed
integer nonlinear models. Then, the solver selects a subset of these
candidate solutions to initialize a series of local optimization.
For non-convex nonlinear models, the quality of the solution returned
by the multistart solver will be superior to that of the general nonlinear
solver. A user adjustable parameter controls the maximum number of
multistarts to be performed. The nonlinear and global license options
are required in order to utilize the multistart feature with What’sBest!.
Quadratic Recognition and Solver
Quadratic Programming (QP) models
are a common class of nonlinear model that is encountered in applications
such as financial portfolio analysis. The new QP recognition tool
in this release of What’sBest! automatically determines if
a nonlinear model is actually a quadratic model. If the model is
linear with a quadratic objective, then it will be passed to the
faster quadratic solver, which is available as part of the Barrier
Solver option.
Improved Integer Solver
The new integer solver benefits from
a number of enhancements that boost performance on many classes of
problems. A partial list of new features include:
More advanced probing/pre-solving—including lifting clique
Special pre-solving of rows containing all binary variables
Additional cut generation
Faster cut generation
Improved rounding heuristic
New enumeration solver for pure binary models
Many new user controllable parameters
Improved reduced cost fixing and bound tightening within the
tree
Improved performance on mixed integer quadratic models
Improved Linearization Capabilities
This release improves handling
and performance on models with non-smooth functions such as IF, MAX,
MIN, and ABS as well as the product of a binary integer and continuous
variable. Linearization can automatically convert these to a series
of linear, mathematically equivalent expressions. Many non-smooth
models may be entirely linearized. This allows the linear solver to
quickly find a global solution to what would have otherwise been an
intractable problem.
