Hessian matrices theory

Quantum mechanical calculations are restricted to systems with relatively small numbers of atoms, and so storing the Hessian matrix is not a problem. As the energy calculation is often the most time-consuming part of the calculation, it is desirable that the minimisation method chosen takes as few steps as possible to reach the minimum. For many levels of quantum mechanics theory analytical first derivatives are available. However, analytical second derivatives are only available for a few levels of theory and can be expensive to compute. The quasi-Newton methods are thus particularly popular for quantum mechanical calculations. 

The transition state theory of Eyring or its extensions due to Truhlar and coworkers (see, for example, D. G. Truhlar and B. C. Garrett, Ann. Rev. Phys. Chem. 35, 159 (1984)) allow knowledge of the Hessian matrix at a transition state to be used to compute a rate coefficient krate appropriate to the chemical reaction for which the transition state applies. 

There are a few points with respect to this procedure that merit discussion. First, there is the Hessian matrix. With elements, where n is the number of coordinates in the molecular geometry vector, it can grow somewhat expensive to construct this matrix at every step even for functions, like those used in most force fields, that have fairly simple analytical expressions for their second derivatives. Moreover, the matrix must be inverted at every step, and matrix inversion formally scales as where n is the dimensionality of the matrix. Thus, for purposes of efficiency (or in cases where analytic second derivatives are simply not available) approximate Hessian matrices are often used in the optimization process - after aU, the truncation of the Taylor expansion renders the Newton-Raphson method intrinsically approximate. As an optimization progresses, second derivatives can be estimated reasonably well from finite differences in the analytic first derivatives over the last few steps. For the first step, however, this is not an option, and one typically either accepts the cost of computing an initial Hessian analytically for the level of theory in use, or one employs a Hessian obtained at a less expensive level of theory, when such levels are available (which is typically not the case for force fields). To speed up slowly convergent optimizations, it is often helpful to compute an analytic Hessian every few steps and replace the approximate one in use up to that point. For really tricky cases (e.g., where the PES is fairly flat in many directions) one is occasionally forced to compute an analytic Hessian for every step. 

A few last technical points merit some discussion prior to an assessment of the relative utilities of different theoretical levels for prediction of IR spectra. First, note that the first derivatives in the Taylor expansion disappear only when the potential is expanded about a critical point on the PES (since then the gradients are all zero). Thus, the form of Eq. (9.49) is not valid if the level of theory used in the computation of the Hessian matrix differs from... 

Each such null vector may be considered an invariant or symmetry of the thermodynamic system, because it corresponds to an operation (change of extensive variables Xt) that produces no response in any intensive state variable and thus leaves the thermodynamic state unaltered (Sidebar 7.2). As described in Sidebar 10.3, these invariants also play a role somewhat analogous to overall rotations and translations ( null eigenmodes of the Hessian matrix) in the theory of molecular vibrations. 

Standard representation of the TS in organic chemistry textbooks is the point of maximum energy on the reaction coordinate. More precise is the definition provided in Section 1.6 the TS is the col, a point where aU the gradients vanish, and all of the eigenvalues of the Hessian matrix are positive except one, which corresponds to the reaction coordinate. In statistical kinetic theories, a slightly different definition of the TS is required. 

The matrix (B —CM" C ) is called the partitioned orbital Hessian matrix because of its connection to matrix partitioning theory. While the matrix B is the matrix of second derivatives with respect to orbital rotations when the CSF mixing coefficients are held constant, the partitioned orbital Hessian matrix is the matrix of second derivatives with respect to orbital rotations when the CSF eoefficients relax optimally with the orbital changes. The matrix of Eq. (152), consisting of the partitioned orbital Hessian matrix and the orbital gradient vector, is called the augmented partitioned orbital Hessian matrix. 

---