The LSA has been developed with three distinct groups of users in mind:
Numerical linear algebra researchers. These users are the experts in
the areas covered by the LSA, and would use it for rapid prototyping of
solution strategies.
Application scientists and engineers who encounter difficult linear
systems in their work. These are technically trained experts in
other fields, who need the rapid prototyping capabilities and also
need some high level guidance on navigating among the possible
solution methods. They also want usable source code that incorporates
successful solution strategies for their class of problems.
Students in computational science. This group needs much more guidance
from the LSA information subsystem, and needs to have good history
and scripting mechanisms to "replay" and revisit sessions.
Although initially the LSA is most usable by the first group, from
the beginning it has been designed and built with mechanisms
to accomodate all of
the users: multilevel, extensible information subsystem, code extraction and
encapsulation, visual interfaces which aid memory of past experiments, and
component wrappers which make it easy for users to add their own
computational components.