[IPOL discuss] [Fwd: IPOL & MathJax]

[eml]

Dear all,

Since serious mathematics sites such as MathOverflow are using MathJax,
it makes sense to at least consider it.

Pros:
* The rendered formulas are beautiful
* The HTML code is beautiful [1]
* The LaTeX form of any formula can be trivially recovered, and it is
  stored only once.

Cons:
* Relies heavily on javascript
* It is much slower than embedding a png image
* It will require some work to be implemented into IPOL, overriding
  previous design decisions.




[1] Compare the current sources of the same paragraph, using
HTML+png or MathJax:



-- begin HTML+png version --

<p>Previous works correct the distortion by minimizing the RMS distance
for all the <em>2D</em> projected points to the &#34;a priori&#34;
known <em>N</em> straight lines (rects), rect of index <em>l</em>
having <em>N<sub>l</sub></em> points. This minimization can be carried
out through any optimization method (gradient-like). For these cases,
the energy function is as follows:</p> 
<p></p><center> 
 <img
alt="D(\mathbf{k})=\frac{1}{N}\sum_{l=1}^{N}\frac{1}{N_{l}}\sum_{i=1}^{N_{l}}%
\frac{\left(  a_{l}{x}^{*}_{l,i}+b_{l}{y}^{*}_{l,i}+c_{l}\right)
^{2}}% {a_{l}^{2}+b_{l}^{2}} ," class="teximg"
src="./2d29dadaec238afd36ba479bd02abcf4.png" /> </center> 
 
<p>where (<em>x<sup>*</sup> <sub>l,i</sub></em> , <em>y<sup>*</sup> 
<sub>l,i</sub></em>) , are the undistorted points using the distortion
model provided by <strong>k</strong>, and where <em>a<sub>l</sub></em>
<em>x<sup>*</sup> + b<sub>l</sub> y<sup>*</sup>  + c<sub>l</sub></em>
is the line that minimizes <em>D(<strong>k</strong>)</em> for a given
choice of points { (<em>x<sup>*</sup>
<sub>l,i</sub></em>,<em>y<sup>*</sup> <sub>l,i</sub></em>) }
<sub>{<em>i=1,....,N<sub>l</sub></em>}</sub>.</p> 

-- end HTML+png version --




-- begin Mathjax version --

<p>Previous works correct the distortion by minimizing the RMS distance
for all the $2D$ projected points to the &#34;a priori&#34;
known $N$ straight lines (rects), rect of index $l$
having $N_l$ points. This minimization can be carried
out through any optimization method (gradient-like). For these cases,
the energy function is as follows:

\[
 D(\mathbf{k})=\frac{1}{N}\sum_{l=1}^{N}\frac{1}{N_{l}}\sum_{i=1}^{N_{l}}%
\frac{\left(  a_{l}{x}^{*}_{l,i}+b_{l}{y}^{*}_{l,i}+c_{l}\right)
^{2}}% {a_{l}^{2}+b_{l}^{2}}
\]
 
<p>where $(x^*_{l,i},y^*_{l,i})$, are the
undistorted points using the distortion model provided by $k$,
and where $a_lx^*+b_ly^*+c_l=0$ is the line that minimizes
$D(\mathbf{k})$  for a given choice of
points $\{(x^*_{l,i},y^*_{l,i})\}_{i=1,\dots,N}$.</p> 

-- end Mathjax version --