Parameters for the procedures using the derivatives

• `ALIAS/maxgradient`: the maximal width of a box for using the gradient for the evaluation of the expression (default value: 1.e10). (C++:ALIAS_Diam_Max_Gradient)
• `ALIAS/maxkraw`: the maximal width of a box before using the Krawczyk test (default value: 1.e10) (C++: ALIAS_Diam_Max_Kraw)
• `ALIAS/maxnewton`: maximal width of a box before using the interval Newton method (default value: 1.e10) (C++: ALIAS_Diam_Max_Newton)
• `ALIAS/newton_max_dim`: for the numerical C++ Newton scheme the iteration is stopped if the absolute value of the i-th parameter is greater than the i-th of this table
• `ALIAS/store_gradient`: the signs of the derivatives for each box processed by the algorithm are usually stored. Setting this variable to 0 allows to decrease the size of the storage memory
• `ALIAS/transmit_gradient`: the master program will usually transmit to the slaves the sign of the elements of the jacobian for the box that is send to the slave. This may be avoided by setting this flag to 0
• `ALIAS/use_inflation`,`ALIAS/eps_inflation`: as soon as a solution is found we will try to inflate the box in which the solution has been found by at least `ALIAS/eps_inflation`. This process may be computer intensive and may be invalidated by setting `ALIAS/use_inflation` to 0
• `ALIAS/type_n_new_boxes`, `ALIAS/allows_n_new_boxes`: if a unicity box has been discovered solutions may be sought in the following boxes only in the complementary part of the box with respect to the unicity box: this is allowed by setting the flag type_n_new_boxes to 1. This may create a large number of boxes and the flag allows_n_new_boxes allows to specify how many new boxes may be created
• `ALIAS/Grad_Equation`: an integer array used by the HessianSolve procedure. If the i-th element of this array is 0 the derivatives of the i-th equation are not used for the interval evaluation of the equation
• `ALIAS/apply_kanto`: an integer used by the HessianSolve procedure. If set to 2 the algorithm will always perform `ALIAS/newton_iteration` (default value: 100) iterations of a Newton scheme with as initial guess of the solution the center of the current box. If the scheme converge it is first verified that a solution has indeed been found and if this solution lie inside the initial search domain. The box in which a solution is found is then inflated and the search domain is reduced by the solution box. Using this method allows to often determine quickly solutions of a system.