SuperSCS  1.3.2
 All Data Structures Files Functions Variables Typedefs Enumerations Enumerator Macros Pages
Data Fields
scs_direction_cache Struct Reference

Memory for the computation of directions (Broyden and Anderson's methods). More...

#include <scs.h>

Data Fields

scs_floatS
 cached values of \(s_i\) (s-memory) More...
 
scs_floatU
 cached values of \(u_i = \frac{s_i - \tilde{s}_i}{\langle s_i, \tilde{s}_i\rangle}\), or cached values of \(y_i\) More...
 
scs_floatS_minus_Y
 The difference \(S-Y\) (for Anderson's acceleration) More...
 
scs_floatt
 Solution of the linear system \(Yt = R(x)\). More...
 
scs_int ls_wspace_length
 Length of workspace used to solve Anderson's linear system. More...
 
scs_floatls_wspace
 Workspace used to solve Anderson's linear system. More...
 
scs_int mem_cursor
 current memory cursor [0..mem-1] More...
 
scs_int mem
 (target/maximum/allocated) memory More...
 
scs_int current_mem
 current memory length More...
 

Detailed Description

Memory for the computation of directions (Broyden and Anderson's methods).

For Broyden's method, a cache of \((s_i, u_i)\) where \(u_i = \frac{s_i - \tilde{s}_i}{\langle s_i, \tilde{s}_i\rangle}\). We do not need to store past values of \(\tilde{s}_i\).

For Anderson's acceleration, it stores pairs \((s_i, y_i)\).

Field Documentation

scs_int current_mem

current memory length

scs_float* ls_wspace

Workspace used to solve Anderson's linear system.

scs_int ls_wspace_length

Length of workspace used to solve Anderson's linear system.

scs_int mem

(target/maximum/allocated) memory

scs_int mem_cursor

current memory cursor [0..mem-1]

cached values of \(s_i\) (s-memory)

scs_float* S_minus_Y

The difference \(S-Y\) (for Anderson's acceleration)

Solution of the linear system \(Yt = R(x)\).

cached values of \(u_i = \frac{s_i - \tilde{s}_i}{\langle s_i, \tilde{s}_i\rangle}\), or cached values of \(y_i\)


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