Skip to content

Julia package to compute the result of expm(t*A)*v when A is a sparse matrix, without computing expm(t*A).

License

Notifications You must be signed in to change notification settings

marcusps/ExpmV.jl

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

42 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ExpmV

Build Status

This is a Julia translation of the MATLAB implementation of Al-Mohy and Higham's function for computing expm(t*A)*v when A is sparse, without explicitly computing expm(A).

The original code can be found at Matlabcentral File Exchange, and the theory is explained in the following article:

Computing the Action of the Matrix Exponential, with an Application to Exponential Integrators, Awad H. Al-Mohy and Nicholas J. Higham, SIAM Journal on Scientific Computing 2011 33:2, 488-511. (preprint)

Installation

Install into Julia using the package manager:

Pkg.clone("[email protected]:marcusps/ExpmV.jl.git", "ExpmV")

Usage

expmv(t,A,v)

Benchmarks

Although both the ExpmV.jl and Expokit.jl implementations are in the early stages of development (ExpmV.jl is a direct translation of MATLAB code, and Expokit.jl is not fully optimized), here are some crude benchmarks (using Benchmark.jl) that indicate large gains over the dense expm. Source can be found in test\benchmark.jl.

Benchmark 1: Complex matrix with density of 0.0191, dimension 100, 100 trials

Row Function Average Relative Replications
1 Expmv.jl 0.0213879 5.24362 100
2 Expokit.jl 0.00407885 1.0 100
3 Julia's dense expm 0.0686303 16.8259 100

Benchmark 2: Complex matrix with density of 0.1913, dimension 100, 100 trials

Row Function Average Relative Replications
1 Expmv.jl 0.00602035 1.74028 100
2 Expokit.jl 0.00345941 1.0 100
3 Julia's dense expm 0.0275519 7.96434 100

Benchmark 3: Complex matrix with density of 0.3651, dimension 100, 100 trials

Row Function Average Relative Replications
1 Expmv.jl 0.0105414 1.97621 100
2 Expokit.jl 0.00533413 1.0 100
3 Julia's dense expm 0.0354484 6.64558 100

Clearly the current ExpmV.jl implementation needs to be looked at more carefully (See Issue #1 in particular, which appears to cause a bottle neck due to my incomplete implementation)

License

Released under the BSD 2-clause license used in Al-Mohy and Higham's original code.

About

Julia package to compute the result of expm(t*A)*v when A is a sparse matrix, without computing expm(t*A).

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 4

  •  
  •  
  •  
  •  

Languages