TomSym provides an easy-to-use modeling layer in TOMLAB. The creates the problem with analytical derivatives and uses standard TOMLAB assign routines to create a solver independent model. When using the TomSym class only the objective function and (nonlinear) constraints have to be provided by the user, the rest is handled automatically. Simple bounds, linear and nonlinear constraints are identified and structured correctly with symbolic pre-computed derivatives, such as the Jacobian matrix.

The generated Matlab code is fully optimized and will eliminate duplicated expressions, simplify and compact the code.

It is possible to model LP, MILP, QP, MIQP, SDP, BMI, NLP, MINLP problems among many others with the class.