Dintegration and getIntRoots

[antonleykin]

@michaelPerlman and I looked at #4027 and found a bug in Dintegration.
It is caused by

• either simply a bug (that can be easily fixed) in getIntRoots (and factorBFunction) which happens to assume that all roots of a b-function (of a cyclic module) are rational.
• or something else.

loadPackage "BernsteinSato"
W = QQ[x, dx, a, Da, WeylAlgebra => {{x, dx}, {a, Da}}]
M = cokernel matrix {{-x*a+1, x^2*a^4*Da^2-2*x^2*dx*a^2*Da+2*x^2*a^3*Da+x^2*dx^2+2*x*a^2*Da-2*x*dx+1}}
errorDepth = 0
Dintegration(0,M,{0,1})

The b-function that emerges is $s(s^2-s-1)$.

The help of D-modules experts would be appreciated!
@lorinczandras @christineBerkesch @mahrud

[mikestillman]

I would also like to get the Dmodules packages working with Complexes, rather than OldChainComplexes. I think it won't be too hard. Should I work off the versions in Macaulay2, or are there more recent versions you are using? @lorinczandras @christineBerkesch @mahrud, and would you rather do it yourself, or with me, or have me do it?

[michaelPerlman]

Hi Mike, I would be interested in updating Dmodules to use Complexes with you. I am unsure if all the changes made by @christineBerkesch @mahrud @antonleykin are included in the current version of M2.

[mahrud]

There aren't any unpublished changes involving me that either involve this issue or would conflict with migrating to Complexes.

ps: I'm going to hide the last three messages (including this one) as off-topic in order to keep this issue focused (this has worked well in other open source communities I believe).