From 4e3f2e891a01599af63faa6998c6cbf214934515 Mon Sep 17 00:00:00 2001 From: Anthony Scemama Date: Mon, 18 Dec 2017 18:23:35 +0100 Subject: [PATCH] CAS-SD --- Excited-states.md | 13 +++++++++++++ Home.mediawiki | 1 + MR-CI.md | 27 +++++++++++++++++++++++++++ 3 files changed, 41 insertions(+) create mode 100644 MR-CI.md diff --git a/Excited-states.md b/Excited-states.md index f852118..1744961 100644 --- a/Excited-states.md +++ b/Excited-states.md @@ -12,3 +12,16 @@ qp_run save_natorb file.ezfio The MOs will be replaced, so the two-electron integrals and the wave function are invalidated as well. + +If you are computing multiple states, it is often a good idea to have the ``s2_eig`` flag set to true. This will force the Davidson algorithm to choose only vectors with a value of S^2 equal to the ``expected_s2``. Otherwise, you can have different spin states coming out in the diagonalization. + +Also, you need to have a number of states in the Davidson algorithm (``n_states_diag``) which is greater than ``n_states``. + + +Usually, it is good practice to use state-averaged MOs so that all states have MOs of comparable quality. For example, if you search for a singly excited state, you can use state-average natural orbitals of a preliminary CIS calculation. + + +The QP doesn't take account of the symmetry. For reasons due to numerical noise, +excited states of different symmetries may enter in the calculation. +Note that it is possible to make state-average calculation of states with different +symmetries and/or different spin multiplicities. diff --git a/Home.mediawiki b/Home.mediawiki index d4de3c1..e922246 100644 --- a/Home.mediawiki +++ b/Home.mediawiki @@ -6,6 +6,7 @@ For users: * [[Installation instructions]] * [[Installation behind a firewall]] * [[Tutorial]] +* [[MR-CI]] * [[Excited states]] diff --git a/MR-CI.md b/MR-CI.md new file mode 100644 index 0000000..4940cc2 --- /dev/null +++ b/MR-CI.md @@ -0,0 +1,27 @@ +If you want to do a selected CI calculation in a space containing the CAS-CI and all singles and doubles on top of it, you need to define different classes of MOs. This is done using the ``qp_set_mo_class`` program. + +Here is an example for F_2 in the cc-pVTZ basis set, using Hartree-Fock MOs. We want to freeze the first 2 MOs, use the MOs 7 and 10 as the active space and make all singles and doubles on top of it: + +``` +$ qp_set_mo_class -core "[1,2]" -inact "[3-6,8-9]" -act "[7,10]" -virt "[11-60]" f2.ezfio +Core : (1, 2) +Inactive : (3, 4, 5, 6, 8, 9) +Active : (7, 10) +Virtual : (11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60) +Deleted +0000000000000000000000000000000000000000000000000000001111111100 +0000111111111111111111111111111111111111111111111111111001000000 +0000000000000000000000000000000000000000000000000000001111111100 +0000111111111111111111111111111111111111111111111111111001000000 +0000000000000000000000000000000000000000000000000000001111111100 +0000111111111111111111111111111111111111111111111111111001000000 +``` + +Now you can run ``qp_run cas_sd`` to make the CIPSI selection in the CAS+SD space. + +Warning #1: If you run ``qp_run save_natorb``, the order of the MOs may change. The active MOs will be placed between the inactive and the virtual, so you may need to re-define the classes of the MOs with the new ordering. + +Warning #2: If you want to do a Full-CI after, you will need to re-set the MO classes such that you don’t have inactive and virtuals MOs. The simplest way is to run ``qp_set_frozen_core.py``. This program runs qp_set_mo_class with an automatically selected valence Full-CI space. + + +CAS-SD calculations are the starting point of MR-CCSD calculations.