Closed benruijl closed 7 years ago
I get
F =
+ x2*rat(-1,1)
+ x1*rat(-1,1)
;
with 320a477 , 5634f18 , 6b38064 .
cf71752 is bad, though, giving
F =
+ x2*rat( - 3*ep - 2,ep)
+ x1*rat( - 3*ep - 2,ep)
;
Thanks Josh! I will git bisect it soon to find the bad commit.
As you can imagine, this bug was introduced by a0b635c.
BAD * c8cdc42 [2017-04-19] [test] Test case for #128 @Takahiro Ueda
BAD * 7976927 [2017-04-19] The permute in transform did not handle local dollars properly in tform. @vermaseren
BAD * 41642ab [2017-04-16] Added the use of $-variables to the permute option in the transform statement. @vermaseren
BAD * 929268a [2017-04-14] [test] An example of perm_, given in 31d4b6b @Takahiro Ueda
BAD * 4e28f92 [2017-04-14] Repaired a little mistake in the manual. @vermaseren
BAD * 31d4b6b [2017-04-14] Added function perm_ to the manual @vermaseren
BAD * 7ded5c9 [2017-04-14] Added the permutations function perm_ @vermaseren
BAD * a995a3c [2017-04-09] Man sorting into a chapter @vermaseren
BAD * 81d838a [2017-04-08] [build][doc] Include spectators.tex @Takahiro Ueda
BAD * af374eb [2017-04-08] Forgot to add spectators.tex @vermaseren
BAD * ff98159 [2017-04-07] Added spectators to the manual @vermaseren
BAD * 1470f0b [2017-04-06] Added trailing - option to polyratfun. Dangerous! @vermaseren
BAD * 72fc216 [2017-04-05] [test] Put a test case from #178 @Takahiro Ueda
BAD * e24fd09 [2017-04-05] Repaired fix.frm @vermaseren
BAD * f74b7b5 [2017-04-05] Repaired a crash reported by Josh Davies @vermaseren
BAD * 5b55fdb [2017-04-04] Changed CLEANPRF into MUSTCLEANPRF @vermaseren
BAD * 6c957e4 [2017-04-04] optimized poly_ratfun_normalize Merge branch 'master' of github.com:vermaseren/form @vermaseren
|\
BAD | * a675751 [2017-04-03] [test] Test case for #180 @Takahiro Ueda
BAD * | c41e6af [2017-04-04] Optimized poly_ratfun_normalize: only dirty functions are normalized. @vermaseren
|/
BAD * 2fb697d [2017-04-03] Fixed up RAT and a few imperfections of poly_ratfun_normalize @vermaseren
BAD * 74cfccd [2017-04-03] Fix an invalid memory access for ParFORM @Takahiro Ueda
BAD * a7392f5 [2017-04-03] Reapply the patch of dd92c10 @Takahiro Ueda
BAD * cda04e9 [2017-04-03] [build] Bug reports should be sent to the Issue Tracker @Takahiro Ueda
BAD * 5fc9a7e [2017-04-03] fixed up the previous fixes and some asserts in the ruby file @vermaseren
BAD * 86a5228 [2017-04-02] Fixed up the expand part of PolyRatFun that went wrong with the polywrap.cc fix @vermaseren
BAD * edd914d [2017-04-02] Repaired inefficiency in PolyRatFun. Added debug code. Etc. Merge branch 'master' of github.com:vermaseren/form @vermaseren
|\
GOOD | * 5634f18 [2017-03-31] Fix for activeexprnames_ with replaced expressions @Takahiro Ueda
BAD * | a0b635c [2017-04-02] Repaired major inefficiency with PolyRatFun. Added lots of debug code. Replaced many Malloc1 calls. Reinstated MaxNumberSize setup option. @vermaseren
|/
GOOD * a4fd5e1 [2017-03-29] Add support for absolute paths in #Append/PrependPath @Takahiro Ueda
GOOD * 29c1794 [2017-03-27] Implement #Append/PrependPath @Takahiro Ueda
GOOD * 1d7338d [2017-03-17] [doc] Put Ben in the author list of the manual @Takahiro Ueda
GOOD * 7c56626 [2017-03-17] Merge branch 'exprvars' @Takahiro Ueda
The following program shows that moving a polyratfun out of a function may produce a wrong result.
I suspect there is a problem with the ordering of the monomials in the polyratfun which changes when it leaves the function and goes to ground level, which causes the add routine to produce a wrong result.