[Gretl-users] Re: Example code of 'eigensym' and way faster 'eiggen2'
On Fri, 10 Jan 2020, Allin Cottrell wrote:
On Fri, 10 Jan 2020, Allin Cottrell wrote:
[...]
>
> Script:
>
> <hansl>
> set verbose off
> set seed 8402212 # or whatever
>
> tiny_mat = 0 # or 0 for bigger input!
Oops. As you might have guessed, that line should have read:
tiny_mat = 1 # or 0 for bigger input!
"tiny_mat" (4 x 4) was supposed to be the default, as in Ioannis's original
script. But the timings I posted were correct, for my system.
Home laptop:
tiny = 0:
eigen() time: 2.6018s
eigensym() time: 0.8643s
ratio of eigen() time to eigensym() time: 3.01019
tiny = 1:
eigen() time: 0.4468s
eigensym() time: 0.2054s
ratio of eigen() time to eigensym() time: 2.17544
Work PC:
tiny = 0:
eigen() time: 0.6631s
eigensym() time: 0.5610s
ratio of eigen() time to eigensym() time: 1.18205
tiny = 1:
eigen() time: 0.1055s
eigensym() time: 0.0857s
ratio of eigen() time to eigensym() time: 1.23118
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------