Improving the performance of GPU-based genetic programming through exploitation of on-chip memory

Genetic Programming (GP) (Koza, Genetic programming, MIT Press, Cambridge, 1992) is well-known as a computationally intensive technique. Subsequently, faster parallel versions have been implemented that harness the highly parallel hardware provided by graphics cards enabling significant gains in the performance of GP to be achieved. However, extracting the maximum performance from a graphics card for the purposes of GP is difficult. A key reason for this is that in addition to the processor resources, the fast on-chip memory of graphics cards needs to be fully exploited. Techniques will be presented that will improve the performance of a graphics card implementation of tree-based GP by better exploiting this faster memory. It will be demonstrated that both L1 cache and shared memory need to be considered for extracting the maximum performance. Better GP program representation and use of the register file is also explored to further boost performance. Using an NVidia Kepler 670GTX GPU, a maximum performance of 36 billion Genetic Programming Operations per Second is demonstrated.     

Prenatal screening for Down syndrome

OBJECTIVE: To investigate the fetal angiotensin II type 1 receptor genotype in pre-eclampsia. DESIGN: Case-control study. POPULATION: Forty-one maternal-fetal pairs from pre-eclamptic pregnancies and 80 maternal-fetal pairs from normotensive pregnancies. METHODS: Maternal and fetal DNA was genotyped at three diallelic polymorphisms, at nucleotides 573, 1062, and 1166, in the coding exon of the angiotensin II type 1 receptor gene, and at a dinucleotide repeat polymorphism in its 3' flanking region. RESULTS: Allele and genotype frequencies at the four polymorphic regions investigated did not differ between pre-eclamptic and normotensive groups, in either fetal or maternal samples. Mothers heterozygous for the dinucleotide repeat allele designated A4 transmitted this allele to the fetus in 15 of 18 informative pre-eclamptic pregnancies and in eight of 26 normotensive pregnancies. This was greater than the expected probability in pre-eclamptic pregnancies (p=0.04) and less than expected in normotensive pregnancies (p<0.005). The 573T variant, which is in partial linkage disequilibrium with the A4 allele, showed a similar distortion of maternal-fetal transmission. CONCLUSION: Angiotensin II type 1 receptor gene expression in the fetus may contribute to the aetiology of pre-eclampsia. It is unclear whether susceptibility is conferred by the fetal genotype acting alone, or by allele sharing by mother and fetus. Possible mechanisms for the effect of the angiotensin II type 1 receptor gene are suggested by the association of the 573T variant with low levels of surface receptor expression on platelets. If receptor expression is similarly genetically determined in the placenta, responsiveness to angiotensin II may be affected, with the potential to influence placentation or placental prostaglandin secretion.     

Control of seismic response of piping systems and components in power plants by base isolation

Standard approaches to the aseismic design of piping systems and components for power plants — such as the floor spectrum method — are costly and neglect equipment-structure interaction. Such interaction has recently been shown to be severe when the natural frequency of a component is close or equal to one of the natural frequencies of the primary structure — a situation referred to as tuning and one almost certain to occur in a large structural system. A base isolation system is described that has been demonstrated to reduce dramatically the accelerations induced in structures under earthquake motion. A series of further experiments is described in the paper; this experimental work demonstrates that the response of equipment in structures so isolated is also greatly reduced. Thus, sensitive internal equipment can be protected directly from seismic attack; interaction need not to be considered and inelastic analyses need not be performed during the design process. Due to rigid body action of the primary system above the base isolation system, multiple support response spectra design methods are not needed. It is anticipated that the use of base isolation will reduce the cost of the design and construction of power plant components, piping systems, and structures.     

Testing of a wind restraint for aseismic based isolation

Base isolation has been proposed as an economical approach to aseismic design for many types of buildings and structural systems. In general, the base of a structure is isolated when a support system with extremely low horizontal stiffness limits the transmittal of horizontal shear from the ground to the structure. The displacement of such a structure under wind loading will, however, be excessive since the horizontal stiffness of the support system is so low. The horizontal deflection of a building with an isolated natural frequency of 0.5 Hz would, for example, be four inches for a wind load of one-tenth the weight of the building. While such a horizontal deflection would not be acceptable, a base isolation system could be equipped with a mechanical fuse which would be sufficiently strong to resist wind loading, but which would fracture during an earthquake, leaving the building free on the isolation system.In this report we describe an extensive series of shaking table tests of such a device. The wind restraint consisted of a notched shear pin. Several shear pins were tested in conjunction with a natural rubber isolation system placed beneath a three-storey 40,000 lb steel frame model structure. Pins with breaking forces ranging from 3% to 20% of the weight of the model were tested. The model structure was subjected to various peak accelerations of three earthquake inputs.The shear pins fractured rapidly and cleanly. The breaking force for each pin was reasonably predictable. Although higher mode accelerations were induced in the model structure when the pins broke, these disappeared rapidly. The design of a shear pin mechanical fuse system for a full-scale structure is discussed at the end of this report in view of the experimental results described.     

Dynamic response of an opening in jointed rock

A large-scale test was performed on a 0.4 m diameter lined cylindrical opening in a 2 m cube of jointed limestone for the purpose of gathering data to develop and validate computational models for the dynamic response of structures in jointed rock. The limestone cube was assembled from over four thousand 51 mm square×0.6 to 1.2 m long limestone bricks and was then embedded in a concrete test bed and loaded explosively with a 100 MPa, cylindrically diverging stress wave. The material velocity around the boundary of the jointed limestone cube was measured with accelerometers and these data were used to generate a continuous velocity boundary condition that can be used for computational simulations. Additional instrumentation measured stress in the limestone, joint slip and deformation of the opening. After the test, the limestone cube was recovered and carefully disassembled to reveal fracture patterns in the limestone bricks. Vertical fractures emanating from the springlines formed a chimney of bricks that slipped measurably relative to the rest of the rock mass above the opening and to a lesser degree below the opening.     

Long-term corrosion of rebars embedded in aerial and hydraulic binders - Parametric study and first step of modelling

The prediction of long-term behaviour of reinforced concrete structures involved in the nuclear industry requires a phenomenological modelling of the rebars corrosion processes. Previous analytical characterisation of archaeological artefacts allowed to identify a typical layout constituted of four layers (the metal, the dense product layer, the transformed medium and the binder). Additional experiments leaded to identify the long-term corrosion mechanisms. Following these results, this paper proposes an analytical model of long-term corrosion of rebars embedded in concrete. This modelling is considering the kinetic of oxygen diffusion through the system and its consumption at the metal/dense product layer interface as a function of concrete water saturation degree. Corrosion products thicknesses estimated with the model are then compared to corrosion product thicknesses measured on archaeological artefacts. A parametric study is performed and demonstrates that the oxygen diffusivity and the kinetic constant of the cathodic reaction affect in a wide range the model results.     

Sri Lanka and Canada: A comparison of telematic policy developments

The Canadian experience, though not wholly relevant to small developing countries like Sri Lanka, may have long term relevance. Canada took the strategic path of developing its telecommunications (including broadcasting) infrastructure as a response to the convergence of technologies and problems of cultural diversity and demographic dispersion. It did not simultaneously provide the impetus for the development of computer products and services which could be demanded by the expanded Canadian telemarket, leading to market penetration by overseas producers. Sri Lanka, despite its low capability in the area of telematics and informatics, does not perceive external penetration to be a threat, because of the relatively underdeveloped telecommunications systems. However, the new technologies have transborder dimensions which must be carefully examined in terms of opportunities and threats presented to a country's industrialization, investment promotion, telecommunications, energy, education and information policies.     

Some Source-Dependent effects of unbounded fires

Measurements of time-mean axial velocity and temperature in simulated fires of low height to width ratio are reported. The flame length scales as Q²/D⁴ and the implications of this on earlier findings of the authors [4], with longer flame systems where length scales as Q^2/5, are discussed. Empirical derivations of virtual source location for use in the plume equations above the flame tip are compared.     

Third world entry into the electronic age

The leadership of Western nations in technology development, especially in telecommunications continues to be strengthened. But where does the third world fit into this high-tech scenario. Many experts believe that Western communication technologies can allow developing countries to leapfrog over the industrial age. However, this article points out that unless third world countries can develop their own infrastructure, skills and knowledge base, competing with Western nations for them could become technological Blind Man's Bluff.