libcerf is a self-contained numeric library that provides an efficient and accurate implementation of complex error functions, along with Dawson, Faddeeva, and Voigt functions.


By construction, it is expected that the relative accuracy is generally better than 1E-13. This has been confirmed by comparison with high-precision Maple computations and with a long double computation using Fourier transform representation and double-exponential transform.

Copyright and Citation

Copyright (C) Steven G. Johnson, Massachusetts Institute of Technology, 2012; Joachim Wuttke, Forschungszentrum J├╝lich, 2013.

License: MIT License

When using libcerf in scientific work, please cite as follows: * S. G. Johnson, A. Cervellino, J. Wuttke: libcerf, numeric library for complex error functions, version [...],

Please send bug reports to the authors, or submit them through the Gitlab issue tracker.