Release 17 of What'sBest! includes a wide range of performance enhancements and new features:
- Support for Multiple Objective Optimization
- Support has been added to handle multiple objectives for linear and integer models using a strategy known as Lexicographic/Preemptive optimization. The feature allows the user to specify an ordered list of objectives. What’sBest! will sequentially solve the model by optimizing the 1st objective, then constraining the 1st objective to its optimal value and optimize the 2nd objective, then constraining the 1st and 2nd objectives and optimize the 3rd objective, and so on down the objective list.
- Improved Support for using Network and Cloud Drives
- A feature has been added to allow users to more easily solve models stored on network and cloud drives.
- Improved Support for Excel® Functions
- Support for the SUMIFS() formula has been added.
- Support for the SUMIF() formula has been extended.
- Improved User Interface
- The format and layout of many dialog boxes have been simplified with smarter choice of defaults but with no loss of functionality.
- Simplified and Improved Installation Process
- The new smart installation process checks for 32-bit vs 64-bit Excel® and installs the appropriate version automatically.
- Support for User VBA Progress Callback
- This feature has been added to allow users to define a callback procedure (via the VBA interface) for accessing the progress information during the solution process.
- Global Solver Improvements
- Improved performance for models with convex-concave functions.
- Models with ratio/fractional terms, including MIP, tend to solve faster, sometimes order of magnitude.
- Non-convex quadratically constrained models are solved substantial faster.
- Improved recognition and exploitation of convexity of various composite functions,
e.g.,
ln(a + b * exp(x)) WBLOGABEXPX
ln(a1 * exp(x1) + . . . +an * exp(xn)) WBLOGSUMAEXP
ln(exp(x1) + . . . + exp(xn)) WBLOGSUMEXP
x^a/(b - x) WBXPOWDIVAB
- Significant Speed Improvements on Quadratic/Conic Models
- Support has been added for exponential and power cones.
- The models with ex and xk terms are solved faster.
- Integer Solver Improvements
- Better performance on many integer models has been achieved through improved selection of default settings for cut generation and solution heuristics.
Improved Performance on Linear Models
- Enhancements to the Simplex solvers allow linear models with many more variables than constraints to solve in a fraction of the time previous releases required.
- Performance has been improved for linear models when using multiple cores with concurrent execution of Primal, Dual, and Barrier solvers.
