Processing math: 0%

Skin weight optimization (lagrange) (4)

Constraint optimization (Lagrange multiplier)

We introduce the Lagrange energy function as:

\mathcal L(\vec w, \vec \lambda) =  E(\vec w) - \sum_{i=1}^v { \lambda_i g_i( \vec w)  }

with \vec \lambda = [ \lambda_0, \lambda_1, \lambda_2, \cdots, \lambda_v ] and g_i a constraint at each vertex to preserve skin weight normalization. Finding the minimum \vec w_{min} of E under the constraints g_i is equivalent to solve the system:

\nabla \mathcal L(\vec w, \vec \lambda) = 0 \iff \left \{ \begin{array}{lll} \nabla E(\vec w) - \sum_{i=1}^v {\lambda_i \nabla g_i (\vec w)} & = & 0\\  g_1(\vec w) & = & 0 \\ \vdots & & \\ g_v(\vec w) & = & 0\\ \end{array} \right .

A single constraint is defined as:
\begin{array}{lll} g_i( \vec w_i) & = &  0 \\ 1 - \sum_{j=1}^{|\vec w_i|} w_{ij}  & = & 0 \\ \left [ \begin{matrix} 1 & \dots & 1 \end{matrix} \right ] \vec w_i & = &  1 \\ \mathbf g_i^T \vec w_i & = &  1 \\ \end{array}

We can represent the system of constraints for every vertices:
 \begin{bmatrix} 1 & \cdots & 1 &           0  &  0           &  0         & \cdots & 0 \\ 0         &    \cdots &        0  &  1 & 1   &  0         & \cdots & 0 \\ 0         &   \cdots  &        0 &           0  &          0    & 1 &   \cdots       & 0 \\ \vdots  &  \vdots &  \vdots &     \vdots &     \vdots   &               0  &   \ddots &        0  \\ 0          &   \cdots  &        0  &            0  &          0   &          0   &         0  &  1 \\  \end{bmatrix} . \begin{bmatrix} \vec w_0 \\ \vec w_1 \\ \vec w_2 \\ \vdots \\ \vec w_v \\ \end{bmatrix}  = \vec 1 \\ \mathbf C \vec w = \vec m

As for the system \nabla E(\vec w) - \sum_{i=1}^v {\lambda_i \nabla g_i (\vec w)} = 0 we already know \nabla E(\vec w) = (\mathbf A \vec w - \vec b) so lets focus on g(\vec w) = - \sum_{i=1}^v {\lambda_i \nabla g_i (\vec w)} of our system:
\begin{array}{lll} \nabla g_i (\vec w) & = & 1 - \sum_{j=1}^{|\vec w_i|} w_{ij} \\ & = & \nabla[1] - \nabla \left [ \sum_{j=1}^{|\vec w_i|} w_{ij} \right ] \\ & = & 0 - \begin{bmatrix} 0 & \dots & 0 & 1_{i0} & \dots & 1_{ij} & \dots & 1_{i|\vec w_i|} & 0 & \dots & 0 \end{bmatrix}^T \\ & = & - \vec 1_i \text{ vector of ones for the entries } |\vec w_i| \text{ of vertex i and zero otherwise} \end{array}

\begin{array}{lll} \nabla g(\vec w) & = & - \sum_{i=1}^v {\lambda_i \nabla g_i (\vec w)} \\ & = & - \sum_{i=1}^v {\lambda_i . (- \vec 1_i)} \\ & = & \sum_{i=1}^v { \vec \lambda_i }\\ & = & \begin{bmatrix} \vec 1_0 & \vec 1_1 & \dots & \vec 1_i & \dots & \vec 1_{v}  \end{bmatrix} \vec \lambda\\ & = & \begin{array}{c|cccccc|} \nabla \mathbf G & 0        & 1        & \dots & i & \dots & v \\ \hline w_{00} & 1        & 0        & \dots  & 0 & \dots & 0 \\ w_{10} & 0        & 1        & 0  & 0  & 0 & 0 \\ w_{11} & 0        & 1        & 0  & 0 & 0 & 0 \\ & 0        & 0        & \ddots  & 0        & 0        & 0 \\ w_{i0} & 0        & 0        & 0         & 1        & 0        & 0 \\ w_{ij} & \vdots & \vdots & \vdots  & \vdots & \vdots  & \vdots \\ w_{in_i}& 0        & 0        & 0         & 1        & 0        & 0 \\ & 0        & 0        & 0         & 0        & \ddots & 0 \\ w_{vn_v} & 0        & 0        & 0         & 0        & 0        & 1 \\ \end{array} \phantom{.} . \begin{bmatrix} \lambda_0 \\ \lambda_1 \\ \vdots \\ \lambda_i \\ \vdots \\ \lambda_v \\ \end{bmatrix} \\ & = & \nabla \mathbf G \vec \lambda\\ \end{array}

Which translates to:

\begin{array}{lll} \nabla E(\vec w) - \sum_{i=1}^v {\lambda_i \nabla g_i (\vec w)} & = & 0\\ (\mathbf A \vec w - \vec b) + \nabla \mathbf G \vec \lambda & = & 0\\ \mathbf A \vec w  + \nabla \mathbf G \vec \lambda & = & \vec b \\ \end{array}

Let's seek the matrix representation \mathbf M \mathbf x = \mathbf m of \nabla \mathcal L(\vec w, \vec \lambda) = 0 :

\begin{bmatrix} \mathbf A  & \nabla \mathbf G \\ \mathbf C  & 0 \\ \end{bmatrix} \begin{bmatrix} \vec w_0 \\ \vdots \\ \vec w_i \\ \vdots \\ \vec w_j \\ \vdots \\ \vec w_v \\ \lambda_0 \\ \vdots \\ \lambda_i \\ \vdots \\ \lambda_v \\ \end{bmatrix} = \begin{bmatrix} \vec b_0 \\ \vdots \\ \vec b_i \\ \vdots \\ \vec b_j \\ \vdots \\ \vec b_v \\ m_0 \\ \vdots \\ m_i \\ \vdots \\ m_v \\ \end{bmatrix}

No comments

(optional field, I won't disclose or spam but it's necessary to notify you if I respond to your comment)
All html tags except <b> and <i> will be removed from your comment. You can make links by just typing the url or mail-address.
Anti-spam question: