1.The Markov Chain Mento Carlo method can combines the two kind of parallelism: the coarse-grained and the fine one. We can implement it using MPI code.
2.GPU-optimized libraries of software functions are also being developed, including cuBLAS for basic linear algebra operations,cuFFT for carrying out fast Fourier transforms and Thrust for large vector operations.
3.The bottleneck of Heterogeneous computing is tranferring data between GPU and CPU memory. The more data there are, the more efficient it is.
4.the Amazon Web Services Management Console is a good example of cloud computing.
5.512 is the number of threads that a block could take under CUDA.
6.It takes google translate more than 3 seconds to work out "epidemiologist".
7.In general, statistical gene-gene interaction is defined as departure from additive or multiplicative joint effects of the genetic risk factors.
8.The values of odds-ratio defined in terms of genotypes depends on how to code indicator variables G and H.
9.What is dominant model?