Re: empty spectrum

Hi, Dave... it's along the lines of what I'd like to do.  I already 
have a bunch
of software in place to work with spectra that extracted correctly, so 
I'd like
to incorporate these special cases without much scaffolding.

I suppose one could use max(all_data) + 1 to pick an unused spectrum slot.  I
think it suffices to verify with you folks that assign_rsp(a, r, x) will
function "as expected" for a zero-count spectrum.  In particular, are 
there any
gotchas with eval_counts, get_model_counts, eval_flux, or define_bgd?

I did contact helpdesk about dmextract.  I agree that it should return a zero
histogram if no counts are present.  (At least, it should not hang
indefinitely.)

Thanks,
Rob


Quoting "David P. Huenemoerder" <dph_at_email.domain.hidden
>
>
> Hi Rob -
>
> 1) if dmextract fails, it's probably a bug, and you should report it.
>   There should at least be noise, or even if no noise, you should
>   still be able to make a histogram of value 0.
>
> 2) Sure, you can load an arf and rmf and assign to a non-existent
>   histogram index.  This is how you fake data:
>
>  a = load_arf("my.arf");
>  r = load_rmf("my.rmf");
>  assign_rsp( a, r, 1 ) ; % assuming no data have been loaded into 1.
>
> But now the tricky part starts. What, exactly, should the histogram
> be?  You need to have *something* to fit - such as background noise,
> or a very weak source spectrum.  If you want to know, "How much flux
> could I have and still have zero counts/all bins?", you could define a
> model:
>
>  fit_fun("Powerlaw(1)");
>
> set your params and
>
>  fakeit;
>
> to get a fake counts spectrum.  You could then fit this and the do the
> conf on the normalization.  You could the high value of the conf is
> then (hopefully) something like your upper limit.
>
> However, if you are background limited, you probably need to extract
> background counts (dmextract will work) and then
>
> () = define_back (1, "background.pha");
>
> to associate the background pha w/ histogram 1.
>
> Then you can set your model norm to 0.0, set the upper range to
> something, and do the fit and conf to see what norm value is an upper
> limit.
>
>
> Is this something like what you are trying to do?
>
>
> --Dave



----
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