odrpack

A modernized Fortran package for weighted orthogonal distance regression (ODR).

Description

odrpack is a package for weighted orthogonal distance regression (ODR), also known as errors-in-variables regression. It is designed primarily for instances when both the explanatory and response variables have significant errors. The package implements a highly efficient algorithm for minimizing the sum of the squares of the weighted orthogonal distances between each data point and the curve described by the model equation, subject to parameter bounds. The nonlinear model can be either explicit or implicit. Additionally, odrpack can be used to solve the ordinary least squares problem where all of the errors are attributed to the observations of the dependent variable.

History

The first version of the library, named ODRPACK, was originally released in 1989 [1], and the last "official" update, named ODRPACK95, dates from 2004 [2]. The full revision history is provided on page iv of the User's Reference Guide for ODRPACK95.

odrpack is a modernization of the ODRPACK95 code [3], intented to make the library easier to use and maintain. The main changes include:

• Conversion from fixed-form (.f) to free-form (.f90).
• Conversion from upper case to lower case.
• Modularization.
• Removal of DATA statements, labeled do loops, and (most) gotos.
• Addition of intent(in/out) to all procedures.
• Addition of explicit interfaces to BLAS routines.
• Implementation of a C API.
• Automatic code documentation with FORD.
• Python bindings, available in the companion repo odrpack-python.

Version	Year	Standard
odrpack	2024	Fortran 2018
ODRPACK95 1.0	2004	Fortran 95
ODRPACK 2.0	1992	FORTRAN 77
ODRPACK 1.0	1989	FORTRAN 77

References

[1] Paul T. Boggs, Janet R. Donaldson, Richaard h. Byrd, and Robert B. Schnabel. 1989. Algorithm 676: ODRPACK: software for weighted orthogonal distance regression. ACM Trans. Math. Softw. 15, 4 (Dec. 1989), 348–364. https://doi.org/10.1145/76909.76913

[2] Jason W. Zwolak, Paul T. Boggs, and Layne T. Watson. 2007. Algorithm 869: ODRPACK95: A weighted orthogonal distance regression code with bound constraints. ACM Trans. Math. Softw. 33, 4 (August 2007), 27–es. https://doi.org/10.1145/1268776.1268782

[3] Original source code from Netlib.

Build instructions

Dependencies

odrpack depends on a small number of functions from BLAS and LINPACK.

• The build configuration files provided with the code (see further below) assume a suitable BLAS library is locally installed. Alternatively, the subset of required BLAS functions is available in external/blas.f; it suffices to move the file to the src folder.

• The subset of required LINPACK functions is available in src/linpack.f and is already included in the build configuration files. No action is required.

With fpm

The easiest way to build/test the code and run the examples is by means of fpm.

To build the library, do:

fpm build --profile release

To run the tests, do:

fpm test --profile release

To run the provided examples, do:

fpm run --example "example_name" --profile release

With meson

First, setup the build:

meson setup builddir -Dbuild_tests=true

To build the libraries, do:

meson compile -C builddir

To run the tests, do:

meson test -C builddir

Licence

• The original ODERPACK95 code is public domain.
• Modifications introduced in this project are covered under the MIT license.