OSR precision

[henweb]

Dear All,

I am computing optimal simple rules using OSR routine.

To check the working of the routine, I tried different sized shocks, e.g., multiplying standard deviations of all! shocks by, e.g., 100.

This seems to not change results in some cases, but in some other cases I obtain different coefficients for the optimal rule.

Q: Is increasing standard deviations of shocks a good way to increasing the precision of the optimizer underlying OSR?

Many thanks for your support!

[jpfeifer]

If the precision is too low, directly change it in osr1.m In line 51 of osr1.m you will find:

Code: Select all

crit = 1e-7;

Currently, it is hardcoded and you have to change it manually.

[henweb]

Dear jpfeifer,

thanks for this quick information, this is very helpful!