The recurring theme in this working group was high sensitivity at high frequencies over a wide range of angular scales. We need the quoted sensitivity, for at least 1 to 2 GHz bandwidths (and preferably for a 10 GHz band!), between 15 and 50 GHz; and we want this for weak (10-microJy) sources, even in the A+ configuration. We must be able to phase-reference at this level, we must have good surface accuracy and pointing, and we must be able to handle the consequent high data rates. E configuration is also very helpful, for surface brightness sensitivity at these high frequencies. Continuous frequency coverage is not necessary, but wide bandwidths are.
The most sensitive possible system at 1 to 2 GHz is also needed for spectral imaging, both of diffuse and of unresolved sources, whether this system is at 1.4 or 2.4 GHz. We need the highest possible sensitivity, perhaps through wide bandwidths with (on-line?) RFI excision; this also implies high dynamic range. 74 MHz receivers on all VLA antennas would be specially useful for studies of radio halos and the of most extended outskirts of galaxy disks. Continuous frequency coverage is not necessary but we must be able to image the entire primary beam.
We need polarimetry of weak sources with 1-2 GHz bandwidths, to
the same accuracy we achieve now (
%). Again, we would like to
image the entire primary beam.
To take full advantage of the upgrade, we would like the
correlator to permit imaging the whole primary beam in the
A configuration or beyond between 
1.3cm and 90cm,
without bandwidth or time-delay smearing. For the A+ configuration,
this means about 2000 channels, 0.3 sec averaging times, and
33 antennas, for a (sustained!) rate of about 
visibilities per second (full polarization mode). This might be
halved by using longer averaging times and fewer channels on the short
baselines, but not much further than that. Without the A+
configuration, the numbers are more reasonable- 400 channels,
1 second averages, and 27 antennas, for a sustainable rate of
visibilities per second (full polarization).
The E configuration is needed mainly to image diffuse low-level emission at high frequencies. Snapshots are less important than long, sensitive observations. Total power measurements from somewhere must be available to allow imaging at high frequencies; mosaicing must also work.
Frequency agility is essential for the study of time-variable
sources, where simultaneous high-frequency (
GHz) and
lower-frequency observations would be very powerful.
Overall, there are three top-priority items, in the order (1) new
correlator, (2) microJy sensitivities at some high frequency
(
GHz), (3) microJy sensitivities at some low frequency
(1-2 GHz, depending on RFI). The new configurations are interesting
but not as crucial. Whether the E configuration or the A+
configuration is more important depends on the kind of science one is
interested in.