Possible bug for the spin 3/2 spin projection in Calc_2pt.cpp

[pittlerf]

As I was investigating the smearing parameters that we used for the omega baryon mass calculation I realized a possible bug in Calc_2pt in the spin-projection for spin 3/2 baryon.

When I compared the dataset produced with PLEGMA_LIGHT_BARYONS=ON

/sx04sy03sz12st01/baryons/deltapp_deltamm_33/twop_baryon_1

with the dataset I produce in #2 :

/sx04sy03sz12st01/Oms[+1.5e-02]

I got complete agreement, however when I do the spin projection in
/sx04sy03sz12st01/baryons/deltapp_deltamm_33/twop_baryon_1

to obtain

DeltaPl/Pm_Cgi_Cgi

I got differences.

Also running chech-baryons.py fails for the spin 3/2 baryons:

FAILED: groups baryons_u[+1.1e-02]d[-1.1e-02]/DeltaPlPl/Pm_Cgi_Cgi ['baryons/deltapp_deltamm_11/twop_baryon_1', 'baryons/deltapp_deltamm_22/twop_baryon_1', 'baryons/deltapp_deltamm_33/twop_baryon_1'] do not match.
The following components do no pass the test: (first 10)
(0, 0) (-0.0940071-0.0231222j) (-0.0884332023561-0.0191249400377j) 0.0366271545788
(0, 1) (-0.0941744-0.0231738j) (-0.0882135257125-0.0194719992578j) 0.0374613959138
(0, 2) (-0.0931717-0.0218716j) (-0.0881368406117-0.018774015829j) 0.0318146858027
(0, 3) (-0.0934833-0.0241543j) (-0.0882496424019-0.0192721504718j) 0.0383048717544
(0, 4) (-0.0945829-0.0219815j) (-0.0882313661277-0.0191632471979j) 0.037082009364
(0, 5) (-0.0934231-0.0249541j) (-0.0883710719645-0.0195909142494j) 0.0393647009894
(0, 6) (-0.0938502-0.0229162j) (-0.0883552059531-0.0188320260495j) 0.0366271337879
(1, 0) (0.114969-0.00194483j) (0.105237729847-0.00150409713387j) 0.0442316863812
(1, 1) (0.115058-0.00156375j) (0.104984328151-0.00138634617906j) 0.0457853266617
(1, 2) (0.114483-0.00116064j) (0.105101559311-0.00114251184277j) 0.0427204379938

The question is that what is the consequence of this small mismatch regarding the QCD spectrum calculation, and where is the problem in the code.

[pittlerf]

On the local Bonn cluster I regenerated the correlators for the cB211a.25.32, now I basically have full statistics (every second config from 0126 with 16 source position). I used the executable omegamass from my pull request #2 . Now this includes contractBaryonsUDSC as well, however for propagators without applying boundary conditions. (I am doing this in the post-production phase now). I attach the calculated effective masses using the 'log' definitions for the contractions coming from the D_diagramms routine from pull request #1 , and contractBaryonsUDSC from the master branch. You can basically see, the platou region the two effective masses basically agree, just the scattering one is less noisy.
omega_calc2pt_scatt1.pdf

[sbacchio]

Hi Ferenc, thank you for reporting about this. In the past all checks were passed for those baryons contractions. A bug probably go in at some point. I need go back in the git history and understand it better.