matrix-market-attoparsec: Parsing and serialization functions for the NIST Matrix Market format

[ bsd2, library, parsers ] [ Propose Tags ]

Parsing and serialization functions for the NIST Matrix Market format.

[Skip to Readme]
Versions [faq],,,,,,,,,,,,,
Change log CHANGELOG.markdown
Dependencies attoparsec (>=0.10), base (>=4.7 && <5), bytestring (>=0.9), exceptions, scientific (>= [details]
License BSD-2-Clause
Copyright (c) 2017-2020 Marco Zocca
Author Marco Zocca
Maintainer zocca marco gmail
Category Parsers
Home page
Source repo head: git clone
Uploaded by ocramz at 2020-01-19T10:11:24Z
Distributions LTSHaskell:, NixOS:, Stackage:
Downloads 6066 total (33 in the last 30 days)
Rating 2.0 (votes: 1) [estimated by Bayesian average]
Your Rating
  • λ
  • λ
  • λ
Status Hackage Matrix CI
Docs available [build log]
Last success reported on 2020-01-19 [all 1 reports]


[Index] [Quick Jump]


Maintainer's Corner

For package maintainers and hackage trustees

Readme for matrix-market-attoparsec-

[back to package description]


Build Status

Attoparsec parser for the NIST Matrix Market format [0].

The library also contains functions for serializing matrix data to text file.

User guide

The module Data.Matrix.MatrixMarket exports the user interface:

readMatrix :: FilePath -> IO (Matrix S.Scientific)

readArray :: FilePath -> IO (Array S.Scientific)

writeMatrix :: Show a => FilePath -> Matrix a -> IO ()

writeArray :: Show a => FilePath -> Array a -> IO ()  

The first two functions contain the parsing logic, and make use of scientific for parsing numerical data in scientific notation.

As of version 0.1.1 there are also intermediate functions readMatrix', readArray', writeMatrix' and writeArray' that do not touch the filesystem but (de-)serialize from/to lazy ByteString.

Naming convention

We follow the MatrixMarket format definitions, by which a "Matrix" is sparse and stored in coordinate format (row, column, entry), whereas an "Array" is a dense grid of numbers, stored in column-oriented form. Algebraic vectors, such as the right-hand sides of equation systems, are stored as n-by-1 Arrays.


test/LibSpec.hs contains a simple read/write/read sanity test for the included Matrix Marked data files (fidapm05.mtx and fidapm05_rhs1.mtx):

m0 <- readMatrix fname   -- load original
writeMatrix ftemp m0     -- save as temp
m1 <- readMatrix ftemp   -- load temp
m0 `shouldBe` m1         -- compare temp with original