hep_events unit tests failing with ifort 16.0

[Juergen Reuter]

The unit test hep_events is failing,at least with the ifort 16.x compiler (maybe a problem of auto allocation?)

$ diff err-output/hep_events_1.out ref-output/hep_events_1.ref 
181c181
<  Polarization: [spin_type = 2, mult = 2, anti = T]
---
>  Polarization: [spin_type = 2, mult = 2]
191,199c191
<  [h(-1)] => ME(1) = ( 3.333333333333E-01, 0.000000000000E+00)
<  [h(-1|0)] => ME(2) = ( 0.000000000000E+00, 0.000000000000E+00)
<  [h(-1|1)] => ME(3) = ( 0.000000000000E+00, 0.000000000000E+00)
<  [h(0|-1)] => ME(4) = ( 0.000000000000E+00, 0.000000000000E+00)
<  [h(0)] => ME(5) = ( 3.333333333333E-01, 0.000000000000E+00)
<  [h(0|1)] => ME(6) = ( 0.000000000000E+00, 0.000000000000E+00)
<  [h(1|-1)] => ME(7) = ( 0.000000000000E+00, 0.000000000000E+00)
<  [h(1|0)] => ME(8) = ( 0.000000000000E+00, 0.000000000000E+00)
<  [h(1)] => ME(9) = ( 3.333333333333E-01, 0.000000000000E+00)
---
>  [h(1)] => ME(1) = ( 0.000000000000E+00, 0.000000000000E+00)
208,211c200
<  [h(-1)] => ME(1) = ( 5.000000000000E-01, 0.000000000000E+00)
<  [h(-1|1)] => ME(2) = ( 0.000000000000E+00, 0.000000000000E+00)
<  [h(1|-1)] => ME(3) = ( 0.000000000000E+00, 0.000000000000E+00)
<  [h(1)] => ME(4) = ( 5.000000000000E-01, 0.000000000000E+00)
---
>  [h(1)] => ME(1) = ( 0.000000000000E+00, 0.000000000000E+00)
217c206
<  Polarization: [spin_type = 2, mult = 2, anti = T]
---
>  Polarization: [spin_type = 2, mult = 2]
219,222c208
<  [h(-1)] => ME(1) = ( 5.000000000000E-01, 0.000000000000E+00)
<  [h(-1|1)] => ME(2) = ( 0.000000000000E+00, 0.000000000000E+00)
<  [h(1|-1)] => ME(3) = ( 0.000000000000E+00, 0.000000000000E+00)
<  [h(1)] => ME(4) = ( 5.000000000000E-01, 0.000000000000E+00)
---
>  [h(1)] => ME(1) = ( 0.000000000000E+00, 0.000000000000E+00)

[Juergen Reuter]

This test also fails with fortran 4.8.2. and 6.0. Apparently, this test is run only when HepMC is present. WK, could you update this test, please!?

[kilian]

Thanks, that test is skipped on my PC (and apparently on Jenkins?) There might be a minor issue.

Could you attach the err-output (hep_events.out)? I can check manually.

[Juergen Reuter]

Yep, this test is run only when HepMC is on.

[Juergen Reuter]

Error output from hep_events test

[kilian]

Updated test result in r7494.

The previous output reflected the insufficient treatment of polarization factorization. The new version has the polarization states initialized correctly. To fill them correctly, more work is needed, #772.