Unified QEL model with spectral function

[LiangLiu212]

Add unified QEL cross-section framework with hadron tensor infrastructure

This PR convert the original Fortran code to c++ code based on the work from Noah Steinberg and Steven Gardiner et, al.

Introduces a hadron-tensor-based formulation for quasi-elastic cross section calculations, replacing direct amplitude evaluation with a contraction of leptonic and hadronic tensors.

New components:

• HadronTensorInterface(I) and Rank2LorentzTensorI: base classes for rank-2 Lorentz tensor manipulation and contraction. New we only implement the one body current. We will introduction two body currents and interference in the future.
• LeptonTensor and ManualResponseTensor: leptonic and hadronic tensor implementations
• UnifiedQELPXSec: tensor-contraction-based QEL cross-section algorithm
• KellyELFormFactorsModel: Kelly parametrization of nucleon EM form factors, it is dipole model, new to GENIE
• LwlynSmithFFEM: electromagnetic variant of the Llewellyn-Smith form-factor model
• onebody_currents_sf: one-body current operators for spectral function calculations
• Spectral function tables for 12C, 16O, and 40Ar (p/n)

Updates SpectralFunc to interface with the new framework, registers new algorithms in master_config.xml and per-algorithm XML configs, and adjusts Makefile and LinkDef entries for the new classes.

[nusense]

I've got a number of questions, but I've looked at all the changes.

[nusense]

These multiple assignments on the same line are hard to read, along with declaring variables ahead and then assigning them here. It would be less fortran-like to have:

double  mNi = fmNi;
double  w = fw;
... 

This is a stylistic suggestion and not a requirement.

[nusense]

I'm a bit confused by this class being Rank2LorentzTensorI as the trailing I by GENIE convention usually means "interface" not a concrete class. And in the \class documentation above you give the name without the I.

In UnifiedQELPXSec it seems to be treated as a concrete class. And while there is a destructor declared, I don't see a constructor.

[LiangLiu212]

Hi, Robert,

Thank you very much for your valuable comments.

This RP is a continue work from #461. At this point, I have mainly converted the one-body current from Fortran to C++ and verified that the C++ code produces the same results as the Fortran version.

I would love to improve the hadron tensor interface according to your comments.

For the Eigen, do we want to introduce a new external library to GENIE?

Best,

Liang

[LiangLiu212]

Hi, Robert,

Here, I rename all the tensors and here is hierarchy of them

               +-------------------------+
                |   Rank2LorentzTensor    |   <<abstract>>
                |   W^mu_nu interface     |
                +-------------------------+
                      ^               ^
                      |               |
          +-----------+               +-----------+
          |                                       |
+---------------------+               +---------------------+
|   NucleonTensor     | <<abstract>>  |   LeptonTensor      | <<concrete>>
|   hadronic side     |               |   L^mu_nu spinors   |
+---------------------+               +---------------------+
          ^
          |
+-------------------------+
|   IASingleNucleonTensor |  <<concrete>>
|   impulse approx.       |
+-------------------------+

I hope it helps.

Best
Liang

[nusense]

The revisions since my previous review have addressed my comments. I'll approve this for when the series of pull requests are applied for version R-3_08_00.