Re: help finding define_counts()

On Fri, Mar 21, 2008 at 04:14 +0100, Manfred Hanke wrote:
> 
> continuing the suggestions of Mike, one could also add some 
> documentation on the variable Minimum_Stat_Err. When it's not an 
> counts-Spectrum one wants to fit, one has often to set it to a value 
> much lower than 1, perhaps even DOUBLE_MIN.
>

When a dataset is being treated as "counts", isis attempts to
validate the input uncertainty values, requiring them all to be
greater than or equal to Minimum_Stat_Err, where
Minimum_Stat_Err has a default value of 1.

For example, defining a single-bin "counts" dataset with has
bin value 1.e-5 and uncertainty 1.e-8:

isis> define_counts (1.0, 2.0, 1.e-5, 1.e-8);
1
isis> print(get_data_counts(1).err);         
1

Note that isis "corrected" the 1.e-8 uncertainty value because
it was inconsistent with default expectations for this kind of
data.

To force isis to retain positive counts-uncertainty values that
"look wrong", define a minimum (positive) allowable uncertainty
value using Minimum_Stat_Err:

isis> Minimum_Stat_Err=1.e-20; 
isis> define_counts (1.0, 2.0, 1.e-5, 1.e-8);
2
isis> print(get_data_counts(2).err);         
1e-08

Here, isis kept the 1.e-8 uncertainty value.

When data is declared to have "flux" values, this kind of
filtering does not take place:

isis> define_flux (1, 2, 1.e-5, 1.e-8);
1
isis> print(get_data_flux(1).err);     
1e-08

Uncertainties for flux values are required to be positive, but
that's the only constraint.  

I'll include this explanation in the manual.

Thanks,
-John
----
You received this message because you are
subscribed to the isis-users list.
To unsubscribe, send a message to
isis-users-request_at_email.domain.hiddenwith the first line of the message as:
unsubscribe