Gaussian elimination

load script permanent into redis

>> r = r.loadGaussian('go-redis/mex/gaussian.lua');

general workflow in GNU Octave/Matlab

To solve A*x = b, let's create A and b.

>> r = redis; % initialize
>> a = rand(10,10); % create a 10x10 random matrix
>> b = (1:10)'; % create b
>> r.array2redis(a); % save a in redis
>> r.array2redis(b); % save b in redis

calculate gaussian elimination

... the calculation is made by lua inside of redis.

>> tic, x = r.gaussian('a','b'); toc  
Elapsed time is 0.019226 seconds.  

compare the result

Result calculated by GNU Octave/Matlab

>> a\b

ans =

    2.9227
   11.7630
  -20.2694
   16.8191
    3.1782
    2.1669
    7.7981
   -5.2681
  -19.7751
   22.5863

Result calculated by Lua in Redis

>> x

x =

    2.9227
   11.7630
  -20.2694
   16.8191
    3.1782
    2.1669
    7.7981
   -5.2681
  -19.7751
   22.5863

...if you get rounding problems, increase r.precision before storing the arrays.

why?

Ever run out of memory in GNU Octave or Matlab? Well,

• simple pump your huge matrices into a redis instance on a larger server.
• pray for fast bandwidth
• worry about visualization or varification methodes of those data