Algebra (symbolic computation) #

math.js has built-in support for symbolic computation (CAS). It can parse expressions into an expression tree and do algebraic operations like simplification and derivation on the tree.

It’s worth mentioning an excellent extension on math.js here: mathsteps, a step-by-step math solver library that is focused on pedagogy (how best to teach). The math problems it focuses on are pre-algebra and algebra problems involving simplifying expressions.

Simplify #

The function math.simplify simplifies an expression tree:

// simplify an expression
console.log(math.simplify('3 + 2 / 4').toString())              // '7 / 2'
console.log(math.simplify('2x + 3x').toString())                // '5 * x'
console.log(math.simplify('x^2 + x + 3 + x^2').toString())      // '2 * x ^ 2 + x + 3'
console.log(math.simplify('x * y * -x / (x ^ 2)').toString())   // '-y'

The function accepts either a string or an expression tree (Node) as input, and outputs a simplified expression tree (Node). This node tree can be transformed and evaluated as described in detail on the page Expression trees.

// work with an expression tree, evaluate results
const f = math.parse('2x + x')
const simplified = math.simplify(f)
console.log(simplified.toString())       // '3 * x'
console.log(simplified.evaluate({x: 4})) // 12

Note that simplify has an optional argument scope that allows the definitions of variables in the expression (as numeric values, or as further expressions) to be specified and used in the simplification, e.g. continuing the previous example,

console.log(math.simplify(f, {x: 4}).toString()) // 12
console.log(math.simplify(f, {x: math.parse('y+z')}).toString()) // '3*(y+z)'

In general, simplification is an inherently dfficult problem; in fact, for certain classes of expressions and algebraic equivalences, it is undecidable whether a given expression is equivalent to zero. Moreover, simplification generally depends on the properties of the operations involved; since multiplication (for example) may have different properties (e.g., it might or might not be commutative) depending on the domain under consideration, different simplifications might be appropriate.

As a result, simplify() has an additional optional argument, options, which controls its behavior. This argument is an object specifying any of various properties concerning the simplification process. See the detailed documentation for a complete list, but currently the two most important properties are as follows. Note that the options argument may only be specified if the scope is as well.

Note that the default context is very permissive (allows a lot of simplifications) but that there is also a math.simplify.realContext that only allows simplifications that are guaranteed to preserve the value of the expression on all real numbers:

const rational = math.parse('(x-1)*x/(x-1)')
console.log(math.simplify(expr, {}, {context: math.simplify.realContext})
  // '(x-1)*x/(x-1)'; canceling the 'x-1' makes the expression defined at 1

For more details on the theory of expression simplification, see:

Derivative #

The function math.derivative finds the symbolic derivative of an expression:

// calculate a derivative
console.log(math.derivative('2x^2 + 3x + 4', 'x').toString())   // '4 * x + 3'
console.log(math.derivative('sin(2x)', 'x').toString())         // '2 * cos(2 * x)'

Similar to the function math.simplify, math.derivative accepts either a string or an expression tree (Node) as input, and outputs a simplified expression tree (Node).

// work with an expression tree, evaluate results
const h = math.parse('x^2 + x')
const x = math.parse('x')
const dh = math.derivative(h, x)
console.log(dh.toString())        // '2 * x + 1'
console.log(dh.evaluate({x: 3}))  // '7'

The rules used by math.derivative can be found on Wikipedia:

Rationalize #

The function math.transform transforms a rationalizable expression in a rational fraction. If rational fraction is one variable polynomial then converts the numerator and denominator in canonical form, with decreasing exponents, returning the coefficients of numerator.


math.rationalize('2x/y - y/(x+1)')
              // (2*x^2-y^2+2*x)/(x*y+y)
math.rationalize('(2x+1)^6')
              // 64*x^6+192*x^5+240*x^4+160*x^3+60*x^2+12*x+1
math.rationalize('2x/( (2x-1) / (3x+2) ) - 5x/ ( (3x+4) / (2x^2-5) ) + 3')
              // -20*x^4+28*x^3+104*x^2+6*x-12)/(6*x^2+5*x-4)

math.rationalize('x+x+x+y',{y:1}) // 3*x+1
math.rationalize('x+x+x+y',{})    // 3*x+y

const ret = math.rationalize('x+x+x+y',{},true)
              // ret.expression=3*x+y, ret.variables = ["x","y"]
const ret = math.rationalize('-2+5x^2',{},true)
              // ret.expression=5*x^2-2, ret.variables = ["x"], ret.coefficients=[-2,0,5]
Fork me on GitHub