model#
Neural quantum states#
It’s recommended to use ResConv when one needs a deep NQS.
|
Network with one dense layer \(\psi(s) = \prod f(W s + b)\). |
|
The restricted Boltzmann machine with one dense layer \(\psi(s) = \prod \cosh(W s + b)\). |
|
Network with one convolutional layer \(\psi(s) = \prod f(\mathrm{Conv}(s))\). |
|
The restricted Boltzmann machine with one convolutional layer \(\psi(s) = \prod \cosh(\mathrm{Conv}(s))\). |
|
Deep convolutional residual network. |
Fermionic Mean-field#
|
General determinant wavefunction \(\psi(n) = \mathrm{det}(n \star U)\). |
|
Restricted determinant wavefunction \(\psi(n) = \mathrm{det}(n_\uparrow \star U) \mathrm{det}(n_\downarrow \star U)\). |
|
Unrestricted determinant wavefunction \(\psi(n) = \mathrm{det}(n_\uparrow \star U_\uparrow) \mathrm{det}(n_\downarrow \star U_\downarrow)\). |
|
Multi-determinant wavefunction \(\psi(n) = \sum_i c_i \mathrm{det}(n \star U_i)\). |
|
General Pfaffian wavefunction \(\psi(n) = \mathrm{pf}(n \star F \star n)\). |
|
Singlet paired wavefunction \(\psi(n) = \mathrm{det}(n_\uparrow \star F \star n_\downarrow)\). |
|
Multi-Pfaffian wavefunction \(\psi(n) = \sum_i \mathrm{pf}(n \star F_i \star n)\). |
Jastrow factors#
|
Traditional two-body Jastrow factor with a symmetric matrix \(W\) of variational parameters. |
|
A reference state multiplied by a generalized (neural-network) Jastrow factor, \(\psi(s) = J(s)\, \psi_\mathrm{mf}(s)\), where \(J(s)\) is the output of an arbitrary network |