Section RI_MP2

• Parameters influencing the RI-MP2 method. RI-MP2 supports gradients.

• Section path: CP2K_INPUT / FORCE_EVAL / DFT / XC / WF_CORRELATION / RI_MP2

• This section cannot be repeated.

• This section cites the following reference: [DelBen2013]

• Subsections

  • CPHF

• Keywords

  • SECTION_PARAMETERS
  • BLOCK_SIZE
  • EPS_CANONICAL
  • FREE_HFX_BUFFER
  • USE_OLD_GRADIENT_CODE

• Keyword descriptions

  SECTION_PARAMETERS
  	&RI_MP2 {Logical}
  	Putting the &RI_MP2 section activates RI-MP2 calculation.
  	This keyword cannot be repeated and it expects precisely one logical.
  	The lone keyword behaves as a switch to .TRUE.
  	Default value: .FALSE.

  BLOCK_SIZE
  	BLOCK_SIZE {Integer}
  	Determines the blocking used for communication in RI-MP2. Larger BLOCK_SIZE reduces communication but requires more memory. The default (-1) is automatic.
  	This keyword cannot be repeated and it expects precisely one integer.
  	Default value: -1
  	Alias names for this keyword: MESSAGE_SIZE

  EPS_CANONICAL
  	EPS_CANONICAL {Real}
  	Threshold for discriminate if a given ij pairs of the unrelaxed MP2 density matrix has to be calculated with a canonical reformulation based on the occupied eigenvalues differences.
  	This keyword cannot be repeated and it expects precisely one real.
  	Default value: 1.00000000E-007

  FREE_HFX_BUFFER
  	FREE_HFX_BUFFER {Logical}
  	Free the buffer containing the 4 center integrals used in the Hartree-Fock exchange calculation. This will be effective only for gradients calculations, since for the energy only case, the buffers are released by default. (Right now debugging only).
  	This keyword cannot be repeated and it expects precisely one logical.
  	The lone keyword behaves as a switch to .TRUE.
  	Default value: .TRUE.

  USE_OLD_GRADIENT_CODE
  	USE_OLD_GRADIENT_CODE {Logical}
  	Use the original RI-MP2 gradient code.
  	This keyword cannot be repeated and it expects precisely one logical.
  	The lone keyword behaves as a switch to .TRUE.
  	Default value: .TRUE.