![]() |
Structural Bioinformatics Library
Template C++ / Python API for developping structural bioinformatics applications.
|
#include <lambda_matrix.h>
Public Member Functions | |
lambda_matrix (graph &g1) | |
float | get_node_col (int col1, int col2, int row2) |
float | get_node_row (int row1, int col2, int row2) |
void | clear_lambda () |
void | prohibit (graph &g, vector< int > &lo, vector< int > &up) |
void | allow (vector< int > &lo, vector< int > &up) |
void | add_node_col (int col1, int col2, int row2, float step) |
void | add_node_row (int row1, int col2, int row2, float step) |
void | sub_node_col (int col1, int col2, int row2, float step) |
void | sub_node_row (int row1, int col2, int row2, float step) |
vector< vector< float > > & | get_lnc () |
Protected Attributes | |
graph & | g |
vector< vector< float > > | lambda_node_col |
vector< vector< float > > | lambda_node_row |
vector< float > | coef_lambda_node |
vector< vector< float > > | coef_lambda_edge |
vector< vector< int > > | nz_sgv_c |
int | nz_c |
vector< vector< int > > | nz_sgv_r |
int | nz_r |
vector< vector< bool > > | lambda_modified_nodes |
Lambda_Mat_Apurva Class. Contain the lambda coeficients of the lagrangian relaxation of the A_purva model.
|
inline |
Constructors
|
inline |
update lambda coeficients for constraint col1 / node N_{col2,row2}. add step value to lambda coeficient. This function also update the lambda modified coeficients for node and edges.
|
inline |
update lambda coeficients for constraint row1 / node N_{col2,row2}. add step value to lambda coeficient. This function also update the lambda modified coeficients for node and edges.
|
inline |
Allow edges forbidden by function prohibit, by removing penalty
|
inline |
Modificators
|
inline |
This should be forbidden !!!! But it allows much faster computation ^^;
|
inline |
Accessors Return the lambda coeficients for the relaxed constraint between column col1 and node N_{col2,row2}.
|
inline |
Return the lambda coeficients for the relaxed constraint between row row1 and node N_{col2,row2}.
|
inline |
Prohibit edges going out of lower and upper limits, by giving them very large penalty
|
inline |
Same as add_node_col, but for substracting the step value. It modified the step value to avoid negative lambda (lambda are in R+)
|
inline |
Same as add_node_row, but for substracting the step value. It modified the step value to avoid negative lambda (lambda are in R+)
|
protected |
Lambda modified coeficients for edges
|
protected |
Lambda modified coeficients for nodes
|
protected |
The Contact Map Alignment graph
|
protected |
Table recording which nodes were modified during last sgd iteration, in order not to recompute the weight of all nodes...
|
protected |
Lambda coeficients for relaxed column/node constraints
|
protected |
Lambda coeficients for relaxed row/node constraints
|
protected |
Number of non-zero components of the subgradient vector for lambda column/node
|
protected |
Number of non-zero components of the subgradient vector for lambda row/node
|
protected |
Non-Zero components of the subgradient vector for lambda column/node
|
protected |
Non-Zero components of the subgradient vector for lambda row/node