Changes after the release of SecDec version 2.1
 03.02.2015 SecDec2.1.6.1

Analytic continuation of myLog fixed in makeintC.pm for the case of negative and purely real arguments.
 24.11.2014 SecDec2.1.6.1

Bug fixed when calculating fivepoint function in combination with cutconstruct=0.
(For now, the bugfix is only valid for lightlike external legs.)
 12.06.2014 SecDec2.1.6 (SVN rev. 282)

Bug fixed when using togetherflag=1 in the general directory regarding
the collation of the results.

Environment variable CUBACORES included in perlsrc/makeintC.pm to restrict the number
of cores CUBA should use in parallel computations.

Number of Mathematica subkernels to be used in parallel computations is set to 4 by default but can
be altered in the perlsrc/header.pm file.

Bug fixed in general/src/deco/template_tail.m: cases where elements of factorlist given by the user could be factorized further were not treated properly.

Zero numerators can now be handled from the start.
 15.01.2014 SecDec2.1.5 (SVN rev. 270)

introduction of a splitlist in the loop directory in the {mytemplate}.m file where those integrals over the Feynman parameters specified there are split at 1/2

bug fixed which occurred when using togetherflag=1 and a value for grouping that lead to the creation of multiple integration files in one order of the regulator epilon in the together folder, which then lead to the collection of too few result files in the end

various small bugfixes for the userdefined option

Numerical instabilities possibly introduced when using rescaleflag=1 in combination with contourdef=true removed

altered treatment of monomial F[z] functions in combination with contourdef=true

bug fixed which occurred when floating point numbers were used in the onshell conditions

issues when using Mathematica 7 resolved

demo files were updated
 22.10.2013 SecDec2.1.4 (SVN rev. 254)

install script updated to Cuba library version 3.2
 04.10.2013 SecDec2.1.3 (SVN rev. 253)

In the /loop directory:
Iterated sector decomposition is now done in parallel when using Mathematica versions 8 or 9

In the /general directory:
Proper treatment of integration of constants and integrals of just one integration variable included

Update of the Cuba library to version 3.2
 06.08.2013 SecDec2.1.2 (SVN rev. 250)

Bug fixed when using togetherflag=1 + integrators Divonne or Cuhre + the functions to integrate contained only one Feynman parameter

Bug fixed which occurred when using odd dimensions

Comments in the param*.input files added regarding the naming of invariants when computing five or six point amplitudes
 17.06.2013 SecDec2.1.1 (SVN rev. 247)

User input of kinematic invariants in mytemplate.m and myparam.input files may
now additionally contain decimals in the former and fractions in the latter user file.
Scientific numbers or usage of 10^xy are also supported now.

Empty lines in multi*myparam*.input are now ignored which earlier on lead to some wrong pickup
of the kinematic invariants.

Some files added to tarball necessary for CUBA installation
 12.06.2013 (SVN rev. 244)

Small changes in C++ makefile regarding the libraries

CUBA3.0 files updated to the newest version

Bug fixed when using rescale=1 and prefactorflag=0 at the same time
 12.05.2013 (SVN rev. 240)

Bug fixed when using "../launch" in combination with "togetherflag=1" from another working directory than the standard one

CUBA library files reverted to working version
 17.04.2013 (SVN rev. 238)

Dimension which is commented out in mytemplatefile.m is now also recognized as comment

CUBA library files updated to the newest version
 05.03.2013 (SVN rev. 233)

Small bugs fixed when using the together flag for C++ in combination with clusterflag=1

Superfluous semicolons removed in subexp*.pl when using the Mathematica option (which is not yet fully implemented)

Commented in some lines in resultsloop.pl file which were commented out during testing.
 26.02.2013 (SVN rev. 231)

Together flag for C++ option implemented

Files src/subexp/form*.m extended regarding the treatment of orders of epsilon in numerators

Added application of symmetry relations of s_ij to numerator for diagrams with externalegs=4
 07.02.2013 (SVN rev. 230)

Bugs fixed concerning the computation of the numerator. Flexibility of the program increased to tensors of arbitrary rank.

CUBA 3.0 library updated to newest version

Small bug in formContourC.m fixed where indeterminate expressions were found during the subtraction and expansion stage.

Small alterations in the output and resultsfiles.

Renaming of primseclist to functionlist in the template file when inserting user defined functions and omitting the primary sector decomposition.
 05.12.2012 (SVN rev. 224)

Small fixes to get the correct error probability in the result file when using integrators other than Divonne

When using no primary sector decomposition in the loop directory: n1 exchanged by feynpars in form*.m where necessary
 05.12.2012 (SVN rev. 223)

General directory: Update of how the integration is written in makeint.pm file, the error probability is now also written to the output files for each integration in each subdirectory and the results are collected accordingly. The user can now see the maximal error probability of the integration in each subfunction of each order in the regulator epsilon directly in the result files *.res

Removed outdated signcheck in decomposegeneral.pl which produced errors.

Added one missing character in the loop/perlsrc/makeint.pm file writing the Fortran integration files.
 03.12.2012

Bug fixed concerning the correct summation of results and errors, eval statement removed which produced rounding errors.

Decreased size of sobol random sampler in src/util directory. Sobol is compiled once during ./install and addded as .o file to speed up compilation time of C++ contourdef=true functions.

Removed superfluous "common" block line for using no parallelizable Mathematica routine and having chosen language=Fortran in the *.input file.
 21.11.2012

Optimization of C++ code during compilation added.
Removed compiler complaints in Fortran integration files.
Removed computation of nonexistent imaginary part when using C++ and contourdef=False (in myparam.input file)

Mathematica 7 issue removed concerning the distribution of the definition of the symbol "feynpars"

Probability that result is erroneous written to the resultfiles *.res and the {point}results{ord}.log files. Error is printed if not all results were computed yet or one integration failed.
 14.11.2012

Bug fixed concerning the correct computation of the exponent of the functions U and F when using no primary decomposition (n option, and Perl scripts *general.pl in loop directory)

Integration of constants now also possible when using the Fortran option

Readme_general added for the option of using no primary sector decomposition, e.g. for own functions which need sector decomposition and probably also a deformation of the integration contour (thresholds in integration region)
 2.11.2012

Bug fixed for functions with more than 1 Feynman parameter but less than the maximal number of Feynman parameters (Nn1)

Some print statements in output files changed for clarification (misleading "chisq" was changed into "error prob" etc)
 25.10.2012

Togetherflag fixed for Fortran BASES version, minor changes to exemplatory .input files, lengthprimseclist always taken from {template}.m file
 19.10.2012

Integrator chosen properly according to numbers of Feynman parameters contained in integrand functions f*.cc or f*.f (Divonne and Cuhre cannot do 1dim integrals)
 16.10.2012

Ambiguity removed which occurred when a loop momentum was named like an iterator when using the cutconstruct=0 option

Implementation of predecomposition of squared variables for the no primary sector decomposition option

Small fixes to demo files
 15.10.2012

When numerical integration not necessary (no Feynman parameters in the final functions to be integrated numerically), the functions are now computed with the standard C math library
 12.10.2012

Variables "Nloops, Nn, Dim" in the definition of the user prefactor can be used again as they are handed over to the correct prefactor.pl script now
 10.10.2012

Files updated such that cutconstruct=0 and rescaleflag=0 is used as default if no choice is given in the *.input file

Corrections to the makesum.pm file when using language=Fortran

Change in formindlist.m when using the n option
 05.10.2012

Bug fixed concerning integrands where the user chose contour deformation in the *.input file but where no deformation
would be needed (a comment regarding this was then found in the *l*h*.log file)
 04.10.2012

When user computes loop integral with only 1 sector, this is recognized now and the *OUT.info files are written correctly

Problems at the subtraction and expansion stage fixed concerning the form*.m files
 02.10.2012

Removed a bug apprearing during the subtraction and expansion step when computing 1loop diagrams

Documentation updated
Release of SecDec version 2.1
 01.10.2012

In the loop directory, the user can now enter a primseclist in the templatefile and put in their own functions similar to Feynman loop integrals.
An example can be seen in templategeneral.m
Launching ./launch with the n option starts the program with the iterated sector decomposition without doing a primary sector decomposition. As functions, the primseclist inserted into
the 'mytemplatefile' is used.
Changes after the release of SecDec version 2.0
 03.09.2012

Update of CUBA library to version of July 3rd; rescale set to 0 in header.pm by default; output of c++ integration file corrected and more info added
 29.06.2012

For MAC compatibility: changes in makefiles; additional link to gsl library eliminated
 27.06.2012

Small bug fixed when using Fortran to compute diagrams in Euclidean region
 18.06.2012

Result files updated in the general demo directory
 13.06.2012

Small bug in general directory fixed where problems occured at Cuba3.0 compilation stage
 08.06.2012

Compatibility with Mathematica 7 established

Small changes to the integration file template
 23.05.2012

Cuba3.0 library updated to most recent version
 26.04.2012

Small bug fixed for the case that the diagram is only wished to be computed up to a certain pole order and not up to the finite part.
 18.04.2012
 Release of SecDec version 2.0
Changes with respect to version 1.0.0
 01.09.2011

version number changed to 1.1.1.

New features: possibility to use implicit functions; option to produce C++ files instead of fortran files
 02.06.2011

added files permitting to loop over different values for numerical constants
in "general" subdirectory

allowed length of output filenames increased
 11.04.2011

version number changed to 1.1.0 to indicate inclusion of Cuba library as significant upgrade

detection of singularities within the integration region refined
 22.03.2011

in the subdirectory "general", epsilonindependent prefactors were not
treated correctly after the update of 02.03.2011. This is fixed.
 02.03.2011

routine findsymmetries in loop/src/subexp/formindlist.m improved
 justnumerics.pl updated
 24.02.2011

more detailed documentation
 bug fixed in the selection of the preferred decomposition set
 16.02.2011

renamed common/ndim into ndimen because naming conflict occurred with some compilers.
 08.02.2011

Bug fixed relating to prefactors and collection of the results in the loop directory.
Version 1.0.0 gave an incorrect result when "togetherflag=1" and "primarysectors" and
"multiplicities" were nonempty.
Specifically if k sectors are in the list "primarysectors", the
stated result was k times too large.

general/cleanall.sh and loop/cleanall.sh added to run any launchclean scripts
remaining, and to remove all intermediate files from the main directory (e.g. job
output/error files from the batch system)
 04.02.2011

New sector decomposition routines added to allow for parallelisation of the
Mathematica part

Cuba library 2.1 integrated into the package and into the user interface

6.12.2010

issue with "togetherflag=1" on 64bit machines fixed

"graphname" starting with a number is now also possible

info about timings and result files added to the "demos" directories

1.12.2010:

general/finishnumerics.pl: fixed option to call the script from any directory

general/demos/param5F4.input: changed parameter settings to have a faster evaluation
last updated: August 09, 2017