Document mathematical Go code beautifully.
- Write mathematical formulae in a LaTeX-ish syntax
- Super/subscripts formatted with Unicode characters:
2^32
becomes2³²
andx_{i+1}
becomesxᵢ₊₁
- Comprehensive symbol library:
\zeta(s) = \sum 1/n^{s}
becomesζ(s) = ∑ 1/nˢ
Inspired by Filippo Valsorda's literate Go
implementation of
Poly1305, which can be reproduced using mathfmt
.
Install mathfmt
with:
go get -u github.com/mmcloughlin/mathfmt
Apply to files just like you would with gofmt
.
mathfmt -w file.go
Here's our variance function in Go, documented with LaTeX-ish equations in comments.
// Variance computes the population variance of the population x_{i} of size N.
// Specifically, it computes \sigma^2 where
//
// \sigma^2 = \sum (x_{i} - \mu)^2 / N
//
// See also: https://en.wikipedia.org/wiki/Variance.
func Variance(X []float64) float64 {
// Compute the average \mu.
mu := Mean(X)
// Compute the sum \sum (x_{i} - \mu)^2.
ss := 0.0
for _, x := range X {
ss += (x - mu) * (x - mu) // (x_{i} - \mu)^2
}
// Final divide by N to produce \sigma^2.
return ss / float64(len(X))
}
Run it through mathfmt
and voila!
// Variance computes the population variance of the population xᵢ of size N.
// Specifically, it computes σ² where
//
// σ² = ∑ (xᵢ - μ)² / N
//
// See also: https://en.wikipedia.org/wiki/Variance.
func Variance(X []float64) float64 {
// Compute the average μ.
mu := Mean(X)
// Compute the sum ∑ (xᵢ - μ)².
ss := 0.0
for _, x := range X {
ss += (x - mu) * (x - mu) // (xᵢ - μ)²
}
// Final divide by N to produce σ².
return ss / float64(len(X))
}
First a warning: mathfmt
does not have a rigorous grammar, it's a
combination of string replacement and regular expressions that appears to
work most of time. However you may run into some thorny edge
cases.
mathfmt
only works on Go source code. Every comment in the
file is processed, both single- and multi-line.
mathfmt
recognizes a huge symbol table that is
almost entirely borrowed from LaTeX packages. Every symbol macro in comment
text will be replaced with its corresponding Unicode character. In addition
to LaTeX symbol macros, mathfmt
supports a limited set of
"aliases" for character combinations commonly used to
represent mathematical symbols.
Like LaTeX, superscripts use the ^
character and subscripts use _
. If the
super/subscript consists entirely of digits, then no braces are required: for
example 2^128
or x_13
. Otherwise braces must be used to surround the
super/subscript, for example 2^{i}
or x_{i+j}
.
Note that Unicode support for super/subscripts is limited, and in particular
does not support the full alphabet. Therefore, if there is not a
corresponding super/subscript character available for any character in braces
{...}
, mathfmt
will not perform any substition at all. For example there
is no superscript q
, so mathfmt
will not be able to process 2^{q}
, and
likewise with x_{K}
.
Thank you to Günter Milde for the exhaustive unimathsymbols
database of Unicode symbols
and corresponding LaTeX math mode commands.
mathfmt
is available under the BSD 3-Clause License.