Compilation of Haskell array comprehensions for scientific computing
Steven R. Anderson, Paul Hudak
Abstract
Open-access reader
Steven R. Anderson, Paul Hudak
Abstract
Open-access reader
Monolithic approaches to functional language arrays, such as Haskell array comprehensions, define elements all at once, at the time the array is created, instead of incrementally. Although monolithic arrays are elegant, a naive implementation can be very inefficient. For example, if a compiler does not know whether an element has zero or many definitions, it must compile runtime tests. If a compiler does not know inter-element data dependencies, it must resort to pessimistic strategies such as compiling elements as thunks, or making unnecessary copies when updating an array.
OpenAlex reports 22 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Monolithic approaches to functional language arrays, such as Haskell array comprehensions, define elements all at once, at the time the array is created, instead of incrementally. Although monolithic arrays are elegant, a naive implementation can be very inefficient. For example, if a compiler does not know whether an element has zero or many definitions, it must compile runtime tests. If a compiler does not know inter-element data dependencies, it must resort to pessimistic strategies such as compiling elements as thunks, or making unnecessary copies when updating an array.
Key concepts: Haskell, Compiler, Computer science, Programming language, Compile time, Element (criminal law), Scheme (mathematics), Parallel computing