SCI Publications

The issue of mesh quality measures for triangular (and tetrahedral) meshes is considered. A new mesh quality measure is based both on geometrical and solution information and is derived by considering the error in the H 1 norm when linear triangular elements are used to approximate a quadratic function. The new measure is then compared with the recent mesh quality measure based on the L 2 norm. Simple examples are used to show that the choice of norm is critical in deciding what is a good triangulation

The biomedical community is increasingly taking advantage of the power of computing, both to manage and analyze data, and to model biological processes. The working group should investigate the needs of NIH-supported investigators for computing resources, including hardware, software, networking, algorithms, and training. It should take into account efforts to create a national information infrastructure, and look at working with other agencies (particularly NSF and DOE) to ensure that the research needs of the NIH-funded community are met.

It should also investigate the impediments biologists face in utilizing high-end computing, such as a paucity of researchers with cross-disciplinary skills. The panel should consider both today's unmet needs and the growing requirements over the next five years (a reasonable horizon for extrapolating the advances in the rapidly changing fields of computing and computational biology).

The result of deliberations should be a report to the NIH Director, which will be presented to the Advisory Committee to the Director. The report should include recommendations for NIH actions to support the growing needs of NIH-funded investigators for biomedical computing.

The molecular geometries and conformational energies of model molecules of poly(vinylidene fluoride) (PVDF) have been determined from high-level quantum chemistry calculations and have been used in parametrization of a six-state rotational isomeric state (RIS) model for PVDF. The model molecules investigated were 1,1,1,3,3-pentafluorobutane, 1,1,1,3,3,5,5,5-octofluoropentane, 2,2,4,4-tetrafluoropentane, and 2,2,4,4,6,6-hexafluoroheptane (HFH). Analysis of the conformations of these molecules revealed split trans minima (t₊ = 195°, t- = 165°), as was seen in previous studies of perfluoroalkanes. In contrast, the gauche minima, which split in perfluoroalkanes, did not do so in the PVDF model compounds. The lowest energy conformer of HFH, g⁺g⁺g⁺g⁺, was found to be at least 0.4 kcal/mol lower in energy per backbone dihedral than any of the conformers of HFH resembling crystalline polymorphs of PVDF, indicating that intermolecular interactions are important in stabilizing conformations of PVDF in the crystalline phases. A six-state RIS model was able to accurately reproduce the conformer energies of the PVDF model compounds. The RIS analysis revealed that, as in n-alkanes and perfluoroalkanes, the trans conformation of the backbone dihedral is intrinsically lower in energy than the gauche conformation in the PVDF model compounds. However, very large unfavorable second-order interactions between fluorine atoms occur in −CH₂− centered t₊t₊ sequences and, to a lesser extent, t₊g⁺ and t₊g^- sequences. The quantum chemistry based RIS model yielded a characteristic ratio for PVDF in good agreement with experiment, but with significantly different conformer populations than predicted by earlier RIS models, including a much higher gauche probability. The high gauche probability of 65% for unperturbed PVDF chains (at 463 K), greater than that for poly(ethylene) and much greater than that for poly(tetrafluoroethylene), is a consequence of the unfavorable second-order interactions occurring in −CH₂− centered sequences containing trans conformations.

Dimethylnitramine (DMNA) is used as a model system for investigating accurate and efficient electronic structure methods for nitramines. Critical points on the potential energy surfaces of DMNA, CH₃NCH₃, CH₃NCH₂, NO₂, HONO, and the transition state to HONO elimination were located through geometry optimizations using the B1LYP, B3LYP, MPW1PW91, and BH&HLYP density functional methods, in addition to MP2, G2(MP2), and QCISD ab initio theories using the cc-pVDZ basis set. For cost-effective determination of nitramine reaction energetics, highly correlated single-point calculations at DFT geometries are recommended. Our best estimated reaction enthalpies for N-N bond scission and HONO elimination are 41.6 and −0.9 kcal/mol, respectively, determined at the QCISD(T)//QCISD level of theory. These numbers can be reproduced to within 1.3 kcal/mol for the N-N bond and to within 0.5 kcal/mol for the HONO reaction by calculating QCISD(T) energies at B1LYP geometries, thus saving considerable computational cost without sacrificing accuracy. Using the same strategy, the transition state energy for HONO elimination can be modeled to within 0.1 kcal/mol of the QCISD(T)//QCISD result.

An embedded cluster model is used to estimate the molecular dipole moment of crystalline dimethylnitramine (DMNA). The electrostatic potential due to the crystal is included in the calculation via the SCREEP (surface charge representation of the electrostatic embedding potential) approach. The resulting dipole moment for DMNA in the crystalline environment is 6.69 D. This number is more than 40% greater than the gas-phase value and 15% greater than the estimated dipole moment in the liquid phase, thus providing evidence of a strong polarization effect in condensed phases of DMNA.