Given a list of objective functions and a list of constraints functions, and a vector x0, findparam returns useful fsqp parameters. The list of objectives should be: (use empty list when necessary: list() )
list_obj=list(... list(f_1,...,f_nf0),... //regular objectives list(F_1,...,F_nfsr) //SR objectives )
The f_i's are functions: y=f_i(x) should return the value of the ith regular objective as a function of x. y can be a column vector if several regular objectives are stacked together. y=F_i(x) is the ith sequentially related objective. y is a column vector which contains the ith set of SR-objectives (mesh_pts(i) is set to size(F_i(x),1) by findparam).
The list of constraints functions should be as follows:
list_cntr=list(... list(g_1,...,g_ng0),... //regular nonlinear inequality list(G_1,...,G_ncsrn),... //SR nonlinear inequality list(c_1,...,c_nc0),... //regular linear inequality list(C_1,...,C_ncsrl),... //SR linear inequality list(h_1,...,h_nh0),... //nonlinear equality list(A_1,...,A_na0) //linear equality )
Functions g_i's, c_i's, h_1 can return column vectors y (e.g. y=g_1(x)) if several constraints are stacked together. Functions G_i's, C_i's, A_i's should return in a column vector the set of appropriate SR constraints. Examples are given at the end of examplei.sce files. See listutils.sci: generic functions obj, cntr, grob, grcn are constructed from the lists list_obj, list_cntr, and similar lists list_grobj and list_cntr which contain the gradients of objectives and constraints (matrices whith nparam=dim(x) columns).