|
|
using | VectorType = Vector< T, N > |
| |
|
| T * | data () |
| | Returns the data to use for a parameter, asserts you have seeded the data before this.
|
| |
| VectorType & | seedVec () |
| | Returns the vector for you to write into and seed.
|
| |
| void | anchor () |
| | Freezes the current value as the point this state is linearized at, unless it already has one.
|
| |
| const T * | anchorData () const |
| | The frozen linearization point, or null for a state that has never been anchored.
|
| |
| void | releaseAnchor () |
| | Forgets the frozen linearization point, so the state linearizes wherever it currently is again.
|
| |
| void | unseed () |
| | Drops everything held here, for a slot about to be handed to a state of its own.
|
| |
◆ anchor()
template<typename T , int N>
◆ anchorData()
template<typename T , int N>
◆ data()
template<typename T , int N>
Returns the data to use for a parameter, asserts you have seeded the data before this.
◆ releaseAnchor()
template<typename T , int N>
◆ seedVec()
template<typename T , int N>
Returns the vector for you to write into and seed.
Assumes you have seeded it after this function is called.
◆ unseed()
template<typename T , int N>
◆ anchored
template<typename T , int N>
◆ first_estimate
template<typename T , int N>
The value this state is held linearized around, once it has been anchored.
A marginalization prior is a statement about one particular linearization of the states it covers, and it goes on making that statement long after the solve has moved them. Everything that touches an anchored state has to keep linearizing it here, or the frozen factor and the live residuals stop describing the same point.
The documentation for this struct was generated from the following file: