Examples
// Use with larger r x = accsum_wilkinson(10); for i = 1 : 10 x = accsum_shuffle ( x ); [s1,e1] = accsum_compsum ( x ); [s2,e2] = accsum_dblcompsum ( x ); mprintf("#%5d CS=%e + %e DCS=%e + %e\n",i,s1,e1,s2,e2); end |  |  |
Randomly shuffles the input.
y = accsum_shuffle ( x )
a m-by-n, matrix of doubles
a m-by-n, matrix of doubles, the randomly permuted values.
Randomly shuffles the input. This function makes use of the grand function and modifies its state.
// Use with larger r x = accsum_wilkinson(10); for i = 1 : 10 x = accsum_shuffle ( x ); [s1,e1] = accsum_compsum ( x ); [s2,e2] = accsum_dblcompsum ( x ); mprintf("#%5d CS=%e + %e DCS=%e + %e\n",i,s1,e1,s2,e2); end | | |
"Stability and numerical accuracy of algorithms", Nicolas Higham
"Handbook of Floating Point Computations", Muller et al