34 lines
989 B
Markdown
34 lines
989 B
Markdown
# RSDFT-CIPSI-QMC
|
|
|
|
Overleaf : https://www.overleaf.com/1388147284fqqydnrxrvnr
|
|
|
|
## TODO
|
|
|
|
* Intro
|
|
Jastrow + single determinant = short-range correlation
|
|
- CIPSI DZ : Shrinks Jastrow
|
|
CIPSI TZ : Shrinks Jastrow more
|
|
=> CIPSI and Jastrow have significant overlap
|
|
- Goal : Treat only what Jastrow can't do with CIPSI
|
|
- Mimic Jastrow by DFT => range-separated CIPSI
|
|
|
|
* Protocol
|
|
- Find mu by matching it with Jastrow or standard correlation factor
|
|
- Compute RS-CIPSI
|
|
- Add Jastrow
|
|
- Compute DMC
|
|
|
|
Plot Evar+EPT2 avec plusieurs mu
|
|
- mu=0 : pure DFT -> Converges with 1 det
|
|
- mu=inf : pure CIPSI -> Converges with Ndet
|
|
- Improvement : Faster convergence with Ndet => Reduces size-consistence error due to truncation
|
|
Demo : Example on atomization energy of H2O or N2H4.
|
|
Goal : We need to find the smallest possible mu enabling a correct physical description
|
|
|
|
|
|
Articles a citer:
|
|
- Optimization of nodal surfaces with CIPSI (LCPQ, Claudia)
|
|
- FeS, excitation H2O
|
|
|
|
|