Warning
This package is currently unmaintained and these capabilities are available in various other Julia packages. If you would like to help maintain this package (since it is a good teaching tool), please chime in on #153.
The Calculus package provides tools for working with the basic calculus operations of differentiation and integration. You can use the Calculus package to produce approximate derivatives by several forms of finite differencing or to produce exact derivative using symbolic differentiation. You can also compute definite integrals by different numerical methods.
Most users will want to work with a limited set of basic functions:
derivative(): Use this for functions from R to Rsecond_derivative(): Use this for functions from R to RCalculus.gradient(): Use this for functions from R^n to Rhessian(): Use this for functions from R^n to Rdifferentiate(): Use this to perform symbolic differentiationsimplify(): Use this to perform symbolic simplificationdeparse(): Use this to get usual infix representation of expressions
There are a few basic approaches to using the Calculus package:
- Use finite-differencing to evaluate the derivative at a specific point
- Use higher-order functions to create new functions that evaluate derivatives
- Use symbolic differentiation to produce exact derivatives for simple functions
using Calculus
# Compare with cos(0.0)
derivative(sin, 0.0)
# Compare with cos(1.0)
derivative(sin, 1.0)
# Compare with cos(pi)
derivative(sin, float(pi))
# Compare with [cos(0.0), -sin(0.0)]
Calculus.gradient(x -> sin(x[1]) + cos(x[2]), [0.0, 0.0])
# Compare with [cos(1.0), -sin(1.0)]
Calculus.gradient(x -> sin(x[1]) + cos(x[2]), [1.0, 1.0])
# Compare with [cos(pi), -sin(pi)]
Calculus.gradient(x -> sin(x[1]) + cos(x[2]), [float64(pi), float64(pi)])
# Compare with -sin(0.0)
second_derivative(sin, 0.0)
# Compare with -sin(1.0)
second_derivative(sin, 1.0)
# Compare with -sin(pi)
second_derivative(sin, float64(pi))
# Compare with [-sin(0.0) 0.0; 0.0 -cos(0.0)]
hessian(x -> sin(x[1]) + cos(x[2]), [0.0, 0.0])
# Compare with [-sin(1.0) 0.0; 0.0 -cos(1.0)]
hessian(x -> sin(x[1]) + cos(x[2]), [1.0, 1.0])
# Compare with [-sin(pi) 0.0; 0.0 -cos(pi)]
hessian(x -> sin(x[1]) + cos(x[2]), [float64(pi), float64(pi)])
using Calculus
g1 = derivative(sin)
g1(0.0)
g1(1.0)
g1(pi)
g2 = Calculus.gradient(x -> sin(x[1]) + cos(x[2]))
g2([0.0, 0.0])
g2([1.0, 1.0])
g2([pi, pi])
h1 = second_derivative(sin)
h1(0.0)
h1(1.0)
h1(pi)
h2 = hessian(x -> sin(x[1]) + cos(x[2]))
h2([0.0, 0.0])
h2([1.0, 1.0])
h2([pi, pi])
using Calculus
differentiate("cos(x) + sin(x) + exp(-x) * cos(x)", :x)
differentiate("cos(x) + sin(y) + exp(-x) * cos(y)", [:x, :y])
The Calculus package no longer provides routines for univariate numerical integration. Use QuadGK.jl instead.
Calculus.jl is built on contributions from:
- John Myles White
- Tim Holy
- Andreas Noack Jensen
- Nathaniel Daw
- Blake Johnson
- Avik Sengupta
- Miles Lubin
And draws inspiration and ideas from:
- Mark Schmidt
- Jonas Rauch