Computational mechanics of the paddlefish rostrum

Purpose – The rostrum of a paddlefish provides hydrodynamic stability during feeding process in addition to detect the food using receptors that are randomly distributed in the rostrum. The exterior tissue of the rostrum covers the cartilage that surrounds the bones forming interlocking star shaped bones. Design/methodology/approach – The aim of this work is to assess the mechanical behavior of four finite element models varying the type of formulation as follows: linear-reduced integration, linear-full integration, quadratic-reduced integration and quadratic-full integration. Also presented is the load transfer mechanisms of the bone structure of the rostrum. Findings – Conclusions are based on comparison among the four models. There is no significant difference between integration orders for similar type of elements. Quadratic-reduced integration formulation resulted in lower structural stiffness compared with linear formulation as seen by higher displacements and stresses than using linearly formulated elements. It is concluded that second-order elements with reduced integration and can model accurately stress concentrations and distributions without over stiffening their general response. Originality/value – The use of advanced computational mechanics techniques to analyze the complex geometry and components of the paddlefish rostrum provides a viable avenue to gain fundamental understanding of the proper finite element formulation needed to successfully obtain the system behavior and hot spot locations.

A Cluster of FPAAs to Recognize Images using Neural Networks

Analog computing has been recovering its relevance in the recent years. FPAAs are the equivalent to FPGAs but in the analog domain. The main drawback of FPAAs is their reduced integration capacity. In order to increase the amount of analog resources, in this paper a cluster of 40 FPAAs is proposed. As a use case, a 19-8-6-4 feedforward Neural Network has been implemented on such cluster. With the help of a DCT-based software framework, this NN is able to classify 28x28 MNIST images. Results show that the analog network is able to obtain almost the same results as the software baseline network.<br>

A Cluster of FPAAs to Recognize Images Using Neural Networks

Analog computing has been recovering its relevance in the recent years. FPAAs are the equivalent to FPGAs but in the analog domain. The main drawback of FPAAs is their reduced integration capacity. In order to increase the amount of analog resources, in this paper a cluster of 40 FPAAs is proposed. As a use case, a 19-8-6-4 feedforward Neural Network has been implemented on such cluster. With the help of a DCT-based software framework, this NN is able to classify 28x28 MNIST images. Results show that the analog network is able to obtain almost the same results as the software baseline network.<br>

A Cluster of FPAAs to Recognize Images Using Neural Networks

Analog computing has been recovering its relevance in the recent years. FPAAs are the equivalent to FPGAs but in the analog domain. The main drawback of FPAAs is their reduced integration capacity. In order to increase the amount of analog resources, in this paper a cluster of 40 FPAAs is proposed. As a use case, a 19-8-6-4 feedforward Neural Network has been implemented on such cluster. With the help of a DCT-based software framework, this NN is able to classify 28x28 MNIST images. Results show that the analog network is able to obtain almost the same results as the software baseline network.<br>

3-DIMENTIONAL MODELING OF CYLINDRICAL CONCRETE UNDER AXIAL COMPRESSION WITH VARIATIONS OF MESH ELEMENT TECHNIC AND EFFECT OF REDUCED INTEGRATION

Research with numerical methods using computer programs has increased in recent years in various fields including the field of structural materials. However, this numerical analysis still needs to be developed in terms of modeling techniques in order to get the results according to the real condition. In this paper, modeling of concrete-cylinder with numerically method using a finite element based computer program, Abaqus®, presented.. Modeling technique to inventigate the mesh element technique and the reduced integration on hourglass control. Modeling specimen is in the form of cylindrical concrete measuring 150 mm x 300 mm with a quality of 25 MPa. Concrete plastic behavior is modeled with the concrete damage plasticity (CDP) feature. The results of the analysis show that the reduction-integration of hourglass control gives a lower peak load effect when the material passes the elastic limit. Meanwhile, a better mesh technique is to use 8-noded hexahedral because it is more stable and the reduction-integration effect of hourglass control can work.