Table of Contents
This documentation component focuses on Chrono mathematical functions and classes. These concepts are quite ubiquitous in the rest of the Chrono API and they are discussed also in the Linear algebra API documentation.
Linear algebra
Handling vectors and matrices is a recurring theme throughout the Chrono API. Chrono uses Eigen3 for representing all matrices (dense and sparse) and vectors.
Dense matrices in Chrono are templated by the scalar type and have row-major storage order. All Chrono matrix and vector types below are simply aliases to appropriate Eigen matrix types; see Linear algebra and the ChMatrix.h header.
Matrices are indexed starting from 0, with (row,column) indexing:
\[ \mathbf{A}=\left[ \begin{array}{cccc} a_{0,0} & a_{0,1} & a_{0,2} & ... \\ a_{1,0} & a_{1,1} & a_{1,2} & ... \\ a_{2,0} & ... & ... & ... \\ a_{n_{rows}-1,0} & ... & ... & a_{n_{rows}-1,n_{cols}-1} \end{array} \right] \]
There are many matrix and vector specializations and some of their basic features are outlined next.
Dynamic size matrices. Use ChMatrixDynamic to create a matrix with generic size, say 12 rows x 4 columns. ChMatrixDynamic is templated by the scalar type, with double
as the default.
Fixed size matrices. Use ChMatrixNM to create matrices that do not need to be resized and whose size is known at compile-time.
When should one use fixed sizes and when should one prefer dynamic sizes? The simple answer is: use fixed sizes for very small sizes where you can, and use dynamic sizes for larger sizes or where you have to. For small sizes, especially for sizes smaller than (roughly) 16, using fixed sizes is hugely beneficial to performance, as it allows Eigen to avoid dynamic memory allocation and to unroll loops.
The limitation of using fixed sizes, of course, is that this is only possible when you know the sizes at compile time. Also, for large enough sizes, say for sizes greater than (roughly) 32, the performance benefit of using fixed sizes becomes negligible. Worse, trying to create a very large matrix using fixed sizes inside a function could result in a stack overflow, since Eigen will try to allocate the array automatically as a local variable, and this is normally done on the stack. Finally, depending on circumstances, Eigen can also be more aggressive trying to vectorize (use SIMD instructions) when dynamic sizes are used.
3x3 fixed size matrices. Use ChMatrix33 to create 3x3 matrices, which are mostly used to represent rotation matrices and 3D inertia tensors.
ChMatrix33 is templated by the scalar type (with double
as the default). This matrix type is derived from a 3x3 fixed-size Eigen matrix with row-major storage and offers several dedicated constructors and methods for coordinate and rotation operations.
Dynamic size column vectors. Use ChVectorDynamic to create a column vector (one-column matrix) with a generic number of rows.
Fixed size column vectors. Use ChVectorN to create a column vector with fixed length (known at compile time).
Row vectors. Use ChRowVectorDynamic and ChRowVectorN to create row vectors (one-row matrices) with dynamic size and fixed size, respectively.
In addition to the above types, specialized 3x4, 4x3, and 4x4 matrices used in multibody formalism are defined in ChMatrixMBD.h.
Basic operations with matrices
Consult the Eigen API for all matrix and vector arithmetic operations, block operations, and linear system solution.
demo_CH_linalg.cpp illustrates basic operations with matrices.
Function objects
These ChFunction objects are used in many places in Chrono::Engine, and are used to represent y=f(x) functions, for example when introducing prescribed displacements in a linear actuator.
These functions are scalar,
\[ x \in \mathbb{R} \rightarrow y \in \mathbb{R} \]
and there are a predefined number of them that are ready to use, such as sine, cosine, constant, etc. If the predefined ones are not enough, the user can implement his custom function by inheriting from the base ChFunction class.
See ChFunction for API details and a list of subclasses.
Example 1
Example 2
Save values of a sine ChFunction into a file.
Example 3
Define a custom function.
The following class will be used as an example of how you can create custom functions based on the ChFunction interface.
There is at least one mandatory member function to implement: GetVal.
Note that the base class implements a default computation of derivatives GetDer() and GetDer2() by using a numerical differentiation, however if you know the analytical expression of derivatives, you can override the base GetDer() and GetDer2() too, for higher precision.
Quadrature
Quadrature is an operation that computes integrals as when computing areas and volumes.
The following code shows how to use the Gauss-Legendre quadrature to compute the integral of a function
\( \mathbb{R} \mapsto \mathbb{R} \) over a 1D interval, or \( f: \mathbb{R}^2 \mapsto \mathbb{R}\) over a 2D interval or \( f: \mathbb{R}^3 \mapsto \mathbb{R}\) over a 3D interval:
\[ F_{1D}=\int^a_b f(x) dx \]
\[ F_{2D}=\int^{a_y}_{b_y}\int^{a_x}_{b_x} f(x,y) dx dy \]
\[ F_{3D}=\int^{a_z}_{b_z}\int^{a_y}_{b_y}\int^{a_x}_{b_x} f(x,y,z) dx dy dz \]
If the function is polynomial of degree N and the quadrature is of order N, the result is exact, otherwise it is approximate (using large N improves quality but remember that this type of integration is often used where N in the range 1..10 suffices, otherwise other integration methods might be better).
For N less than 10, the quadrature uses precomputed coefficients for maximum performance.
Note that thanks to templating, one can also integrate m-dimensional (vectorial, tensorial) functions \( \mathbf{f}: \mathbb{R}^n \mapsto \mathbb{R}^m \) , for example:
\[ \mathbf{F}=\int^{a_y}_{b_y}\int^{a_x}_{b_x} \mathbf{f}(x,y) dx dy \quad \mathbf{F} \in \mathbb{R}^2 \]