1

i am working in a small project where i code a distributed matrix solver where each matrix dots is represented by a function itself and the final computation of each threaded solver is queued to the next matrix solver function calculator similar to s-boxes but from a neat python software implementation.

  • Help me out what mathematics research fields i am into
  • How would those interconnected matrixed would be finally represented in an algebraic form?
  • any ideas how to proced implementing both the solver and threaded?

the project aimst to apply small fast solvers to FFT F(q) byproducts so i can analyze several noise signals among frequency peaks.

ogeid
  • 111
  • 2

1 Answers1

0

Some 30 years ago, the research area of Systolic Arrays gained a lot of attention: Rather simple processors in a regular mesh, cleverly connected to compute FFT, matrix multiplication and other numerical algorithms. The ideas might still be inspiring for you.

Another area to look for is CUDA FFT. CUDA is a programming model for PC graphics boards with massive parallelism of graphical CPUs (aka "GPU").

The basic Cooley–Tukey FFT algorithm is described in Wikipedia.

Community
  • 1
Axel Kemper
  • 4,943