There was considerable discussion over these points. Both operational/economical considerations should also be considered. Overall, the array design must be suitable for the science it is to do.

Image assessment can be most effectively measured by employing a standard image or visibility dataset of some sort. At the discussion, it was thought that a standard visibility dataset made much sense, perhaps with pointing errors as built-in corruptors. Autocorrelation data can be a separate dataset. A favored way to do this was to have a captive account, probably on an NRAO machine, running AIPS++ with user-defined station coordinates, on the standard dataset. Whether this is consonant with the NRAO computing policy will be examined.

Bryan made a list of known simulation programs, and their capacities, which I will eventually place here, along with updates.

Compact configurations are an integral element of most array styles. Accordingly, the siting of the compact configuration will depend upon the style chosen. Inasmuch as the topography limits siting, it will impact style. On the other hand, the detailed design of the most compact array depends less upon style than the extended arrays, and its design may be split off.

Radford has posted new topographic images in various formats at the ALMA maps site. Included are coordinates of known things, with varying degrees of accuracy. Butler has produced a new mask set. Iteration will ensure that the criteria he used are acceptable. Consultation with VLA civil engineering experts will be sought.