Feature: support list input parse and multiple values for bessel_nao_rcut

[kirk0830]

Feature

Present implementation

Under project of ABACUS Pseudopot-Nao Square (APNS), for quick generation of numerical orbitals and corresponding quick tests in batch, the orbital generation procedure needs to be fast as much as possible. I find present workflow of orbital generation will run identical SCF (differs only with bessel_nao_rcut) for times as many as rcut requested by user. Therefore I make ABACUS support mutiple values of bessel_nao_rcut.
For example, If only one value is given to keyword bessel_nao_rcut in INPUT:

...
bessel_nao_rcut     7
...

One can still find the overlap matrices calculated and named in original way:

root@legion:~/documents/simulation/abacus/test_bessel_nao_rcut/OUT.Er-STRU1-8-2.8# ll
total 28880
drwxr-xr-x 2 root root     4096 Jan  9 11:24 ./
drwxr-xr-x 4 root root     4096 Jan  9 11:24 ../
-rw-r--r-- 1 root root 14778818 Jan  9 11:24 orb_matrix.0.dat
-rw-r--r-- 1 root root 14779187 Jan  9 11:24 orb_matrix.1.dat

If four values given, which means in INPUT it is:

...
bessel_nao_rcut     7 8 9 10
...

Then stdout wil be like:

                                                                                     
                              ABACUS v3.5.0

               Atomic-orbital Based Ab-initio Computation at UStc                    

                     Website: http://abacus.ustc.edu.cn/                             
               Documentation: https://abacus.deepmodeling.com/                       
                  Repository: https://github.com/abacusmodeling/abacus-develop       
                              https://github.com/deepmodeling/abacus-develop         
                      Commit: 8afa4d05d (Tue Jan 9 10:57:24 2024 +0800)

 Tue Jan  9 11:43:18 2024
 MAKE THE DIR         : OUT.Er-STRU1-8-2.8/
 RUNNING WITH DEVICE  : CPU / Intel(R)%Xeon(R)%Platinum
 UNIFORM GRID DIM        : 128 * 128 * 128
 UNIFORM GRID DIM(BIG)   : 128 * 128 * 128
 DONE(0.0601203  SEC) : SETUP UNITCELL
 DONE(0.0604865  SEC) : INIT K-POINTS
 ---------------------------------------------------------
 Self-consistent calculations for electrons
 ---------------------------------------------------------
 SPIN    KPOINTS         PROCESSORS  
 1       1               16          
 ---------------------------------------------------------
 Use plane wave basis
 ---------------------------------------------------------
 ELEMENT NATOM       XC          
 Er      2           
 ---------------------------------------------------------
 Initial plane wave basis and FFT box
 ---------------------------------------------------------
 DONE(0.0734675  SEC) : INIT PLANEWAVE
 MEMORY FOR PSI (MB)  : 4.24988
 DONE(0.173127   SEC) : LOCAL POTENTIAL
 DONE(0.23356    SEC) : NON-LOCAL POTENTIAL
 DONE(0.390492   SEC) : INIT BASIS
 -------------------------------------------
 SELF-CONSISTENT : 
 -------------------------------------------
 START CHARGE      : atomic
 DONE(0.708638   SEC) : INIT SCF
 ITER   ETOT(eV)       EDIFF(eV)      DRHO       TIME(s)    
 CG1    -2.448479e+03  0.000000e+00   4.493e-01  1.959e+00  
 CG2    -2.449854e+03  -1.374351e+00  2.098e-01  1.336e+00  
 CG3    -2.450345e+03  -4.913768e-01  3.520e-03  1.030e+00  
 CG4    -2.450357e+03  -1.196023e-02  1.163e-03  2.260e+00  
 CG5    -2.450359e+03  -1.459728e-03  2.785e-04  1.256e+00  
 CG6    -2.450359e+03  -5.938095e-04  1.842e-04  1.465e+00  
 CG7    -2.450360e+03  -8.549907e-04  5.625e-05  1.024e+00  
 CG8    -2.450361e+03  -5.595582e-04  3.645e-05  1.481e+00  
 CG9    -2.450361e+03  -4.486177e-04  1.996e-05  1.287e+00  
 CG10   -2.450361e+03  -3.566487e-04  1.292e-05  1.585e+00  
 CG11   -2.450361e+03  -1.622941e-04  5.167e-07  1.731e+00  
 CG12   -2.450362e+03  -3.426222e-05  4.295e-07  2.575e+00  
 CG13   -2.450362e+03  -1.644875e-05  5.872e-08  1.714e+00  
update value: bessel_nao_rcut <- 7.000 a.u.
 DONE(26.973     SEC) : cal_overlap_Q
 DONE(39.772     SEC) : cal_overlap_Sq
 DONE(43.926     SEC) : cal_overlap_Q
 DONE(57.017     SEC) : cal_overlap_Sq
update value: bessel_nao_rcut <- 8.000 a.u.
 DONE(61.169     SEC) : cal_overlap_Q
 DONE(74.182     SEC) : cal_overlap_Sq
 DONE(78.270     SEC) : cal_overlap_Q
 DONE(91.244     SEC) : cal_overlap_Sq
update value: bessel_nao_rcut <- 9.000 a.u.
 DONE(95.460     SEC) : cal_overlap_Q
 DONE(108.472    SEC) : cal_overlap_Sq
 DONE(112.553    SEC) : cal_overlap_Q
 DONE(125.577    SEC) : cal_overlap_Sq
update value: bessel_nao_rcut <- 10.000 a.u.
 DONE(129.774    SEC) : cal_overlap_Q
 DONE(142.820    SEC) : cal_overlap_Sq
 DONE(147.045    SEC) : cal_overlap_Q
 DONE(160.081    SEC) : cal_overlap_Sq
 DONE(161.197    SEC) : BASIS OVERLAP (Q and S) GENERATION.

One can find four groups of files in OUT.${suffix}:

root@legion:~/documents/simulation/abacus/test_bessel_nao_rcut/OUT.Er-STRU1-8-2.8# ll
total 115496
drwxr-xr-x 2 root root     4096 Jan  9 11:28 ./
drwxr-xr-x 4 root root     4096 Jan  9 11:28 ../
-rw-r--r-- 1 root root 14778818 Jan  9 11:28 orb_matrix_rcut10deriv0.dat
-rw-r--r-- 1 root root 14779187 Jan  9 11:28 orb_matrix_rcut10deriv1.dat
-rw-r--r-- 1 root root 14778818 Jan  9 11:28 orb_matrix_rcut7deriv0.dat
-rw-r--r-- 1 root root 14779187 Jan  9 11:28 orb_matrix_rcut7deriv1.dat
-rw-r--r-- 1 root root 14778818 Jan  9 11:28 orb_matrix_rcut8deriv0.dat
-rw-r--r-- 1 root root 14779187 Jan  9 11:28 orb_matrix_rcut8deriv1.dat
-rw-r--r-- 1 root root 14778818 Jan  9 11:28 orb_matrix_rcut9deriv0.dat
-rw-r--r-- 1 root root 14779187 Jan  9 11:28 orb_matrix_rcut9deriv1.dat

For python-end the PTG_dpsi, I will submit PR to repo ABACUS-orbitals: https://github.com/abacusmodeling/ABACUS-orbitals

Future heterogeneous container demand

Present implementation suggests use std::vector<std::string> to get value from input, then cast values one-by-one and case-by-case to values in datatype required.

Reminder

• Have you linked an issue with this pull request?
• Have you noticed possible changes of behavior below or in the linked issue?
• Have you explained the changes of codes in core modules of ESolver, HSolver, ElecState, Hamilt, Operator or Psi? (ignore if not applicable)

Linked Issue

Fix #3334

What's changed?

• Add template function template <typename T> read_value2stdvector(std::ifstream& ifs, std::vector<T>& var) in module_io/input, Input class, along with two instantiation for int and double.
• Add corresponding unittest.
• Add variable std::vector<double> bessel_nao_rcuts
• Support multiple values input of bessel_nao_rcut, save the first value to INPUT.bessel_nao_rcut and all values to INPUT.bessel_nao_rcuts by read_value2stdvector.
• Therefore support a external loop over calculation of OVERLAP_Q and OVERLAP_S by numerical basis instance in esolver_ks_pw.cpp.
• still compatible with the old one, if only one value is given to keyword bessel_nao_rcut

Any changes of core modules? (ignore if not applicable)

• expand the behavior of numerical basis

[WHUweiqingzhou]

This PR offers a new function Input::read_value2stdvector to read a vector<T> from one line. For heterogeneous vector, you can use string first, and cast which type you want.