04-09-16 (Build 64 /Subversion repository revision 21). v2.2.1. A bug in the code used for word reduction when a coset table is present has been fixed. Previously the code removed the coset symbol if present, which caused the isnormal utility to give incorrect results. While this is a minor fix the version number is 2.2.1 rather than 2.1.2 because this is the first update in five years.
14-06-11 (Build 63 /Subversion repository revision 20). v2.1.1. A bug in the code used for FSA minimisation has been fixed. The bug caused composite multipliers for coset systems to be generated incorrectly under some circumstances and was introduced in build 46. The most likely effect of this was that axiom checking would erroneously fail, or that automata, gpsubpres, or gpsublowindex could fail while generating a subgroup presentation. A minor change has been made to gpmult. The operation of the -sub option has been slightly changed. Previously inverse words were added for each specified subgroup generator. This is no longer done.
11-06-11 (Build 62 /Subversion repository revision 18). v2.1.0 . Two new utilities are added. The more important one gpxlatwa, allows a word-acceptor using one set of generators to be translated into a word-acceptor for another generating set. There is also a bug fix to the parsing of expressions in input and substructure files. Previously a closing parenthesis without a matching opening parenthesis was treated as the end of the expression instead of generating an input error.
02-06-11 (Build 61 /Subversion repository revision 15). v2.0.5 . Some serious bugs have been fixed in the code which generates subgroup presentations. These bugs affected the code which is used when the presentation is generated from an automatic coset structure. There are two versions of this code, a "parallel" version and a "serial" version. The latter code, which is used if it is estimated that building a general multiplier for all the relators will be difficult did not work properly.
20-05-11 (Build 60 /Subversion repository revision 14). v2.0.4 . A few more corrections to the documentation have been made, and a bug in gpsubslowindex has been fixed.
17-07-10 (Build 58 /Subversion repository revision 13). v2.0.3 . A few corrections to the documentation have been made.
17-07-10 (Build 58 /Subversion repository revision 13). v2.0.2 . gpgeowa was broken in v2.
08-07-10 (Build 57 /Subversion repository revision 12). v2.0.1 . Fixes several bugs from v2.0. The example input files have been fully revised and have been used to perform extensive testing of this release. The documentation has also had many minor corrections.
18-06-10 (Build 54 /Subversion repository revision 7). V2.0 (Beta). Many new utilities, including coset enumerator and low index subgroup program. 64 bit version has reduced memory requirements. Documentation has been rewritten and now includes several tutorials.
11-11-09 (Build 46). Performance of FSA minimisation has been substantially improved, so that construction of large automatic structures is now much faster. There is a new utility gpcclass which creates an FSA that solves the conjugacy problem for a finite group. There are some new command line options for automata, mostly connected with using coset systems with named subgenerators to find subgroups of large but finite index.
14-10-09 (Build 45). Numerous minor changes to improve compatibility of various utilities with KBMAG. automata will now produce a file with a .akbprog suffix when it has found an automatic structure. This contains the set of equations used to establish the correct word-difference machines. The latest MacPro version has been built on Snow Leopard. The author does not know if it works on earlier versions of OS X.
19-09-09. All source files now have a copyright notice and a reference to the GPL.
16-09-09. Further minor changes to ensure clean compilation with most recent g++. Also a minor change to mafauto.cpp to improve handling of difficult to build word acceptors.
14-09-09. Code now compiles cleanly with several compilers, and should now be fully 64-bit portable. There are some minor fixes to the routines that build automatic structures for orderings other than shortlex. Some new options added to automata and kbprog.
2-09-09. Two fixes for normal subgroup coset systems: the subgroup word acceptor was not calculated correctly, and if the -nokb option was used MAF failed to ensure the subgroup was actually normal. The comments in the source code have been updated.
25-08-09. Various minor fixes, and some new utilities. MAF now works better on presentations with a large number of generators - in particular memory usage should be much lower than before in this case.
12-11-08. Various fixes, all relating to coset systems. Sometimes the multiplier for an automatic coset system was incorrect when a group generator was both eliminated and in the subgroup.