Open bchretien opened 9 years ago
QP solvers usually describe quadratic functions as:
f(x) = 0.5 x^T H x + x^T g + λ
but in RobOptim, GenericNumericQuadraticFunction considers:
GenericNumericQuadraticFunction
f(x) = x^T A x + x^T b + c
Thus, we have H = 2 A, which may be a source of error for users used to traditional QP solvers.
H = 2 A
QP solvers usually describe quadratic functions as:
but in RobOptim,
GenericNumericQuadraticFunction
considers:Thus, we have
H = 2 A
, which may be a source of error for users used to traditional QP solvers.