Autodiff is a header-only C++ library that facilitates the automatic differentiation (forward mode) of mathematical functions of single and multiple variables.
This implementation is based upon the Taylor series expansion of an analytic function f at the point x₀:
The essential idea of autodiff is the replacement of numbers with polynomials in the evaluation of f. By inputting the first-order polynomial x₀+ε, the resulting polynomial in ε contains the function's derivatives within the coefficients. Each coefficient is equal to a derivative of its respective order, divided by the factorial of the order.
Assume one is interested in calculating the first N derivatives of f at x₀. Then without any loss of precision to the calculation of the derivatives, all terms O(εN+1) that include powers of ε greater than N can be discarded, and under these truncation rules, f provides a polynomial-to-polynomial transformation:
C++'s ability to overload operators and functions allows for the creation of a class fvar
that represents
polynomials in ε. Thus the same algorithm that calculates the numeric value of y₀=f(x₀) is also used to
calculate the polynomial Ʃnynεⁿ=f(x₀+ε). The derivatives are then found from the
product of the respective factorial and coefficient:
#include <boost/math/differentiation/autodiff.hpp>
#include <iostream>
template<typename T>
T fourth_power(T x)
{
x *= x;
return x *= x;
}
int main()
{
using namespace boost::math::differentiation;
constexpr int Order=5; // The highest order derivative to be calculated.
const autodiff_fvar<double,Order> x = make_fvar<double,Order>(2.0); // Find derivatives at x=2.
const autodiff_fvar<double,Order> y = fourth_power(x);
for (int i=0 ; i<=Order ; ++i)
std::cout << "y.derivative("<<i<<") = " << y.derivative(i) << std::endl;
return 0;
}
/*
Output:
y.derivative(0) = 16
y.derivative(1) = 32
y.derivative(2) = 48
y.derivative(3) = 48
y.derivative(4) = 24
y.derivative(5) = 0
*/
The above calculates
#include <boost/math/differentiation/autodiff.hpp>
#include <boost/multiprecision/cpp_bin_float.hpp>
#include <iostream>
template<typename T>
T f(const T& w, const T& x, const T& y, const T& z)
{
using namespace std;
return exp(w*sin(x*log(y)/z) + sqrt(w*z/(x*y))) + w*w/tan(z);
}
int main()
{
using cpp_bin_float_50 = boost::multiprecision::cpp_bin_float_50;
using namespace boost::math::differentiation;
constexpr int Nw=3; // Max order of derivative to calculate for w
constexpr int Nx=2; // Max order of derivative to calculate for x
constexpr int Ny=4; // Max order of derivative to calculate for y
constexpr int Nz=3; // Max order of derivative to calculate for z
using var = autodiff_fvar<cpp_bin_float_50,Nw,Nx,Ny,Nz>;
const var w = make_fvar<cpp_bin_float_50,Nw>(11);
const var x = make_fvar<cpp_bin_float_50,0,Nx>(12);
const var y = make_fvar<cpp_bin_float_50,0,0,Ny>(13);
const var z = make_fvar<cpp_bin_float_50,0,0,0,Nz>(14);
const var v = f(w,x,y,z);
// Calculated from Mathematica symbolic differentiation.
const cpp_bin_float_50 answer("1976.319600747797717779881875290418720908121189218755");
std::cout << std::setprecision(std::numeric_limits<cpp_bin_float_50>::digits10)
<< "mathematica : " << answer << '\n'
<< "autodiff : " << v.derivative(Nw,Nx,Ny,Nz) << '\n'
<< "relative error: " << std::setprecision(3) << (v.derivative(Nw,Nx,Ny,Nz)/answer-1)
<< std::endl;
return 0;
}
/*
Output:
mathematica : 1976.3196007477977177798818752904187209081211892188
autodiff : 1976.3196007477977177798818752904187209081211892188
relative error: 2.67e-50
*/
Additional details are in the autodiff manual.