Canonical Form

Many mathematical objects can be represented by several equivalent expressions.

For example, the expressions in each row below represent the same mathematical object:

The Compute Engine stores expressions internally in a canonical form to simplify the implementation of some algorithms.

The value of expr.simplify(), expr.evaluate() and expr.N() are canonical expressions.

The ce.box() and ce.parse() functions return a canonical expression by default, which is the desirable behavior in most cases.

To get a non-canonical version of an experssion set the canonical option of ce.parse() or ce.box() to false.

The non-canonical version will be closer to the literal LaTeX input, which may be desirable to compare a “raw” user input with an expected answer.

ce.parse('\\frac{3}{-5}');
// ➔ ["Rational", -3, 5]
// The canonical version moves the sign to the numerator

ce.parse('\\frac{3}{-5}', { canonical: false });
// ➔ ["Divide", 3, -5]
// The non-canonical version does not change the arguments, so this is
// interpreted as a regular fraction ("Divide"), not a rational.

The value of expr.json may not be strictly in canonical form: some “sugaring” is applied to the internal representation before being returned, for example ["Add", -1, "x"] may be returned as ["Subtract", "x ", 1].

You can further customize how an expression is interpreted by using ce.jsonSerializationOptions.

ce.parse('\\frac{3}{5}', { canonical: false });
// ➔ ["Rational", 3, 5]
// This is a rational without modifying the arguments, so a `["Rational"]`
// expression is returned

ce.jsonSerializationOptions = { exclude: ['Rational'] };
ce.parse('\\frac{3}{5}', { canonical: false });
// ➔ ["Divide", 3, 5]
// We've excluded `["Rational"]` expressions, so it is interepreted as a
// division instead.

The canonical form of an expression is always the same when used with a given Compute Engine instance. However, do not rely on the canonical form as future versions of the Compute Engine could provide a different result.

To obtain the canonical representation of an non-canonical expression, use the expr.canonical property.

console.log(ce.box(['Add', 2, 'x', 3]).canonical);
// ➔ ["Add", 5, "x"]

To check if an expression is canonical use expr.isCanonical.

If the expression is already canonical, expr.canonical immediately returns expr.

Canonical Form Transformations

The canonical form used by the Compute Engine follows common conventions. It is not always “the simplest” way to represent an expression.

Calculating the canonical form of an expression is applying some rewriting rules to an expression to put sums, products, numbers, roots, etc… in canonical form. In that sense, it is similar to simplifying an expression with expr.simplify(), but it is more conservative in the transformations it applies, and it will not take into account any assumptions about symbols or their value.

Below is a list of some of the transformations applied to obtain the canonical form:

• Idempotency: \( f(f(x)) \to f(x) \)

• Involution: \( f(f(x)) \to x \)

• Associativity: \( f(a, f(b, c)) \to f(a, b, c) \)

• Literals

  • Rationals are reduced, e.g. \[(\frac{6}{4} \longrightarrow \frac{3}{2}\]
  • The denominator of rationals is made positive, e.g. \[(\frac{5}{-11} \longrightarrow \frac{-5}{11}\]
  • A rational with a denominator of 1 is replaced with a number, e.g. \[(\frac{19}{1} \longrightarrow 19\]
  • Square roots of rationals have their perfect squared factored out, e.g. \[(\sqrt{63} \longrightarrow 3\sqrt{7}\]
  • Complex numbers with no imaginary component are replaced with a real number

• Abs

  • The absolute value of literals is evaluated

• Add

  • Arguments are sorted
  • Literal 0 is removed
  • Sum of a literal and the product of a literal with the imaginary unit are replaced with a complex number.

• Multiply

  • Arguments are sorted
  • $x \times x$ is replaced with ["Square", x]
  • The product of two integers literals is evaluated But not exact literals? I.e. rationals or Square Root of fractional?
    • If any argument is NaN or Undefined evaluates to NaN Might be too aggressive

• Divide

• Power

  • $x^{\tilde\infty} \longrightarrow \operatorname{NaN}$
  • $x^0 \longrightarrow 1$
  • $x^1 \longrightarrow x$
  • $(\pm 1)^{-1} \longrightarrow -1$
  • $(\pm\infty)^{-1} \longrightarrow 0$
  • $0^{\infty} \longrightarrow \tilde\infty$
  • $(\pm 1)^{\pm \infty} \longrightarrow \operatorname{NaN}$
  • $\infty^{\infty} \longrightarrow \infty$
  • $\infty^{-\infty} \longrightarrow 0$
  • $(-\infty)^{\pm \infty} \longrightarrow \operatorname{NaN}$

• Square

• Sqrt

• Root

• Subtract

• Negate

• For Multiply, literal 1 is removed, small integers and small rations are multiplied together.

• For Divide, replaced by Multiply / Power

• For Subtract, replaced by Add

• For Sqrt and Root, replaced by Power