This is a little about mathematics, and a little about writing for the web, but mostly about the nuts and bolts of putting mathematics on the web. I want to record how, mainly with the pandoc
program, I have converted some mathematics from a LaTeX file into html
. Like “Computer Recovery” then, this post is a laboratory notebook.
The mathematics is a proof of Dirichlet’s 1837 theorem on primes in arithmetic progressions. This is the theorem that, if to some number you keep adding a number that is prime to it, there will be no end to the primes that you encounter in this way.
For some reason, I wanted to learn the proof. Maybe this had to do with having given courses on the Prime Number Theorem of 1896 at the Nesin Mathematics Village, but not having been able to teach there this summer, owing to the Covid pandemic. Dirichlet’s theorem could be part of a course at the Village.
I read the proof in Landau’s Elementary Number Theory, originally published in 1927, ninety years after Dedekind’s theorem. I wrote out the argument, according to my understanding. If you want to read what I wrote, there are:

a pdf file (24 pages, size A5) based on the LaTeX file that I composed;

an html file derived from my LaTeX file by means of the
pandoc
program.
Recently I encountered a page of links to over two hundred expository articles by a mathematician. I looked at one article, and I wondered what kind of audience would both

need to be told that an automorphism of a field is a bijective homomorphism from the field to itself,

already know the field of padic numbers.
I might have thought somebody who knew the padic numbers would also know something of Galois theory; but maybe not.
In my own article on Dirichlet’s theorem, I had already given accounts of

what the reader should know,

what I know as an amateur of number theory.
I become increasingly aware of how webpages can be visited from anywhere, although their composers seem unaware of this. If you are a newspaper, what city are you in? If a university, what state or province? If a business, what country? If you are a blog, what are you trying to do, and how can your visitors decide whether to spend time with it?
I have tried to make my own “About” page useful in this way.
I wrote my article on Dirichlet’s theorem, to satisfy my curiosity. Then I remembered posting on this blog my memoir of life on a farm. I had obtained the html
file from a LaTeX file using pandoc
, and I had been pleased with the results. Notably, pandoc
had kept my footnotes as such: they were at the end of the same html
file.
I decided to try to convert the Dirichlet article to html
.
WordPress allows the embedding of TeX and LaTeX code, but converts the code to images. The tex4ht
program does this as well. I wanted to avoid images.
The pandoc
program does not create images, but tries to express mathematics (along with everything else) as text. However, there are mathematical expressions that LaTeX accommodates, but html
does not, or not so well. The pandoc
program leaves those untouched.
I therefore edited my original LaTeX file, to turn all of the mathematics into something that pandoc
could interpret. I tried to make it as easy as possible to switch between a LaTeX file as such—a file to be compiled by the latex
program—and a file to be converted to html
by pandoc
.
Here is what I did.

Apparently
pandoc
can handle the commands that you define, even with arguments; but not default arguments. For example, my\newcommand{\Zmod}[1][k]{\mathbb Z/#1\mathbb Z}
didn’t work until I removed
[k]
. 
I like the
compactitem
andcompactenum
environments of theparalist
package, but apparentlypandoc
does not recognize these, so I switched back toitemize
andenumerate
. 
Fractions are the big challenge. I used
\renewcommand{\frac}[2]{#1/#2}
.One will then need to use parentheses if a numerator or denominator is a sum; but I had this problem in only one case. The reader still has to understand a/bc as meaning a/(bc), although I have allowed (a/b)c to appear as a/b⋅c (this happens when c is a summation with ∑).

Summations themselves are a problem; I rewrote each
\sum_{j=i}^{n}
as\sum_{i\leq j\leq n}
. (One could also define a new command with three arguments here.) 
For one use, I defined:
\DeclareMathOperator{\nCk}{C}
\renewcommand{\binom}[2]{\nCk(#1,#2)}

None of the environments
align
,gather
, andmultline
for displaying more than one line of mathematics together gets interpreted bypandoc
; therefore I put each line to be displayed into its ownequation
environment. 
There is a similar problem with the
cases
environment; I recast usingitemize
. 
Apparently
pandoc
cannot handle a negated symbol such as\not\equiv
, so I used\nequiv
fromtxfonts
. 
I rewrote
\pmod{#1}
as\;(\text{mod }#1)
.
Doing that much gave a file that pandoc
would render as pure html
. However, there were remaining issues.
The program was not dealing properly with the bibliography I had created with BibTeX. Also pandoc
gave my section headings <h1>
tags. I took care of this by running the following command:
pandoc baseheaderlevel=2 bibliography ../../../references.bib filter pandocciteproc dirichletsimple.tex o dirichletsimple.html
The online documentation says baseheaderlevel=
is deprecated, and one should use shiftheadinglevelby=
; but this didn’t work for me. (If pandoc
didn’t come with my Ubuntu Linux installation, I may have installed it when converting the docx
file of somebody else’s philosophy paper to LaTeX.)
At this point, the problem remained that pandoc
did not deal properly with

theorem environments,

labels of equations.
I found discussion of these on the Google group for pandoc
. I have not understood why they should be a problem, or at least why references should be a problem.
For example, pandoc
prints the label of a theorem as plain text. It could print the label of an equation in the same way, but apparently it doesn’t.
One can apparently customize pandoc
, but for now I don’t know how. Therefore I have done the following (probably not something to be done on a regular basis):

formatted and numbered by hand my theorems and lemmas;

formatted the proofs by hand, inserting
$\Box$
at the end if this is text, and\qquad\Box
if it’s an equation (pandoc
could not deal with\qed
); 
changed
\label{#1}
(which I habitually place just after\begin{equation}
) to\mylabel{#1}
, defined as(#1)\qquad
; 
redefined
\eqref{#1}
as(#1)
; 
changed all of my equation labels to the desired serial numbers;

changed every
{equation}
to{equation*}
.
After all of this, I cleaned up the html
file created by pandoc
by:

putting blank lines between paragraphs (for ease in reading the
html
file); 
changing
<span class = "math display">
to<div style= "textalign:center;">
; 
putting
<div style = "textalign:justify; marginleft:10%; marginright:10%;">
at the head, and</div>
at the foot, as with all of my blog posts and pages.
Now I am using pandoc
to create the present html
file from the plain text file that I originally typed.
One Comment
Most useful — thanks!
5 Trackbacks
[…] I did not try to define compactness in the last post. Perhaps one does not normally learn about this concept until one has spent some time with calculus. Calculus is the practical side of what has the theoretical side called analysis. From high school I have Apostol’s Mathematical Analysis (second edition, Reading, Mass.: AddisonWesley, 1974); we used selections from it with Mr Brown to learn about uniform convergence, a concept needed also for the proof of Dirichlet’s Theorem that I discussed in “LaTeX to HTML.” […]
[…] the blog page called, and about, “Primes in Arithmetic Progressions”; that page was the subject of the post called “LaTeX to HTML.” […]
[…] A proof of Dirichlet’s theorem on primes in arithmetic progressions, in pdf and html format (as discussed in “LaTeX to HTML”) […]
[…] The underlying html file is now created by pandoc from a text file, edited from the one that I obtained by applying pandoc to the old html file. I talk about pandoc in “LaTeX to HTML.” […]
[…] edit online with the WordPress editor. My current practice, initiated with the composition of “LaTeX to HTML” three months ago, is to edit a text (txt) file, using “Pandoc’s enhanced version of […]