Three-dimensional seismic refl ection data reveal the presence of a low seismic velocity zone (LVZ) with weak refl ectivity character along the Nankai accretionary prism. This LVZ is intercalated between an upper, offscraped layer and a lower, underthrusting layer in the outer accretionary wedge. Wide-angle ocean bottom seismograph data also support the presence of the LVZ, which is estimated to be a maximum of ~2 km thick, ~15 km wide, and ~120 km long. The LVZ could be an underthrust package underplated in response to the lateral growth of the Nankai accretionary prism. Underplating of the underthrusting layer beneath the overlying offscraped layer would maintain a critical taper of the accretionary prism so that the offscraped layer can continue to grow seaward. The LVZ could have elevated fl uid pressure, leading to rigidity reduction of the entire outer accretionary wedge. The rigidity-lowered outer wedge, containing the LVZ, may be more easily uplifted and thus eventually foster tsunami generation during a Nankai megathrust earthquake. If the fl uid-rich LVZ supplies a signifi cant amount of the fl uid to the megasplay fault zone at depth, it may affect stick-slip behavior of the fault.

This paper presents a new heroic computing method for unstructured, low-order, finite-element, implicit nonlinear wave simulation: 1.97 PFLOPS (18.6% of peak) was attained on the full K computer when solving a 1.08T degrees-of-freedom (DOF) and 0.270T-element problem. This is 40.1 times more DOF and elements, a 2.68-fold improvement in peak performance, and 3.67 times faster in time-to-solution compared to the SC14 Gordon Bell finalist's state-ofthe-art simulation. The method scales up to the full K computer with 663,552 CPU cores with 96.6% sizeup efficiency, enabling solving of a 1.08T DOF problem in 29.7 s per time step. Using such heroic computing, we solved a practical problem involving an area 23.7 times larger than the state-of-the-art, and conducted a comprehensive earthquake simulation by combining earthquake wave propagation analysis and evacuation analysis. Application at such scale is a groundbreaking accomplishment and is expected to change the quality of earthquake disaster estimation and contribute to society. Categories: Time-to-solution, Scalability, Peak performance I. CONTRIBUTIONS OF SUPERCOMPUTERS TO REDUCING EARTHQUAKE DISASTERS A. Overview and importance of the problem An earthquake can affect many people. The 2011 Tohoku Earthquake in Japan killed 20,000 and more than 200,000 people are still in temporary housing. The loss of lives, damage to the economy, and catastrophic damage are fresh in our memory. This damage occurred in Japan, a country that leads the world in earthquake disaster mitigation, and there are concerns over similar disasters in earthquake-prone mega-cities such as Los Angeles, San Francisco, and Tokyo. Reliable earthquake disaster estimation plays an important role in mitigating such disasters. Physics-based comprehensive earthquake simulation is the only way to make reliable estimations of such infrequent and untestable events, and is

No abstract

SUMMARYThis paper presents the particle discretization scheme (PDS) to analyze brittle failure of solids. The scheme uses characteristic functions of Voronoi and Delaunay tessellations to discretize a function and its derivatives, respectively. A discretized function has numerous discontinuities so that these discontinuities are utilized as a candidate of crack path segment in modeling propagating cracks, without making any extra computation to accommodate new displacement discontinuities. When the scheme is implemented to a finite element method (FEM), the resulting stiffness matrix coincides with the one that is obtained by using linear elements. The accuracy of computing a stress intensity factor at a crack tip is examined. It is shown that the accuracy is better than that of a FEM with linear elements when the rotational degree of freedom is included in discretizing displacement functions. Three three-dimensional growing crack problems are solved by means of the PDS and the results are presented.

This paper proposes meta-modeling as a methodology for constructing a model that is consistent with continuum mechanics. Consistency means solving a Lagrangian of continuum mechanics by using a particular subset of continuum mechanics' function space, so that a solution of a consistent model is rigorously converted to a solution of continuum mechanics or vice versa. This conversion enables us to make smart use of solid and structure element analysis. Meta-modeling is applied to beam and plate problems, and it is shown that Rayleigh beam and Kirchhoff-Love plate are consistent with continuum mechanics.