update

Author: fishjojo

source/develop/ao2mo_developer.rst

@@ -0,0 +1,6 @@
+.. _developer_ao2mo:
+
+AO to MO transformation
+***********************
+
+*Modules*: :mod:`ao2mo`

source/develop/cc_developer.rst

@@ -0,0 +1,6 @@
+.. _developer_cc:
+
+Coupled Cluster Methods
+***********************
+
+*Modules*: :mod:`cc`, :mod:`pbc.cc`

source/develop/df_developer.rst

@@ -0,0 +1,6 @@
+.. _developer_df:
+
+Density fitting
+***************
+
+*Modules*: :mod:`df`, :mod:`pbc.df`

source/develop/dft_developer.rst

@@ -1,4 +1,48 @@
+.. _developer_scf:
 
-Include here I think the main mods of DFT vs HF, the numint module, the
-key path in numint, and important functions e.g. eval_ao, etc.
+*******************************
+Density functional theory (DFT)
+*******************************
+
+*Modules*: :mod:`dft`, :mod:`pbc.dft`
+
+Overview
+========
+The DFT methods in modules :mod:`pyscf.dft` and :mod:`pyscf.pbc.dft` 
+are implemented as derived classes of the base SCF class :class:`pyscf.scf.hf.SCF`. 
+The major modifications of DFT against HF include re-implementation of the 
+:meth:`get_veff` and :meth:`energy_elec` methods.
+In addition to :meth:`get_jk`, :meth:`get_veff` also calls 
+:func:`numint.nr_rks` (for spin-restricted cases) or 
+:func:`numint.nr_uks` (for spin-unrestricted cases) 
+to compute the exchange-correlation (XC) energy and potential.
+The XC energy is added to the total electronic energy in :meth:`energy_elec`.
+
+Numerical integration
+=====================
+The XC functionals are evaluated on numerical quadrature grids.
+These grids are defined in modules :mod:`pyscf.dft.gen_grid` and
+:mod:`pyscf.pbc.dft.gen_grid`  for molecules and solids, respectively.
+The actual methods to evaluate those XC functionals and their related integrals
+are implemented in modules :mod:`pyscf.dft.numint` and :mod:`pyscf.pbc.dft.numint`.
+For example, the XC energy and potential matrix for a given density matrix are computed by 
+:meth:`nr_rks` (or :meth:`nr_uks`), which internally calls
+
+- :func:`eval_ao` -- to compute the atomic orbitals (AOs) and their derivatives on the grid
+
+- :func:`eval_rho` -- to compute the electron density and density derivatives on the grid
+
+- :func:`eval_xc` -- to compute the XC energy and potential through an interface to the external library Libxc (:func:`pyscf.dft.libxc.eval_xc`)
+  or XCFun (:func:`pyscf.dft.xcfun.eval_xc`). 
+
+Other convenient functions implemented in :mod:`numint` include
+
+- :func:`eval_mat` -- evaluating the XC potential matrix with AO, electron density, and XC potential values on the grid as the input 
+
+- :func:`nr_vxc` -- evaluating the XC energy and potential matrix with the density matrix as the input
+
+- :func:`nr_sap_vxc` -- evaluating the superposition of atomic potentials matrix, which is used as the initial guess for :math:`v_{\rm eff}`
+  when setting :attr:`mf.init_guess` to ``'vsap'``.
+
+- :func:`nr_rks_fxc`, :func:`nr_uks_fxc` -- evaluating the XC kernel matrix
 

source/develop/fci_developer.rst

@@ -0,0 +1,6 @@
+.. _developer_fci:
+
+Full CI solver
+**************
+
+*Modules*: :mod:`fci`

source/develop/mcscf_developer.rst

@@ -0,0 +1,6 @@
+.. _developer_mcscf:
+
+Multi-configuration SCF
+***********************
+
+*Modules*: :mod:`mcscf`

source/develop/scf_developer.rst

@@ -114,13 +114,9 @@ The molecular HF operator relies on the following underlying AO integral functio
   If enough memory is provided, the two-electron repulsion integral is computed (:meth:`mol.intor('int2e')`)
   and saved as the attribute :attr:`_eri`, 
   and is then contracted with the DM (:func:`pyscf.scf.hf.dot_eri_dm`) (incore algorithm).
-  Otherwise, a "direct" algorithm, where the AO integrals are computed on the flow,
-  is used (:func:`pyscf.scf._vhf.direct`). CHECK------- The incore path can be forced by 
-  setting :attr:`.incore_anyway` of the :class:`Mole` object to `True`. 
-
-.. The eigenvalues and eigenvectors of the Hamiltonian is computed by solving the 
-.. generalized eigenvalue problem (:func:`pyscf.scf.hf.eig`).
-.. And the electronic energy is computed by contracting the Hamiltonian with the DM (:func:`pyscf.scf.hf.energy_elec`).
+  Otherwise, a "direct" algorithm, where the AO integrals are computed on the fly,
+  is used (:func:`pyscf.scf._vhf.direct`). The incore path can be forced by 
+  setting :attr:`.incore_anyway` of the :class:`Mole` object to ``True``. 
 
 Custom Hamiltonians
 ===================
@@ -138,7 +134,7 @@ The following shows an example of HF with a Hubbard model Hamiltonian::
 
     # incore_anyway=True ensures the customized Hamiltonian (the _eri attribute)
     # is used.  Without this parameter,  MO integral transformation used in
-    # subsequent post-HF calculations may <-----------CHECK CHECK
+    # subsequent post-HF calculations may 
     # ignore the customized Hamiltonian if there is not enough memory.
     mol.incore_anyway = True
 

source/develop/soscf_developer.rst

@@ -0,0 +1,6 @@
+.. _developer_soscf:
+
+Second-order SCF solver
+***********************
+
+*Modules*: :mod:`soscf`

source/index.rst

@@ -43,6 +43,7 @@ See :ref:`installing_plugin`.
 
    code-rule.rst
    develop/scf_developer.rst
+   develop/dft_developer.rst
    advanced.rst
 
 .. toctree::

source/theory/cc.rst

@@ -1,7 +1,7 @@
-
+.. _theory_scf:
 
 ********************
 Coupled cluster (CC)
 ********************
 
-*Modules*: :ref:`cc <cc>`, :ref:`pbc.cc <pbc_cc>`
+*Modules*: :mod:`cc`, :mod:`pbc.cc`