v1.0.4

Description

Add a new finite_jacobian_tensor<N> function which follows the convention that even-order tensors (e.g., matrices) are stored in column-blocks, while odd-order tensors are stored in row-blocks.

This is based on the tensor vectorization presented in "Dynamic Deformables" by Kim and Eberle [2022].

• Update to C++17

Examples

• finite_jacobian_tensor<3> (odd tensor order → row-block layout)
  • $f: \mathbb{R}^n \mapsto \mathbb{R}^{p \times q}$
  • $\nabla f: \mathbb{R}^n \mapsto \mathbb{R}^{pq \times n}$ where every $p$ rows represent one derivative w.r.t. $x_i$ for $0 \leq i < n$
• finite_jacobian_tensor<4> (even tensor order → column-block layout)
  • $f: \mathbb{R}^n \mapsto \mathbb{R}^{p \times q \times r}$
  • $\nabla f: \mathbb{R}^n \mapsto \mathbb{R}^{pq \times rn}$ where every $q$ columns represent one derivative w.r.t. $x_i$ for $0 \leq i < n$

---

Full Changelog: v1.0.3...v1.0.4

v1.0.3

• Switch to CPM for dependencies
• Add support for 3rd-order tensor Jacobians
  • $f: \mathbb{R}^n \mapsto \mathbb{R}^{p \times q}$
  • $\nabla f: \mathbb{R}^n \mapsto \mathbb{R}^{p \times qn}$ where every $q$ columns represent one derivative w.r.t. $x_i$ for $0 \leq i < n$

Full Changelog: v1.0.2...v1.0.3

v1.0.2

Use std::abs

v1.0.1

Use Eigen::Ref for input

v1.0.0

Update continuous.yml