n2p2 - A neural network potential package
Public Member Functions | Public Attributes | List of all members
nnp::Training::SubCandidate Struct Reference

Contains update candidate which is grouped with others to specific parent update candidate (e.g. More...

Collaboration diagram for nnp::Training::SubCandidate:
Collaboration graph
[legend]

Public Member Functions

bool operator< (SubCandidate const &rhs) const
 Overload < operator to sort in descending order. More...
 

Public Attributes

std::size_t a
 Atom index (only used for force candidates). More...
 
std::size_t c
 Component index (x,y,z -> 0,1,2, only used for force candidates). More...
 
double error
 Absolute value of error with respect to reference value. More...
 

Detailed Description

Contains update candidate which is grouped with others to specific parent update candidate (e.g.

forces belonging to same structure).

Definition at line 369 of file Training.h.

Member Function Documentation

◆ operator<()

bool nnp::Training::SubCandidate::operator< ( SubCandidate const &  rhs) const
inline

Overload < operator to sort in descending order.

Definition at line 379 of file Training.h.

379 {
380 return this->error > rhs.error;
381 }
nnp::Training::SubCandidate::error
double error
Absolute value of error with respect to reference value.
Definition: Training.h:376

References error.

Member Data Documentation

◆ a

std::size_t nnp::Training::SubCandidate::a

Atom index (only used for force candidates).

Definition at line 372 of file Training.h.

Referenced by nnp::Training::update().

◆ c

std::size_t nnp::Training::SubCandidate::c

Component index (x,y,z -> 0,1,2, only used for force candidates).

Definition at line 374 of file Training.h.

Referenced by nnp::Training::update().

◆ error

double nnp::Training::SubCandidate::error

Absolute value of error with respect to reference value.

Definition at line 376 of file Training.h.

Referenced by operator<().

The documentation for this struct was generated from the following file: