UVFIT improvement?

[len067]

Hi All,

I was listening in on one of the radio school presentations where it was mentioned that sources could be modelled using uvfit in miriad. I didn't get a chance to raise this as a concern during the meeting but I don't believe that uvfit is able to fit for spectral index (I could be wrong but the only package I'm aware of that allows this is difmap) - this is a problem for wide-band data particularly at longer wavelengths. Has there been any consideration into adding this as an option? It would be absolutely fantastic if it could fit multi-order terms too ... but even just alpha would be a great improvement compared to what is possible now.

Cheers,

Emil.

[Mark.Wieringa]

Hi Emil,

two answers:

- the current uvfit lets you fit the data in a number of frequency bins, if the number is > 1 it will give an estimate of the spectral index. The residuals or model are calculated by interpolation between bins.
- Dave McConnell is putting the finishing touches on a version of uvfit (which will be called uvsfit) that can fit the spectral index and second and third order terms. It should appear in the next week or two.

Cheers,

Mark

[len067]

Hi Mark,

That's great Mark - thanks for letting me know.

Cheers,

Emil.

[hzhangshao2011]

hi, Emil,
Could you tell me how to download this audio presentation from the internet?
I am learning the radio data process, these audios will be useful for me. Thanks!

Hui

[len067]

Hi Hui,

I'm not actually sure if there was an audio recording of the presentation. Sorry, perhaps you misunderstood - I was in the audience listening to the presentation during the radio school itself.

Cheers,

Emil.

[Mark.Wieringa]

Just announcing that uvsfit (written by Dave McConnell) is now available,

UVSFIT is a Miriad task which fits model components to a visibility
dataset. UVSFIT differs from UVFIT by fitting explicitly for the
frequency dependence of source flux. Optionally the model or
residual visibilities can be written out.

The standard uvfit was enhanced last year with the option of binning in time and frequency,
so there are now two ways to model frequency dependence in wide bands.

Cheers,

Mark

[len067]

Awesome!

Thanks Mark/Dave!

Cheers,

Emil.

[davidmcconnell]

As mentioned by Mark, uvsfit is now distributed (although I will shortly make a minor improvement to the output log format). I would be happy to receive users' comments and suggestions.

Dave