https://github.com/tuwien-cms/xprec

Full quadruple precision (double-double) data type for numpy
https://github.com/tuwien-cms/xprec

extended-precision numpy

Last synced: 7 days ago
JSON representation

Full quadruple precision (double-double) data type for numpy

• Host: GitHub
• URL: https://github.com/tuwien-cms/xprec
• Owner: tuwien-cms
• License: mit
• Created: 2021-08-23T14:53:50.000Z (about 3 years ago)
• Default Branch: mainline
• Last Pushed: 2024-06-24T14:39:39.000Z (5 months ago)
• Last Synced: 2024-10-31T19:37:04.901Z (14 days ago)
• Topics: extended-precision, numpy
• Language: C
• Homepage:
• Size: 270 KB
• Stars: 12
• Watchers: 5
• Forks: 4
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.txt

Awesome Lists containing this project

README

        
Library for double-double arithmetic calculation

================================================

Extension module for numpy providing the `ddouble` data type.

Loading this module registers an additional scalar data type `ddouble` with

numpy implementing double-double arithmetic.  You can use use the data type

by passing `dtype=xprec.ddouble` to numpy functions.

The `xprec.linalg` module provides some linear algebra subroutines, in

particular QR, RRQR, SVD and truncated SVD.

Installation

------------

    $ pip install xprec

Quickstart

----------

    import numpy as np

    x = np.linspace(0, np.pi)

    # import double-double precision data type

    from xprec import ddouble

    x = x.astype(ddouble)

    y = x * x + 1

    z = np.sin(x)

    # do some linalg

    import xprec.linalg

    A = np.vander(np.linspace(-1, 1, 80, dtype=ddouble), 150)

    U, s, VT = xprec.linalg.svd(A)

Trouble shooting

---

* icc


You may suffer from a long runtime when xprec is built with icc. If you encounter this problem, please try the following:

```

CFLAGS="-fp-model=precise" pip install xprec

```

Licence

-------

The xprec library is

Copyright (C) 2021 Markus Wallerberger.

Licensed under the MIT license (see LICENSE.txt).

Contains code from the QD library, which is

Copyright (C) 2012 Yozo Hida, Xiaoye S. Li, David H. Bailey.

Released under a modified BSD license (see QD-LICENSE.txt).