Static Analysis by Abstract Interpretation of the Functional Correctness of Matrix Manipulating Programs

Abstract : We present new abstract domains to prove automatically the functional correctness of algorithms implementing matrix operations, such as matrix addition, multiplication, GEMM (general matrix multiplication), or more generally BLAS (Basic Linear Algebra Subprograms). In order to do so, we introduce a family of abstract domains parameter-ized by a set of matrix predicates and by a numeric domain. We show that our analysis is robust enough to prove the functional correctness of several versions of matrix addition and multiplication codes resulting from loop reordering, loop tiling, inverting the iteration order, line swapping, and expression decomposition. Finally, we extend our method to enable modular analysis on code fragments manipulating matrices by reference, and show that it results in a significant analysis speedup.
Document type :
Conference papers
Complete list of metadatas

Cited literature [18 references]  Display  Hide  Download

https://hal.sorbonne-universite.fr/hal-01360556
Contributor : Antoine Miné <>
Submitted on : Tuesday, September 6, 2016 - 10:05:25 AM
Last modification on : Thursday, March 21, 2019 - 1:04:43 PM
Long-term archiving on : Wednesday, December 7, 2016 - 12:36:51 PM

File

article-journault-al-sas16.pdf
Files produced by the author(s)

Identifiers

Citation

Matthieu Journault, Antoine Miné. Static Analysis by Abstract Interpretation of the Functional Correctness of Matrix Manipulating Programs. 23rd Static Analysis Symposium (SAS), Sep 2016, Edimbourg, United Kingdom. pp.257-277, ⟨10.1007/978-3-662-53413-7_13⟩. ⟨hal-01360556⟩

Share

Metrics

Record views

533

Files downloads

250