We've just released Analytic Solver V2025 Q1, with new versions of Analytic Solver Desktop and Cloud (for Excel users), Solver SDK (for developers), and our cloud platform RASON. This release features both usability improvements and a powerful new way to solve a wide range of nonlinear optimization models using the Gurobi Solver Engine -- traditionally known for solving linear (and quadratic) mixed-integer models. This new approach yields globally optimal solutions every time -- not possible with other nonlinear Solvers -- and often much faster solutions.
Usability: New Flexible Dialogs
Analytic Solver has always offered an easy user interface with an Office-style Ribbon, Task Pane, and dialogs that you can move around and resize. But the content of many dialogs used fixed-size fonts and graphics. To help with both accessibility, and better use of larger, high-resolution screens when you're "consuming" this content, in Analytic Solver V2025 Q1 we've revised the Task Pane and scores of other dialogs with "zoomable" content. You can now press the CTRL key, then use the mouse wheel to enlarge or shrink the text and graphics in almost any dialog or pane of Analytic Solver. At the bottom of this page is an example, where we've enlarged the text in the top and bottom parts of the Task Pane, but left the middle part (Model Diagnosis) unchanged -- so you can see the difference.
Performance: New Nonlinear Solver Power
Analytic Solver V2025 Q1 features a very special level of support for the just-released (November 19) Gurobi Solver 12.0. This popular Solver again improves performance on linear programming (about +4%) and mixed-integer programming (about 13%) models, and even better performance (about +28%) on non-convex mixed-integer quadratic models. But most interesting for our Excel users is Gurobi's improved support for (often non-convex) mixed-integer nonlinear optimization problems. Such problems arise in specialty chemicals, hydroelectric power, and some aeronautics and space applications.
Unlike other nonlinear Solvers (such as our LSGRG, LSSQP and KNITRO Solvers), Gurobi takes a different approach, that can yield globally optimally solutions IF you can formulate your model with (only) certain algebraic expressions -- defined via special Python API calls -- that use just over a dozen mathematical operators (plus, minus, times, divide, square, square root, exponential and logarithm, etc.). If you're really good at Python coding and really good with algebraic manipulation, you can code this yourself. But in Analytic Solver V2025 Q1, you don't have to do any extra work to define your nonlinear model for the Gurobi Solver.
Standard Excel Formulas, New Solving Speed
You can write your model in Excel, using any of Excel's mathematical operators and any of Excel's 470+ built-in functions. Analytic Solver's PSI Interpreter will analyze your formulas (including formulas that depend on other formulas, to any level), diagnose your model as linear, quadratic, smooth nonlinear or non-smooth, and -- In V2025 Q1 -- tell you if your model is "nonlinear and solvable via the new Gurobi Solver". Analytic Solver can generate the internal "expression trees" that the Gurobi Solver requires.
Analytic Solver will even expand various Excel functions, based on their mathematical definitions, into chains of "Gurobi-supported primitive operators". For example, you can just go ahead and use the STDEV Excel function -- Analytic Solver will recognize this as "Gurobi-eligible" and create (internally) an expression tree using square, plus and square root operators, based on the definition of "standard deviation". The bottom line: With Analytic Solver, you can use the Gurobi 12.0 Solver to solve a much wider range of nonlinear models, much more quickly and easily, than you could any other way.
For "eligible" nonlinear optimization models, the Gurobi 12.0 Solver is often faster than other nonlinear Solvers, especially when integer variables are also used in the model, and when you require a globally optimal solution. This isn't always true however -- occasionally Gurobi is (much) slower -- so it's good that you have alternative nonlinear Solvers available to you, even in the basic Analytic Solver Optimization product. We do hope and expect to see even better performance in future versions of the Gurobi Solver.
Analytic Solver V2025 Q1 is a free upgrade for anyone with a paid Analytic Solver Optimization (or higher) license. If you've been a customer for years, you're probably accustomed to such dramatic enhancements -- but we're always working hard to bring you more value, sooner than anyone else in our "world" of advanced analytics.
