We present a molecular dynamics scheme which combines first-principles and machine-learning (ML) techniques in a single information-efficient approach. Forces on atoms are either predicted by Bayesian inference or, if necessary, computed by on-the-fly quantum-mechanical (QM) calculations and added to a growing ML database, whose completeness is, thus, never required. As a result, the scheme is accurate and general, while progressively fewer QM calls are needed when a new chemical process is encountered for the second and subsequent times, as demonstrated by tests on crystalline and molten silicon. [18] to fit "once and for all" the DFT potential energy surface (PES), after which atomic forces are obtained by analytic differentiation. Similar to classical FFs, a fixed highquality parametrized PES simultaneously ensures fast force evaluation and reliable interpolation. However, accuracy is still not guaranteed to transfer to chemical situations not represented in the fitting database.Here, we propose an alternative machine-learning (ML) based scheme where we allow a stream of fresh quantummechanical (QM) calculations to augment the ML database during each MD simulation, enabling safe interpolation. The scheme could equally be viewed as an efficient FPMD approach where we seek to compute only the QM information necessary to progress the simulation, while retaining the very broad applicability of FPMD. To minimize the QM workload of the MD simulation, one can start by noticing that ML-predicted atomic forces will suffice as long as the dynamics visits configuration is "well represented" within the existing database. Thus, an ideal ML MD scheme should not attempt to increase its database through additional QM calculations until "something new" happens that necessitates this. This is a central guideline for the present work, significantly improving on an earlier scheme [19,20] where all QM information was used once and afterwards discarded. Our new scheme has the potential to reduce the cost of FPMD for the vast range of problems where it is already applied [2], and to extend its use to problems currently beyond reach because of prohibitive time and/or length scales. Here, we test it on standard benchmark physical systems [3,17,18,20], comprising crystalline and molten silicon over a wide range of temperatures and bonding geometries, in both insulating and metallic conditions (Figs. 1-3). FIG. 1 (color online).Comparison of the bulk Si phonon spectrum calculated with DFTB (blue), and SW (black) and with the ML on-the-fly (MLOTF) approach (red), computed with the finite displacement Parlinksi-Li-Kawazoe method [21,22] using a standard σ err ¼ 0.05 eV=Å (dotted lines) and a highaccuracy σ err ¼ 5 × 10 −4 eV=Å value (solid lines) for the ML data noise parameter. The ML database was constructed from a 300 K MD trajectory.

The search for the stable monatomic forms of solid nitrogen is of great importance in view of its potential application as a high-energy-density material. Based on the results of evolutionary structure searches, we proposed two high-pressure polymeric structures to be stable beyond the stability field of the synthesized cubic gauche structure--the layered Pba2 or Iba2 (188-320 GPa) and the helical tunnel P2_{1}2_{1}2_{1} structure (>320 GPa). We rule out the low-temperature stability of the earlier proposed black phosphorus structure. Stability fields of the newly predicted polymorphs are within the reach of current experimental techniques.

There is great interest in the exploration of hydrogen-rich compounds upon strong compression where they can become superconductors. Stannane (SnH 4 ) has been proposed to be a potential high-temperature superconductor under pressure, but its high-pressure crystal structures, fundamental for the understanding of superconductivity, remain unsolved. Using an ab initio evolutionary algorithm for crystal structure prediction, we propose the existence of two unique high-pressure metallic phases having space groups Ama2 and P6 3 ∕mmc, which both contain hexagonal layers of Sn atoms and semimolecular (perhydride) H 2 units. Enthalpy calculations reveal that the Ama2 and P6 3 ∕mmc structures are stable at 96-180 GPa and above 180 GPa, respectively, while below 96 GPa SnH 4 is unstable with respect to elemental decomposition. The application of the Allen-Dynes modified McMillan equation reveals high superconducting temperatures of 15-22 K for the Ama2 phase at 120 GPa and 52-62 K for the P6 3 ∕mmc phase at 200 GPa.hydrogen-rich compounds | metallization | electron-phonon coupling R elatively high-temperature superconductivity is now documented in light-element metals such as Li under pressure (1-3) and MgB 2 (4), where transition temperatures T c up to 20 K and 39 K, respectively, are observed. There is great interest in exploration of unique superconducting phases in other lightelement materials because their high phonon frequencies can enhance electron-phonon coupling (see ref. 5). As the lightest element, hydrogen at very high densities is also predicted to be a superconductor with high transition temperatures (6-8). Experiments indicate that the predicted metallic and superconducting states of hydrogen remain above ∼300 GPa (9-11). It has been proposed that hydrogen-rich compounds (e.g., group IVa hydrides (12)) are expected to metallize at pressures considerably lower than pure hydrogen due to the chemical "precompression" caused by heavier elements; these metallization pressures may fall within the range of current capabilities of static compression techniques. The exploration of potential superconductivity in these hydrogen-rich compounds (e.g., SiH 4 , GeH 4 , and SnH 4 ) is thus desirable and numerous studies have been performed (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25). Strikingly, recent experiments (15,18) show that SiH 4 transforms to a metallic phase near 50-60 GPa with a superconducting T c of 17 K at 96 and 120 GPa, though debate remains (26). We have recently predicted (17) that GeH 4 becomes a high-temperature superconductor with a T c of 64 K at 220 GPa. A theoretical study of SnH 4 (21) predicts that its T c can be even higher, reaching the value of 80 K. Using simulated annealing and geometry optimization, that study found that the high-pressure phase of SnH 4 has P6∕mmm symmetry with a layered structure intercalated by molecular H 2 units, wherein the nearest H-H distance, 0.84 Å, is short enough to be considered as covalent bonding, but significantly longer than the 0.74 Å in the free H ...

The present study was designed to identify the changes in microvesicle-dipeptidyl peptidase-IV (DPP IV) levels in human urine and serum, and to determine whether there were correlations with the severity of diabetic kidney disease (DKD). A total of 127 patients with type 2 diabetes mellitus (T2DM) were divided into three groups according to the urinary albumin/ creatinine ratio (UACR): microalbuminuria group (n = 50); macroalbuminuria group (n = 34) and normoalbuminuria group (n = 43), and 34 age-and sex-matched non-diabetic healthy subjects were selected as controls. Microvesicle-bound DPP IV and free urinary DPP IV were separated by a filtra-centrifugation method. The total microvesicles were captured by a specific monoclonal antibody, AD-1. DPP IV activity was determined by measuring the cleavage of chromogenic free 4-nitroaniline from Gly-Pro-p-nitroanilide at 405 nm with an ELISA plate reader. DPP IV protein levels were determined by ELISA and Western blot. Our results showed that the microvesicle-bound type was the major form of DPP IV in urine; the urinary microvesicle-DPP IV excretion of each T2DM group was significantly higher compared with controls. The urinary microvesicle-DPP IV level was positively correlated with UACR in patients with T2DM. These findings suggest that the urinary level of microvesicle-bound DPP IV is associated with the severity of DKD.

No abstract