Advancing biomolecular simulation through exascale HPC, AI and quantum computing.

Biomolecular simulation can act as both a digital microscope and a crystal ball; offering the potential for a deeper understanding of experimental observations whilst also presenting a forward-looking avenue for the in silico design and evaluation of hitherto unsynthesized compounds. Indeed, as the intricacy of our scientific inquiries has grown, so too has the computational prowess we seek to deploy in our pursuit of answers. As we enter the Exascale era, this mini-review surveys the computational landscape from both the point of view of the development of new and ever more powerful systems, and the simulations that are run on them. Moreover, as we stand on the cusp of a transformative phase in computational biology, this article offers a contemplative glance into the future, speculating on the profound implications of artificial intelligence and quantum computing for large-scale biomolecular simulations.

Semiempirical Molecular Dynamics (SEMD) I: Midpoint-Based Parallel Sparse Matrix-Matrix Multiplication Algorithm for Matrices with Decay.

In this paper, we present a novel, highly efficient, and massively parallel implementation of the sparse matrix-matrix multiplication algorithm inspired by the midpoint method that is suitable for matrices with decay. Compared with the state of the art in sparse matrix-matrix multiplications, the new algorithm heavily exploits data locality, yielding better performance and scalability, approaching a perfect linear scaling up to a process box size equal to a characteristic length that is intrinsic to the matrices. Moreover, the method is able to scale linearly with system size reaching constant time with proportional resources, also regarding memory consumption. We demonstrate how the proposed method can be effectively used for the construction of the density matrix in electronic structure theory, such as Hartree-Fock, density functional theory, and semiempirical Hamiltonians. We present the details of the implementation together with a performance analysis up to 185,193 processes, employing a Hamiltonian matrix generated from a semiempirical NDDO scheme.

Image distortions of a partially fluorinated hydrocarbon molecule in atomic force microscopy with carbon monoxide terminated tips.

The underlying mechanisms of image distortions in atomic force microscopy (AFM) with CO-terminated tips are identified and studied in detail. AFM measurements of a partially fluorinated hydrocarbon molecule recorded with a CO-terminated tip are compared with state-of-the-art ab initio calculations. The hydrogenated and fluorinated carbon rings in the molecule appear different in size, which primarily originates from the different extents of their π-electrons. Further, tilting of the CO at the tip, induced by van der Waals forces, enlarges the apparent size of parts of the molecule by up to 50%. Moreover, the CO tilting in response to local Pauli repulsion causes a significant sharpening of the molecule bonds in AFM imaging.

Changing computing paradigms towards power efficiency.

Power awareness is fast becoming immensely important in computing, ranging from the traditional high-performance computing applications to the new generation of data centric workloads. In this work, we describe our efforts towards a power-efficient computing paradigm that combines low- and high-precision arithmetic. We showcase our ideas for the widely used kernel of solving systems of linear equations that finds numerous applications in scientific and engineering disciplines as well as in large-scale data analytics, statistics and machine learning. Towards this goal, we developed tools for the seamless power profiling of applications at a fine-grain level. In addition, we verify here previous work on post-FLOPS/W metrics and show that these can shed much more light in the power/energy profile of important applications.

Characterizing and Understanding Divalent Adsorbates on Carbon Nanotubes with Ab Initio and Classical Approaches: Size, Chirality, and Coverage Effects.

The study of oxygen chemisorption on single-walled carbon nanotubes generally relies on simple atomistic models and hence hampers the possibility to understand whether nanotube size or adduct concentration have a role in determining the surface-adsorbate interaction. Our large-scale DFT-based simulations show that structural and electronic properties as well as diffusion barriers strongly depend on both nanotube diameter and adsorbate concentration. Our atomistic models cover nanotube of different chirality with diameters from 0.6 to 1.5 nm and oxygen concentration from 0.1 to 1%. In particular, the tendency to cluster increases with concentration and stabilizes ether (ET) groups but affects hopping barriers only to a minor extent. Significant differences with graphene are found, also for 1.5 nm diameter nanotubes. Extension to species isoelectronic to oxygen reveals dissimilarities, and especially for sulfur that tends to form epoxides (EP), to diffuse more easily and to rapidly close the energy gap for increasing concentration. The relative ET-EP stability can be described in terms of the bare-bond curvature, a concentration-dependent chemical descriptor here introduced. Comparison of these DFT calculations-using different exchange-correlation functionals-and our additional investigation with a reactive force-field (ReaxFF) clarifies several similarities but also discrepancies between the predictions of the two schemes.

Improved coarse-grained model for molecular-dynamics simulations of water nucleation.

We developed a new coarse-grained (CG) model for water to study nucleation of droplets from the vapor phase. The resulting potential has a more flexible functional form and a longer range cutoff compared to other CG potentials available for water. This allowed us to extend the range of applicability of coarse-grained techniques to nucleation phenomena. By improving the description of the interactions between water molecules in the gas phase, we obtained CG model that gives similar results than the all-atom (AA) TIP4P model but at a lower computational cost. In this work we present the validation of the potential and its application to the study of nucleation of water droplets from the supersaturated vapor phase via molecular-dynamics simulations. The computed nucleation rates at T = 320 K and 350 K at different supersaturations, ranging from 5 to 15, compare very well with AA TIP4P simulations and show the right dependence on the temperature compared with available experimental data. To help comparison with the experiments, we explored in detail the different ways to control the temperature and the effects on nucleation.

Understanding the self-healing hydrophobic recovery of high-voltage insulators.

Amorphous siloxane polymers are designed to have high dielectric strength for use as high-voltage insulation materials. Surface hydrophobicity is essential and can be impaired by environmental, electrical, or mechanical factors, leading to leakage currents due to dielectric breakdown. Self-recovery is possible and is generally observed over a period of several hours. Using large-scale, all-atom molecular dynamics simulations, the surface wetting of water droplets on the polymer surface is simulated for various surface conditions, including oxidation and coating with small molecules, to understand the driving forces of the recovery process at the atomistic level, which is of primary importance for the developments of novel materials. In this work, we shed light onto the self-recovery mechanism and propose the use of low-molecular-weight (LMW) siloxane to accelerate the recovery of hydrophobicity.

Mechanisms of propylene glycol and triacetin pyrolysis.

Propylene glycol and triacetin are chemical compounds, commonly used as food additives. Though the usage of the pure chemicals is not considered harmful when used as dietary supplements, little is known about the nature of their thermal degradation products and the impact they may have on human health. For these reasons, in this manuscript we investigate the thermal decomposition mechanisms of both neutral propylene glycol and triacetin in the gas phase by a novel simulation framework. This is based on a free energy sampling methodology followed by an accurate energy refinement. Structures, Gibbs free energy barriers, and rate constants at 800 K were computed for the different steps involved in the two pyrolytic processes. The thermal decomposition mechanisms found theoretically for propylene glycol and triacetin were validated by a qualitative experimental investigation using gas-phase chromatography-mass spectroscopy, with excellent agreement. The results provide a validation of the novel simulation framework and shed light on the potential hazard to the health that propylene glycol and triacetin may have when exposed to high temperatures.

A new piece in the puzzle of lithium/air batteries: computational study on the chemical stability of propylene carbonate in the presence of lithium peroxide.

The electrolyte role in non-aqueous lithium/air batteries is attracting a lot of attention in several research groups, because of its fundamental importance in producing the appropriate reversible electrochemical reduction. While recent published works identify the lithium superoxide as the main degrading agent for propylene carbonate (PC), there is no clear experimental evidence that the oxygen at the cathode interface layer does not reduce further to peroxide before reacting with PC. Here, we investigate the reactivity of lithium peroxide versus propylene carbonate and find that Li(2)O(2) irreversibly decomposes the carbonate solvent, leading to alkyl carbonates. We also show that, compared with a single Li(2)O(2) unit in PC, a crystalline surface of Li(2)O(2) exhibits an enhanced reactivity. Our findings support the possibility that in lithium/air cells, oxygen may still be reduced to peroxide, with the formation of solid Li(2)O(2), which degrades by decomposing PC.

Reaction Dynamics of ATP Hydrolysis in Actin Determined by ab Initio Molecular Dynamics Simulations.

Energy released by the hydrolysis of the high-energy phosphate bond of nucleoside triphosphate (NTP) cofactors is the driving force behind most biological processes. To understand how this energy is used to induce differences in protein structure and function, we examine the transfer of vibrational energy into the nucleotide-bound actin active site immediately after reaction activation. To this end, we perform Born-Oppenheimer molecular dynamics simulations of the active site at the level of density functional theory (DFT) starting at the calculated transition state (TS) structure. Similarly to the mechanism determined in many nucleotide-bound protein systems, the Os-Pγ bond is first elongated. Then, nucleophilic attack of the lytic water on Pγ occurs. Subsequently, protons are transferred in a cycle formed by water molecules, a protein residue, Asp154, and the γ-phosphate group, resulting in the formation of H2PO4(-). To investigate the possible creation of excited vibrational states in the products, power spectra of bond-length autocorrelation functions for relevant bonds within the active site are compared for simulations that start at the TS, at reactants, and at reaction end products. The hydroxyl bond formed in the final proton transfer to the phosphate molecule is observed to exhibit relatively high kinetic energies and large oscillations during reaction. It is also likely that some of the energy released by the reaction is captured by the low-energy stretching vibrations of the phosphoryl bonds of orthophosphate, which oscillate with large amplitudes in nonequilibrium simulations of end products.

Surface dynamics of amorphous polymers used for high-voltage insulators.

Amorphous siloxane polymers are the backbone of high-voltage insulation materials. The natural hydrophobicity of their surface is a necessary property for avoiding leakage currents and dielectric breakdown. As these surfaces are exposed to the environment, electrical discharges or strong mechanical impact can temporarily destroy their water-repellent properties. After such events, however, a self-healing process sets in and restores the original hydrophobicity within some hours. In the present study, we investigate possible mechanisms of this restoration process. Using large-scale, all-atom molecular dynamics simulations, we show that molecules on the material surface have augmented motion that allows them to rearrange with a net polarization. The overall surface region has a net orientation that contributes to hydrophobicity, and charged groups that are placed at the surface migrate inward, away from the vacuum interface and into the bulk-like region. Our simulations provide insight into the mechanisms for hydrophobic self-recovery that repair material strength and functionality and suggest material compositions for future high-voltage insulators.

High-resolution molecular orbital imaging using a p-wave STM tip.

Individual pentacene and naphthalocyanine molecules adsorbed on a bilayer of NaCl grown on Cu(111) were investigated by means of scanning tunneling microscopy using CO-functionalized tips. The images of the frontier molecular orbitals show an increased lateral resolution compared with those of the bare tip and reflect the modulus squared of the lateral gradient of the wave functions. The contrast is explained by tunneling through the p-wave orbitals of the CO molecule. Comparison with calculations using a Tersoff-Hamann approach, including s- and p-wave tip states, demonstrates the significant contribution of p-wave tip states.

A revisited picture of the mechanism of glycerol dehydration.

The dehydration mechanism of neutral glycerol in the gas phase was investigated by means of metadynamics simulations. Structures, vibrational frequencies, Gibbs free energy barriers, and rate constants at 800 K were computed for the different steps involved in the pyrolytic process. In this article, we provide a novel mechanism for the dehydration of neutral glycerol, proceeding via formation of glycidol with a barrier of 66.8 kcal/mol. The formation of glycidol is the rate limiting step of the overall decomposition process. Once formed, glycidol converts into 3-hydroxypropanal with a barrier of 49.5 kcal/mol. 3-Hydroxypropanal can decompose further into acrolein or into formaldehyde and vinyl-alcohol with barriers of 53.9 and 35.3 kcal/mol, respectively. These findings offer new insights to available experimental data based on glycerol pyrolysis studies performed with isotopic labeling and on the interpretation of the chemistry of glycerol and sugars in pyrolytic conditions.

Molecular motion of amorphous silicone polymers.

In this paper, we have studied silicone polymers based on poly(dimethylsiloxane) (PDMS) molecules, which have versatile applications in many fields because of their flexible molecular properties. These polymers are of interest because when used for high-voltage insulation, surfaces exposed to weather need to be hydrophobic because a hydrophilic surface can cause leakage currents. Indeed, after damaging electrical discharges, self-recovery of the hydrophobic surface occurs, requiring molecular diffusion and surface reconstruction for repair. We use large-scale, all-atom molecular dynamics simulations that enable an atomic-level description of molecular motion in mixed, amorphous, PDMS-based materials. The local properties that contribute to enhanced molecular motion are characterized based on their local structural and electrostatic environment. With this knowledge, molecular components with desirable diffusion properties may be designed for improved material functionality.

Large-scale simulations of a-Si:H: the origin of midgap states revisited.

Large-scale classical and quantum simulations are used to generate a-Si:H structures. The bond-resolved density of the occupied electron states discloses the nature of microscopic defects responsible for levels in the gap. Highly strained bonds give rise to band tails and midgap states. The latter originate mainly from stretched bonds, in addition to dangling bonds, and can act as hole traps. This study provides strong evidence for photoinduced degradation (Staebler-Wronski effect) driven by strain, thus supporting recent work on a-Si, and sheds light on the role of hydrogen.

Ab initio simulation of the equation of state and kinetics of shocked water.

We report herein first principles simulations of water under shock loading and the chemical reactivity under these hot, compressed conditions. Using a recently developed simulation technique for shock compression, we observe that water achieves chemical equilibrium in less than 2 ps for all shock conditions studied. We make comparison to the experimental results for the Hugoniot pressure and density final states. Our simulations show that decomposition occurs through the reversible reaction H(2)O <--> H(+) + OH(-), in agreement with experiment. Near the approximate intersection of the Hugoniot and the Neptune isentrope, we observe high concentrations of charged species that contribute electronic states near the band gap.

Ultrafast transformation of graphite to diamond: an ab initio study of graphite under shock compression.

We report herein ab initio molecular dynamics simulations of graphite under shock compression in conjunction with the multiscale shock technique. Our simulations reveal that a novel short-lived layered diamond intermediate is formed within a few hundred of femtoseconds upon shock loading at a shock velocity of 12 kms (longitudinal stress>130 GPa), followed by formation of cubic diamond. The layered diamond state differs from the experimentally observed hexagonal diamond intermediate found at lower pressures and previous hydrostatic calculations in that a rapid buckling of the graphitic planes produces a mixture of hexagonal and cubic diamond (layered diamond). Direct calculation of the x-ray absorption spectra in our simulations reveals that the electronic structure of the final state closely resembles that of compressed cubic diamond.

Metal-carbon nanotube contacts: the link between Schottky barrier and chemical bonding.

The field effect transistor based on carbon nanotubes (CNT) is a very promising candidate for post-CMOS microelectronics. Transport in the CNT channel is dominated by the Schottky barriers existing at the metal source contacts. The nature of the metal and the geometry of the contact appear to influence strongly the electrical behavior, but the mechanism is still rather obscure. Extensive calculations based on density functional theory performed for both end and side contacts and for two metals of very different nature, namely, Al and Pd, allow us to identify a clear connection between the character of the chemical bonding and the height of the Schottky barrier (SBH). Our results emphasize that a low SBH for hole conduction in a CNT implies that the pi-electron system of the latter is almost exclusively involved in the chemical bonding with the metal atoms at the interface and that the bonding is not too strong so that both orbital hybridization and topology are preserved. This is the case for Pd in both end and side configurations and to a large extent for Al but in the side geometry only. On the other hand, the coupling of the metal states with the sigma-like system or, in other words, the perturbation of the conjugation of the pi-system via sp3 C-hybridization is the mechanism that enhances the SBH. This is especially evident in the end contact with Al. By showing how the chemistry at the interfaces determines the SBH, our findings open the possibility of better controlling and designing "good contacts".

Anomalous behavior of the dielectric constant of hafnium silicates: a first principles study.

We present an extensive ab initio study of the structural and dielectric properties of hafnium silicates Hf(x)Si(1-x)O(2) that accounts for the observed anomalous dependence on composition of the static dielectric constant in the entire x range. The results reveal that this complex behavior reflects that of the structural development with x, from silica to hafnia, and clarify how different growth processes can also lead to scattered sets of data. Several simple models proposed thus far to explain part of the experimental data are shown to be inadequate. It is argued that silicate layers with low hafnium content form at the HfO(2)/Si interface and play a crucial role in preserving high electron mobility in the channel.