Symmetric and asymmetric buried explosive hazards (BEHs) present real, persistent, deadly threats on the modern battlefield. Current approaches to mitigate these threats rely on highly trained operatives to reliably detect BEHs with reasonable false alarm rates using handheld Ground Penetrating Radar (GPR) and metal detectors. As computers become smaller, faster and more efficient, there exists greater potential for automated threat detection based on state-of-the-art machine learning approaches, reducing the burden on the field operatives. Recent advancements in machine learning, specifically deep learning artificial neural networks, have led to significantly improved performance in pattern recognition tasks, such as object classification in digital images. Deep convolutional neural networks (CNNs) are used in this work to extract meaningful signatures from 2-dimensional (2-D) GPR B-scans and classify threats. The CNNs skip the traditional "feature engineering" step often associated with machine learning, and instead learn the feature representations directly from the 2-D data. A multi-antennae, handheld GPR with centimeter-accurate positioning data was used to collect shallow subsurface data over prepared lanes containing a wide range of BEHs. Several heuristics were used to prevent over-training, including cross validation, network weight regularization, and "dropout." Our results show that CNNs can extract meaningful features and accurately classify complex signatures contained in GPR B-scans, complementing existing GPR feature extraction and classification techniques.

The experimental shock tube data recently reported by Kiefer et al. [J. Phys. Chem. A 2004, 108, 2443-2450] for the title reaction at temperatures between 1600 and 2400 K have been compared to master equation simulations using three models: (a) standard RRKM theory, (b) RRKM theory modified by local random matrix theory, which introduces dynamical corrections arising from slow intramolecular vibrational energy randomization, and (c) an ad hoc empirical non-RRKM model. Only the third model provides a good fit of the Kiefer et al. unimolecular reaction rate data. In separate simulations, all three models accurately reproduce the experimental 300 K chemical activation data of Marcoux and Setser [J. Phys. Chem. 1978, 82, 97-108] when the energy transfer parameters are freely varied to fit the data. When experimental energy transfer parameters for a geometrical isomer (1,1,2-trifluoroethane) are used, the standard RRKM model fits the chemical activation data better than the other models, but if energy transfer in the 1,1,1-trifluoroethane is significantly reduced in comparison to the 1,1,2 isomer, then the empirical ad hoc non-RRKM model also gives a good fit. While the ad hoc empirical non-RRKM model can be made to fit the data, it is not based on theory, and we argue that it is physically unrealistic. We also show that the master equation simulations can mimic the Kiefer et al. vibrational relaxation data, which was the first shock tube observation of double-exponential relaxation. We conclude that, until more data on the trifluoroethanes become available, the current evidence is insufficient to decide with confidence whether non-RRKM effects are important in this reaction, or whether the Kiefer et al. data can be explained in some other way.

Classical trajectory calculations on intramolecular vibrational energy redistribution (IVR) involving the torsion in 1,1,1-trifluoroethane (TFE) are reported. Two potential energy functions (PEFs) are used to describe the potential energy surface. The "full" PEF gives excellent agreement with the experimental vibrational frequencies. The "simple" PEF omits nondiagonal interaction terms, but still gives very good agreement with the experimental frequencies. The "simple" PEF is intended to minimize mode-mode coupling. Neither PEF includes the HF elimination reaction. Calculations are carried out both with nominal microcanonical selection of initial coordinates and momenta, and with a modified selection method that places controlled amounts of energy in the torsion. Total (classical) vibrational energies from 0.005 to 140 kcal mol(-1) are investigated. The calculated time constants describing energy flow out of the torsional mode are <10 ps for classical vibrational energies near the classical reaction threshold energy (approximately 75 kcal mol(-1)) and greater. It is found that the rate of decay from the torsion largely depends on the amount of energy in the other vibrational modes. Analysis using power spectra shows that the torsional mode in TFE is strongly coupled to the other vibrational modes. These results strongly suggest that vibrational energy in TFE will not be sequestered in the torsion for time periods greater than a few tens of picoseconds when the molecule has enough energy to react via HF elimination.