A learning scheme for SPICE simulations diagnostics

A prototype learning scheme for SPICE simulations diagnostics is discussed. The scheme, called ALEA, is run under the MS-DOS operating system and written in C. It consists of two subsystems: AUSPICE, a batch translation of the homonymous expert system, and LEA, which is the actual learning functionality. These two subsystems are meant to cooperate according to different modalities (policies) in different times. A simple numerical example is introduced in order to explain how ALEA can learn from AUSPICE's rules and from experience. For the sake of brevity, only two SPICE parameters are considered: TMAX and TSTOP.<>