[Bug-gsl] bug in manual - gsl_odeiv_control_standard_new

[zkoza]

I think there's a bug in the description of the 
algorithm used in the standard dynamic step control objects
(http://www.gnu.org/software/gsl/manual/gsl-ref_25.html#SEC384)

The description does not match the implementation of function 
static int std_control_hadjust
implemented in file cstd.c.

First, its not true that 
"D/E is taken to be the maximum of the ratios D_i/E_i.". 
As a matter of fact the algorithm uses
the ratio E/D computed as the maximum of the ratios E_i/D_i.

Then there's an error in the formula
"h_new = h_old * S * (E/D)^(1/(q+1))"
which should read
h_new = h_old * S * (E/D)^(-1/(q+1))

It might also be helpful to mention that the scaling factor for h is always 
between 1/5 and 5.


Anyway, thank you very, very much for such an exceptionally useful tool as 
GSL is!
 
regards, 
Zbigniew Koza