Module ao_mpi

This file implements ao_mpi, a function which returns a parallel version of ao, the function which computes the model tendencies.

The function returned by ao_mpi is a rank-specific function which distributes the calculation of the model tendencies over multiple cores.

It uses the lua-mpi module by Peter Colberg: https://colberg.org/lua-mpi/

Functions

ao (t, y, buf) Rank-specific function that calculates the time derivative of the n variables.

Local Functions

ao_mpi (rank, size, comm) Get a rank-specific function which calculates part of the tensor contraction and reduces the result to all cores.


Functions

ao (t, y, buf)
Rank-specific function that calculates the time derivative of the n variables. The function reduces to a sparse tensor contraction due to the bilinear nature of the equations.

Parameters:

  • t time
  • y array with variables at time t
  • buf n-array (buffer) to store derivatives.

Local Functions

ao_mpi (rank, size, comm)
Get a rank-specific function which calculates part of the tensor contraction and reduces the result to all cores.

Parameters:

  • rank
  • size
  • comm partition the tensor into subtensors for each core.
generated by LDoc 1.4.3 Last updated 2016-03-11 16:19:27