Classic Newton

Molecular dynamics (MD) studies the time evolution of N interacting particles via the solution of classical Newton s equations of motion. 

The strategy in a molecular dynamics simulation is conceptually fairly simple. The first step is to consider a set of molecules. Then it is necessary to choose initial positions of all atoms, such that they do not physically overlap, and that all bonds between the atoms have a reasonable length. Subsequently, it is necessary to specify the initial velocities of all the atoms. The velocities must preferably be consistent with the temperature in the system. Finally, and most importantly, it is necessary to define the force-field parameters. In effect the force field defines the potential energy of each atom. This value is a complicated sum of many contributions that can be computed when the distances of a given atom to all other atoms in the system are known. In the simulation, the spatial evolution as well as the velocity evolution of all molecules is found by solving the classical Newton equations of mechanics. The basic outcome of the simulation comprises the coordinates and velocities of all atoms as a function of the time. Thus, structural information, such as lipid conformations or membrane thickness, is readily available. Thermodynamic information is more expensive to obtain, but in principle this can be extracted from a long simulation trajectory. 

Existence of multiple solutions. Existence of solutions to (4.3.10a), (4.3.10c), constructed numerically and described above, immediately follows from the classical Newton-Kantorovitch theorem (see, e.g., [26]) which asserts the following. 

Using the classic Newton s law Fj = nijxaj (F , mj and a are the force on the atom, its mass, and its acceleration, respectively) and taking into consideration a Maxwellian distribution for a given temperature, a simulation of the atoms velocities is performed for a few picoseconds. 

A first model is used to compute the flowrates allowing to perform the separation with the greatest productivity. Then, the "mixed cell in series" model takes into account thermodynamic, hydrodynamic and kinetic properties of the system and compute the concentration profile inside the columns [14], In this model, we make the assumptions that the pressure drop inside the column is negligible compared to the pressure drop realized and controlled with the analogical valves, and we model the true moving bed assuming that the performance of SMB and TMB are equivalent. A mass balance equation is written for each stage and a classical Newton Raphson numerical method is used to solve the permanent state of the process [14],... 

In the classic Newton method, the Newton direction is used to update each previous iterate by the formula xfe+1 = x + pfe, until convergence. The reader may recognize the one-dimensional version of Newton s method for solving a nonlinear equation f(x) = 0 x +1 = xk — f(xk)/f (xk). The analogous iteration process for minimizing f x) is x +1 — xk — f xk)lf"(xk). Note that the one-dimensional search vector, -f xit)lf"(.xk), is replaced by the Newton direction -Hk lgt in the multivariate case. This direction is defined for nonsingular Hk. When x0 is sufficiently close to a solution x, quadratic convergence can be proven for Newton s method.3-6 That is, a constant 3 exists such that... 

Unfortunately, there is a disparity between this theoretical convergence result and the practical behavior of the method in general. Thus, modifications of the classic Newton iteration are essential for guaranteeing global convergence, with quadratic convergence rate near the solution. 

Table 1 compares computational effort, storage requirements, and performance characteristics among five methods classic Newton (for reference), nonlinear CG, full-memory QN, limited-memory QN, and TN methods. 

In the classical Newton-Raphson technique, the Jacobian matrix is inverted every iteration in order to compute the corrections AT] and Al]. The method of Tomich, however, uses the Broyden procedure (Broyden, 1965) in subsequent iterations for updating the inverted Jacobian matrix. 

Second, we should consider the effect of the separation distance between molecules on the interaction forces. It was found that these fall off exponentially as the distance between atoms increases. The explanation for this behavior can be found in classical Newton mechanics according to Newton s laws, a force, F, is a push or pull exerted on a body it is a vector quantity with magnitude (newton, N = kgms 2, in the SI system) and direction. The work, W, done by the force acting on a body is given as... 

Even though Eq. (1.9) obtains from classical (Newton) dynamics, it is not hard to prove that the relation 6.E = d(wc ) = 0 d is valid also in the theory of special relativity as well, see e.g. Lowdin [4], From Eqs. (1.6-1.8), we obtain the general result (using 3c = Or)... 

Newton s Method The classical Newton s method is a technique that instead of specifying a step length at each iteration uses the inverse of the Hessian matrix, H(x)" to deflect the direction of steepest descent. The method assumes that /(x) may be approximated locally by a second order Taylor approximation and is derived quite easily by determining the minimum point of this quadratic approximation. Assuming that H(x ) is nonsingular, then the algorithmic process is defined by... 

The Bom-Oppenheimer separation of the electronic and nuclear motions is a cornerstone in computational chemistry. Once the electronic Schrodinger equation has been solved for a large number of nuclear geometries (and possibly also for several electronic states), the potential energy surface (PES) is known. The motion of the nuclei on the PES can then be solved either classically (Newton) or by quantum (Schrodinger) methods. If there are N nuclei, the dimensionality of the PES is 3N, i.e. there are 3N nuclear coordinates that define the geometry. Of these coordinates, three describe the overall translation of the molecule, and three describe the overall rotation of the molecule with respect to three axes. For a linear molecule, only two coordinates are necessary for describing the rotation. This leaves 3N - 6(5) coordinates to describe the internal movement of the nuclei, which for small displacements may be chosen as vibrational normal coordinates . 

In a molecular dynamics calculation, equations of motion are integrated to determine the trajectories of all atoms in the molecule. The equations of motion can. in principle, be either classical (Newton s laws of motion) or quantum mechanical. But, in practice, due to the very laige number of atoms in a macromolecule. Newton s equations of motion are used. Quantum mechanical methods are too time consuming, complicated, and at this stage too inaccurate to be popular in the field of polymer chemistry. 

Inexact Newton s methods update only a portion of the Hessian and solve the linear system using an iterative method. This family of methods is a hybrid (classical Newton method and conjugate direction method). These methods are useful in solving very large problems (see Chapter 4). 

The classical Newton method has quadratic convergence properties whereas the successive substitution method has linear rate of convergence. However, because of the overshoot, the Newton iteration may fail to converge when the initial estimate is not a good estimate of the solution of the system of nonlinear equations. 

Molecular dynamics [MD] is a well-established technique for simulating the structure and properties of materials in the solid, liquid, and gas phases and will only be briefly overviewed here [1]. In these simulations, N atoms are placed in a simulation cell with an initial set of positions and interact via an interatomic potential 7[r]. The force on each particle is determined by its interaction with all other atoms to within an interaction cutoff specified by 7[r]. For a given set of initial particle positions, velocities, and a specification of the position- or time-dependent forces acting on the particles, MD simulations solve the classical Newton s equations of motions numerically via finite-difference methods to calculate the time evolution of the particle trajectories. 

Hence, the classical Newton process converges very rapidly if the initial guess belongs to the domain of attraction of a stationary point or if an iterate falls into this domain. Exactly that property... 

As explained at the beginning of this section, a quasi-Newton method is obtained when in the classical Newton process the Hessian matrix is approximated by estimate matrices which are computed by an update formula. Subsequently a few results pertaining to the behaviour of convergence of some quasi-Newton methods are presented. Furthermore, the applicability of these methods for locating minimizers and/or saddle points of energy functionals is discussed. 

Remark 3 In contrast to the classical Newton process, a quasi-Newton procedure is not self-correcting, i.e. the errors are accumulated, because a matrix k depends on all preceding matrices i=0(l)k-l. 

It is necessary to notice that the crude representation (28.30) is the first and the last one where the quantization of nuclear motion can be accomplished by means of classical Newton mechanical separations of the degrees of freedom. All other representations will mix the vibrational, rotational and translational modes, and they will not be separable any more. 

---