Name: Vladyslav Shtabovenko Date: 07/19/15-01:39:44 AM Z

Hi,

this should be now fixed. I’ve spent quite some time
on improving the heuristics in SPC, so let us hope that
this kind of problems will not trouble us anymore.

OneLoop[q, FAD[{q, m}, {-2 p + q, m}, {-k - 2 p + q, m}, -p + q] SPD[p,
q] SPD[q, q]] // FCE

I \[Pi]^2 (1/2 B0[SP[k, k], m^2, m^2] -
   1/2 B0[SP[k, k] + 2 SP[k, p] + SP[p, p], 0, m^2] -
   1/2 m^2 C0[SP[k, k], 4 SP[p, p],
     SP[k, k] + 4 SP[k, p] + 4 SP[p, p], m^2, m^2, m^2] +
   1/2 m^4 D0[SP[p, p], 4 SP[p, p],
     SP[k, k] + 4 SP[k, p] + 4 SP[p, p],
     SP[k, k] + 2 SP[k, p] + SP[p, p], SP[p, p], SP[k, k], 0, m^2,
     m^2, m^2] +
   3/2 C0[SP[k, k], SP[p, p], SP[k, k] + 2 SP[k, p] + SP[p, p], m^2,
     m^2, 0] SP[p, p] +
   1/2 m^2 D0[SP[p, p], 4 SP[p, p],
     SP[k, k] + 4 SP[k, p] + 4 SP[p, p],
     SP[k, k] + 2 SP[k, p] + SP[p, p], SP[p, p], SP[k, k], 0, m^2,
     m^2, m^2] SP[p, p])

Cheers,
Vladyslav

Am 08.07.2015 um 11:32 schrieb Vladyslav Shtabovenko:
> Hi,
>
> thanks for reporting this and sorry for the late reply (I’m on a
> conference right now). There is an issue with ScalarProductCancel, where
> sometimes it fails to cancel all the scalar products, that can be
> cancelled. I was trying to fix this in a “non-invasive” way, but it
> seems like to get rid of this problem once and for all, I need to
> rewrite the internal logic of SPC almost from scratch, as it was done
> for TID and Tdec. However, since SPC is used by many other functions I
> will need some time for additional tests.
>
> As a workaround, for now you can use TID (from the latest git snapshot
> with today’s commits). With the option UsePaVeBasis set to True it will
> reduce tensor integrals to PaVe scalar functions just as OneLoop does.
> The purely scalar integrals can be converted to A0, B0, C0 etc. via
> ToPaVe. For example:
>
> TID[test1, q] // ToPaVe[#, q] &
>
> or
>
> TID[test1, q, UsePaVeBasis -> True] // ToPaVe[#, q] &
>
> The first one is the full reduction into the PaVe scalar integrals
> A0,B0,C0,D0. The second one is for the case that you choose your
> kinematics such, that there will be vanishing Gram determinants.
> This is also more close to what OneLoop does.
>
> Cheers,
> Vladyslav
>
> Am 03.07.2015 um 04:08 schrieb Sun Qingfeng:
>> Hi, another pronlem:
>> When I test the FC9.0 patched with FA3.9,
>> Input1:
>> test1=Pair[Momentum[p, D],
>> Momentum[q, D]] (Pair[Momentum[q, D],
>> Momentum[q, D]]) FeynAmpDenominator[
>> PropagatorDenominator[Momentum[q, D], m],
>> PropagatorDenominator[-2 Momentum[p, D] + Momentum[q, D], m],
>> PropagatorDenominator[-Momentum[k, D] - 2 Momentum[p, D] +
>> Momentum[q, D], m],
>> PropagatorDenominator[-Momentum[p, D] + Momentum[q, D], 0]]
>> Input2:
>> OneLoop[q, test1]
>>
>> Output:
>> Something went wrong in the cancelling of scalar products. Evaluation aborted!
>>
>> But when the Denominator is:
>> test2=Pair[Momentum[p, D],
>> Momentum[q, D]] (Pair[Momentum[q, D],
>> Momentum[q, D]]) FeynAmpDenominator[
>> PropagatorDenominator[Momentum[q, D], m],
>> PropagatorDenominator[-Momentum[p, D] + Momentum[q, D], m],
>> PropagatorDenominator[-Momentum[k, D] - 2 Momentum[p, D] +
>> Momentum[q, D], m],
>> PropagatorDenominator[-Momentum[p, D] + Momentum[q, D], 0]]
>> The problem disappear!
>>
>