The U-machine is a computer architecture oriented toward
general-purpose analog and digital parallel computation in new
technologies, such as those emerging from nanotechnology. Most of
the computation is implemented by means of matrix multiplication and
the application of simple functions to the matrix elements. This
simulator will be used as part of the U-Machine Research Project in
order to test alternative ways of solving problems on the U-machine and
to investigate its accuracy and robustness in the face of failures,
noise, manufacturing defects, etc.
For the input phase, the simulator reads a file of simple textual
commands (like a command-line interpreter), although a graphical user
interface would be an acceptable alternative. As a result, the
simulator defines a number of matrices in its internal memory, reads
input data files, etc. While it is simulating the operation of
the machine, it performs matrix multiplications and other matrix
operations by means of standard library procedures (e.g., from LINPACK,
ScaLAPACK, etc.). Therefore, you will not need to program any of
the matrix operations yourself. The simulator is also able to
dump and load its matrices to disk so that they can be carried over
from one simulation run to another.
For more information on this project, please send me mail or make an appointment to discuss it.