knitro_ms: Call the KNITRO C++ interface using multiple start values
In jtilly/knitroR: R Interface for Non-Linear Constraint Optimizer Knitro

This is a multi start version of knitro(). Uses a matrix as startvalues where each row corresponds to one set of startvalues to be used. This version of multi-start gives the user more control over the start values than KNITRO's built-in version of multi-start. If you want to use the built-in version of multi-start instead, you can do so via the options file.

1
2
3

knitro_ms(objFun, objGrad = NULL, c_equality = NULL, c_inequality = NULL,
  jac = NULL, jacIndexCons = NULL, jacIndexVars = NULL, x0 = NA,
  lb = NULL, ub = NULL, optionsFile = "options.opt")

`objFun`	is a scalar valued R function that returns the objective function
`objGrad`	is a vector-valued R function with the gradient
`c_equality`	is a vector-valued R function with equality constraints
`c_inequality`	is a vector-valued R function with inequality constraints
`jac`	is a vector with the content of the Jacobian (sparse)
`jacIndexCons`	refers to each element of jac and contains the number of the constraint it refers to. Indexing is C++ compatible, i.e. the first constraint has index 0
`jacIndexVars`	refers to each element of jac and contains the number of the variable it refers to. Indexing is C++ compatible, i.e. the first variable has index 0
`x0`	is a matrix with starting values
`lb`	is a vector of lower bounds
`ub`	is a vector of upper bounds
`optionsFile`	is the path and filename of the options file. If it does not exist, the function will create it