columns of the matrix

rows of the matrix

returns the dimensions of the error function

Prints a text message. And, optionally, if a Numpy array is provided it prints it's shape, min, and max values.

set the ith vertex on the hyper-edge to the pointer supplied */

true => this node is fixed during the optimization

temporary index of this node in the parameter vector obtained from linearization

true => this node is marginalized out during the optimization

set the estimate for the vertex also calls updateCache()

return b, the right hand side of the system

* lock for the block of the hessian and the b vector associated with this vertex, to avoid * race-conditions if multi-threaded. */

* Update the position of the node from the parameters in v. * Depends on the implementation of oplusImpl in derived classes to actually carry

* unlock the block of the hessian and the b vector associated with this vertex */

return x, the solution vector

The identity_block is the block that has no conv layer at shortcut # Arguments input_tensor: input tensor kernel_size: defualt 3,

if NOT NULL, error of this edge will be robustifed with the kernel

where does the row at block-row r starts?

return the size of the solution vector (x) and b

* Returns whether the node is a leaf node * @return true iff the node is a leaf */

set the temporary index of the vertex in the parameter blocks

sets the level of the edge

* @brief Ellipse * @return the ellipse */