A model potential-energy function comprising both twoand three-atom contributions is proposed to describe interactions in solid and liquid forms of Si. Implications of this potential are then explored by molecular-dynamics computer simulation, using 216 atoms with periodic boundary conditions. Starting with the diamond-structure crystal at low temperature, heating causes spontaneous nucleation and melting. The resulting liquid structurally resembles the real Si melt. By carrying out steepest-descent mappings of syste~ configurations onto potential-energy minima, two main conclusions emerge: (1) a temperature-independent inherent structure underlies the liquid phase, just as for "simple" liquids with only pair interactions; (2) the Lindemann melting criterion for the crystal apparently can be supplemented by a freezing criterion for the liquid, where both involve critical values of appropriately defined mean displacements from potential minima.

No abstract

Classification of potential energy minima-mechanically stable molecular packings-offers a unifying principle for understanding condensed phase properties. This approach permits identification of an inherent structure in liquids that is normally obscured by thermal motions. Melting and freezing occur through characteristic sequences of molecular packings, and a defect-softening phenomenon underlies the fact that they are thermodynamically first order. The topological distribution of feasible transitions between contiguous potential minima explains glass transitions and associated relaxation behavior.

The "inherent structures" which underlie the liquid state are those stable particle packings (potential minima) which can be reached by a steepest-descent quench on the potential-energy hypersurface. This paper explores the dynamics of transition between distinct inherent structures for a simple classical model of monatomic substances. Molecular-dynamics calculations with 32 and 108 particles have been carried out with running construction of the mapping to potential minima. This determines the distribution of stable packings according to their potential energy and shows how transition rates between alternative structures vary with total system energy. Melting and freezing events have been monitored in this manner. %'e observe occasional transitions in localized "twostate" (bistable) systems in strongly supercooled amorphous states. Transitions in fIuid states display a peculiar intermittency that may have relevance to self-diffusion and viscous flow.

Fully-microscopic No-core Shell Model (NCSM) calculations of all stable s and p shell nuclei are used to determine a realistic N N interaction, JISP16, describing not only the two-nucleon data but the binding energies and spectra of nuclei with A ≤ 16 as well. The JISP16 interaction, providing rapid convergence of the NCSM calculations, is obtained in an ab exitu approach by phase-equivalent transformations of the JISP6 N N interaction. To complement the successful but computationally intensive 'ab initio' No-core Shell Model (NCSM) [1], we introduce the 'ab exitu' NCSM. While the former has proven very successful for light nuclei when one includes three-body (N N N ) forces [2,3], the computational complexity motivates us to introduce an approach that simultaneously minimizes N N N forces while providing more rapid convergence with a pure nucleon-nucleon (N N ) force. We invoke directly an end-goal of nuclear theory (hence the term 'ab exitu'), a successful description of nuclear properties, including the available N N data, to develop a new class of N N potentials that provide accurate descriptions of a broad range of nuclear data. PACSTo achieve this, we form a union of two recent techniques -the J-matrix inverse scattering [4,5,6] and the NCSM [1]. A major ingredient of our approach is the form of the N N interaction (a small matrix in the oscillator basis), which is chosen to provide rapid convergence of manybody observables within the NCSM. Indeed, we show below that results up through A = 16 obtained directly with the bare interaction (one that accurately describes the N N data) are close to those obtained with the effective interaction and are very useful to establish the confidence region for the binding energy.Since this is a departure from the more traditional approach, we motivate our development with observations concerning the successful ab initio approaches to light nuclei. Indeed several promising microscopic approaches have been introduced and tested extensively with realistic N N interactions (see [7] and references therein) and with realistic N N + N N N interactions [8,2,3]. Progress towards heavier nuclei appears limited only by scientific manpower and by available computers. However, all approaches face the exponentially rising computational complexity inherent in the quantum many-body problem with increasing particle number and novel schemes are needed to minimize the computational burden without sacrificing realism and precision.The earliest and most successful in reaching nuclei beyond A = 4 is the Green's-function Monte Carlo (GFMC) approach [8] whose power has been used to determine a sequence of everimproving N N N interactions [8,9,10], in conjunction with highly precise N N interactions [11] that 1