@phdthesis{10.5555_910997,
    doi = {10.5555/910997},
    author = {Johnson, Steven Dexter},
    title = {Synthesis of digital designs from recursion equations},
    year = {1983},
    publisher = {Indiana University},
    address = {USA},
    abstract = {The discipline of applicative program design style is adapted to the design of synchronous systems. The result is a powerful and natural methodology for engineering correct hardware implementations. This dissertation makes a formal connection between functional recursion and component connectivity that is pleasantly direct, suggesting that applicative notation is the appropriate basis for digital design. Synthesis is a theory for constructing concretely descriptive realizations from abstractly descriptive specifications. Functional recursion equations serve here as specifications; the realization language uses signal equations to describe circuit connectivity. A semantics for realizations assigns an infinite value sequence to each signal. The register equation X = a ! t, where t is an applicative term, defines signal X to be the output a register initialized to a. Synchronous systems are characterized by iterative specifications over a single function symbol (simple while-loops). Moreover, the transcription from F(x(,1),..., x(,n)) p (--->) r, F(t(,1),..., t(,n)) to the canonical realization {X(,i) = a(,i) ! t(,i)} is immediate and correct. Realizations can be compiled from arbitrary iterative specifications by a familiar construction. In non-iterative cases synthesis of iterative form is the main tactic used here for deriving realizations. This subgoal formalizes the conventional technique of decomposing a circuit into an architecture and a finite state controller.This approach suggests yet another way to implement "function recursion" with "data recursion." An interpreter has been implemented for Daisy, a demand-driven list processing language. By representing signals as infinite lists, expressed realizations can be directly interpreted, resulting in a simulation of logical behavior. Thus, the engineering notation serves also as a vehicle for experimentation.A non-trivial exercise in language-driven design reveals that global design factorization techniques, such as hierarchical decomposition and data abstraction, are inherited from the functional description style. Next, a specialized transformation system is defined to address a typical local refinement problem: serialization of inputs.},
    note = {AAI8321375},
    date-added = {2025-4-18 18:29:4 +0100}
}

@phdthesis{10.5555_910997, doi = {10.5555/910997}, author = {Johnson, Steven Dexter}, title = {Synthesis of digital designs from recursion equations}, year = {1983}, publisher = {Indiana University}, address = {USA}, abstract = {The discipline of applicative program design style is adapted to the design of synchronous systems. The result is a powerful and natural methodology for engineering correct hardware implementations. This dissertation makes a formal connection between functional recursion and component connectivity that is pleasantly direct, suggesting that applicative notation is the appropriate basis for digital design. Synthesis is a theory for constructing concretely descriptive realizations from abstractly descriptive specifications. Functional recursion equations serve here as specifications; the realization language uses signal equations to describe circuit connectivity. A semantics for realizations assigns an infinite value sequence to each signal. The register equation X = a ! t, where t is an applicative term, defines signal X to be the output a register initialized to a. Synchronous systems are characterized by iterative specifications over a single function symbol (simple while-loops). Moreover, the transcription from F(x(,1),..., x(,n)) p (--->) r, F(t(,1),..., t(,n)) to the canonical realization {X(,i) = a(,i) ! t(,i)} is immediate and correct. Realizations can be compiled from arbitrary iterative specifications by a familiar construction. In non-iterative cases synthesis of iterative form is the main tactic used here for deriving realizations. This subgoal formalizes the conventional technique of decomposing a circuit into an architecture and a finite state controller.This approach suggests yet another way to implement "function recursion" with "data recursion." An interpreter has been implemented for Daisy, a demand-driven list processing language. By representing signals as infinite lists, expressed realizations can be directly interpreted, resulting in a simulation of logical behavior. Thus, the engineering notation serves also as a vehicle for experimentation.A non-trivial exercise in language-driven design reveals that global design factorization techniques, such as hierarchical decomposition and data abstraction, are inherited from the functional description style. Next, a specialized transformation system is defined to address a typical local refinement problem: serialization of inputs.}, note = {AAI8321375}, date-added = {2025-4-18 18:29:4 +0100} }