Modeling Parallel Processes in Biosystems

The problem of building an AI system basing on exhausted symbolic AI paradigm forces to search for a new one in optics and neurophysiology. A pseudooptical neural network approach is considered. Prerequisites of this approach are examined. The model is an artificial network of interfering neurons. Potentials of these neurons change as a result of interference of periodical signals. Modeling of holographic and other optical phenomena is possible in networks of this kind. This pseudo-optical neuronal model is a step toward the artificial holographic brain and brain-distributed associative memory. The process of calculation of the neuron potentials in the network is massively parallel. The model is implemented in software simulator. Parallel computing is considered. The modern state of HPC domain is examined and the trends are observed. A educational Beowulf cluster, supporting MPI, is constructed and tested. The simulator is parallelized to run on cluster.