Off-line data analysis requirements

We have to perform a survey of off-line analysis requirements for ALMA.

This covers science data processing and includes what should be under the responsibility of the PI:

what we have called `project calibration': but not `array calibration' which are under responsibility of the observatory: their results should be provided to the PI.
imaging: image production, deconvolution, ...
do we include more image analysis and interpretation tools ?

we should list and describe the rules for selecting in the available off-line software items that will be eligible to be included in the calibration and science data pipelines.
There is of course a strong interest into using available off-line software as the `production engines' of the pipeline, allowing the pipeline development effort to concentrate into the software needed to move the data in and out of those engines, as well as to manage quality information.
Is this realistic ?

list some sample problems along with the breakpoint at which the problems are assumed to be runnable in a "normal" machine. Some remarks about parallelization, memory and I/O use might be in order.

make sure that we include off-line analysis requirements for all modes that we have listed in memo 293 !
identify the algorithms that must be made available for each of these modes, describe the input and output, the range of computing resources for a typical project, ...
In particular we should think about the tools that will enable the user to evaluate the quality of the data: phase noise, dynamic range, ...
Include priorities; some features will be needed from the start.