Accelerated precomputation of reduced deformable models
Technologies are disclosed for precomputation of reduced deformable models. In such precomputation, a Krylov subspace iteration may be used to construct a series of inertia modes for an input mesh. The inertia modes may be condensed into a mode matrix. A set of cubature points may be sampled from the input mesh, and cubature weights of the set of cubature points may be calculated for each of the inertia modes in the mode matrix. A training dataset may be generated by iteratively adding training samples to the training dataset until a training error metric converges, wherein each training sample is generated from an inertia mode in the mode matrix and corresponding cubature weights. The reduced deformable model may be generated, including inertia modes in the training dataset and corresponding cubature weights.
1. A method to precompute a reduced deformable model for an object represented by an input mesh, comprising:
applying a Krylov subspace iteration to construct a series of inertia modes for the input mesh;
condensing the inertia modes into a mode matrix;
sampling a set of cubature points from the input mesh, and calculating cubature weights for the set of cubature points, for each of the linear inertia modes in the mode matrix;
generating a training dataset by iteratively adding training samples to the training dataset until a training error metric converges, wherein each training sample is generated from an inertia mode in the mode matrix and corresponding cubature weights; and
generating the reduced deformable model, wherein the reduced deformable model includes inertia modes in the training dataset and corresponding cubature weights.
2. The method of claim 1 , wherein the set of cubature points are uniformly sampled from the input mesh.
3. The method of claim 1 , wherein the series of inertia modes comprises linear inertia modes and nonlinear inertia modes, and further comprising computing asymptotic inertia derivatives for the input mesh in order to generate the nonlinear inertia modes.
4. The method of claim 1 , further comprising using the reduced deformable model in one or more online simulations.
5. The method of claim 1 , wherein iteratively adding training samples to the training dataset comprises adding more than one inertia mode at each iteration.
6. The method of claim 1 , further comprising applying a Gram-Schmidt orthogonalization scheme to regularize the constructed series of inertia modes.
7. The method of claim 1 , wherein a random projection method is used to condense the inertia modes in the mode matrix.
8. A computing device equipped to precompute reduced deformable models, comprising:
a processor;
a memory; and
a precomputation application stored in the memory and executable by the processor, wherein the precomputation application is adapted to precompute a reduced deformable model for an object represented by an input mesh by causing the processor to:
apply a Krylov subspace iteration to construct a series of inertia modes for the input mesh;
condense the inertia modes into a mode matrix;
sample a set of cubature points from the input mesh, and calculating cubature weights for the set of cubature points, for each of the linear inertia modes in the mode matrix;
generate a training dataset by iteratively adding training samples to the training dataset until a training error metric converges, wherein each training sample is generated from an inertia mode in the mode matrix and corresponding cubature weights; and
generate the reduced deformable model, wherein the reduced deformable model includes inertia modes in the training dataset and corresponding cubature weights.
9. The computing device of claim 8 , wherein the set of cubature points are uniformly sampled from the input mesh.
10. The computing device of claim 8 , wherein the series of inertia modes comprises linear inertia modes and nonlinear inertia modes, and wherein the precomputation application causes the processor to compute asymptotic inertia derivatives for the input mesh in order to generate the nonlinear inertia modes.
11. The computing device of claim 8 , further comprising a simulator application which causes the processor to use the reduced deformable model in one or more online simulations.
12. The computing device of claim 8 , wherein iteratively adding training samples to the training dataset comprises adding more than one inertia mode at each iteration.
13. The computing device of claim 8 , wherein the precomputation application causes the processor to apply a Gram-Schmidt orthogonalization scheme to regularize the constructed series of inertia modes.
14. The computing device of claim 8 , wherein a random projection method is used to condense the inertia modes in the mode matrix.