Sophia Borowka, Gudrun Heinrich, Stephan Jahn, Stephen Jones, Matthias Kerner, Johannes Schlenk, Tom Zirke
A program to evaluate dimensionally regulated parameter integrals numerically
home download program user manual faq changelog
The first release version of pySecDec can be downloaded as pySecDec-1.0.tar.gz. The manual is available here.
See also the corresponding paper arXiv:1703.09692.
Version 3.0 of the program can be downloaded as SecDec-3.0.9.tar.gz. The manual for this version is available here.
A beta version of the program where the algebraic part is implemented in python can be downloaded as pySecDec-0.9.tar.gz.
The manual for pySecDec-0.9 is available online in html format and downloadable as pySecDec-0.9.pdf.
The previously uploaded alpha version is still available as pySecDec-0.1.1.tar.gz. Its documentation can be downloaded as pySecDec-0.1.1.pdf.
To install the program SecDec-3.0.9:
- tar xzvf SecDec-3.0.9.tar.gz
- cd SecDec-3.0.9
The program Normaliz needed by the geometric decomposition strategies can be obtained from Normaliz 2.10.1
Documentation and examples are contained in the program and are described e.g. in Comput.Phys.Commun. 196 (2015) 470-491 [arXiv:1502.06595]
and in Comput.Phys.Commun. 184 396-408 (2013), Comput.Phys.Commun. 184 2552-2561 (2013) and in Sophia Borowka's PhD thesis
Please send bug reports, comments, etc to email@example.com
New features in version 3.0 are:
- Inclusion of two new sector decomposition strategies avoiding an infinite recursion
- New user interface
- Improvements in speed, e.g. inclusion of fast 1-dimensional integrator
- Possibility to specify inverse and linear propagators, divergent prefactors and dummy functions depending on the dimensional regulator
- Choice of numerical integrators extended to Mathematica's NIntegrate
- Omitting the primary sector decomposition is now possible. A user can now compute Feynman parameter integrals which need no primary sector decomposition for arbitrary kinematics.
- User friendliness improved by printing the maximal error probability to the result files.
- Computation of integrals of in principle arbitrary rank possible
- The C++ compile times are reduced.
Multi-scale loop integrals can be evaluated without restriction to the Euclidean region
Warning: For kinematic points near a pinch singularity, the result can depend critically on the settings for the numerical integration. In this case the user should watch carefully the convergence of the Monte Carlo integration. In case of very bad convergence the error given by the Monte Carlo program may underestimate considerably the true error.
- Parallelisation of the algebraic part in Mathematica possible if several cores are available
- Newest version of the integration routines from the Cuba library are included
- Possibility to loop over ranges of parameter values is automated
- Possibility to loop over sets of numerical constants
- Inclusion of implicit functions possible in the algebraic part for general functions
- Option to create C++ instead of fortran functions
- Integration routines from the Cuba library are included
- Parallelisation of the decomposition in Mathematica possible if several cores are available
- Correction of a bug: in the combination "togetherflag=1" and "primarysectors" and "multiplicities" non-empty, the prefactor was multiplied several times
An older version of the program (v 1.1.1) can be downloaded as SecDec-1.1.1.tar.gz
For further reading:
- documentation and examples of SecDec version 1 can be found in Comput.Phys.Commun. 182: 1566 (2011)
[arXiv: 1011.5493 hep-ph]
- review article on sector decomposition
- original article on multi-loop integrals Nucl.Phys.B585:741 (2000)
last updated: Mar 29, 2017