Rendering of small equations with labels has layout issues

[jlperla]

The jupinx master seems to have flaws in the display with equations. Create something like

.. math::
    :label: test

    v = p + \beta v

And the generated HTML is

<a id='equation-test'></a>
<table width=100%><tr style='background-color: #FFFFFF !important;'>
<td width=10%></td>
<td width=80%>
$$
v = p + \beta v
$$
</td><td width=10% style='text-align:center !important;'>
(1)
</td></tr></table>

which wrapped the equation in a weird way because of the decrease in the width of the table

cc: @mmcky any thoughts on easy workaround?

[mmcky]

thanks @jlperla. I will add your example to our test set and check it out. Perhaps our css is altering the look of labeled equations and not working correctly on default nbconvert

[mmcky]

I also like you're idea of using latex \tag{#}

[mmcky]

• [x] add to test case math

@jlperla if you get the chance can you post the output. The test case packaged with the extension has a labelled equation that is a bit smaller than the original equation (without labels) but I don't know what you mean with strange wrapping.

[mmcky]

I have added the above test case to the test notebook math.rst and it renders like this:

[mmcky]

is your issue at the notebook level or post nbconvert to html?

[jlperla]

Notebooks. I am not sure how to generate html yet!

[mmcky]

are you getting rendering that is different to the screenshot above?

[jlperla]

The above is a rendered from the following rst in Jupytelab (generated using ubuntu on WSL, not that it matters, and displaying in chrome)

Fixed-Point Maps
----------------------------

Consider the simple equation, where the scalars :math:`p,\beta` are given, and  :math:`v` is the scalar we wish to solve for

.. math::
    :label: vfixed

    v = p + \beta v

Of course, in this simple example, with parameter restrictions this can be solved as :math:`v = p/(1 - \beta)`

Rearrange the equation in terms of a map :math:`f(x) : \mathbb R \to \mathbb R`

.. math::

    v = f(v)

where

.. math::

    f(v) := p + \beta v

Therefore, a fixed point :math:`v^*` of :math:`f(\cdot)` is a solution to the above problem

[mmcky]

Hmm. Thanks @jlperla. I can't replicate it on my local. Are you by chance using sphinx >= 1.8? I haven't tested on the newly released 1.8

[jlperla]

Key version numbers from conda list are

sphinx                    1.7.4                    py36_0
sphinxcontrib             1.0              py36h6d0f590_1
sphinxcontrib-bibtex      0.4.0                     <pip>
sphinxcontrib-jupyter     0.2b0.dev20180929           <pip>
sphinxcontrib-jupyter     0.2b0.dev20180827           <pip>
sphinxcontrib-websupport  1.0.1            py36hb5cb234_1

But I think this is probably a html/css problem, as opposed to a generation problem

[mmcky]

I'm using sphinx=1.7.8 but as you say I don't think that is the issue (as the html output is the same). Could a local extension be altering the alignment of equations. My environment has not jupyter extensions installed with no specific html/css adjustment to base jupyter.

[jlperla]

My jupyter has no extensions, but perhaps it is old. The versions are

jupyter                   1.0.0                    py36_4
jupyter_client            5.2.3                    py36_0
jupyter_console           5.2.0            py36he59e554_1
jupyter_core              4.4.0            py36h7c827e3_0
jupyterlab                0.32.1                   py36_0
jupyterlab_launcher       0.10.5                   py36_0

I will to upgrade jupyterlab tomorrow and see what happens

[mmcky]

I mainly just jupyter notebook rather than jupyterlab ... I will test with jupyterlab on my end also. Unfortunately jupyter notebook and jupyterlab don't always produce the same output: https://github.com/QuantEcon/sphinxcontrib-jupyter/issues/79

[jlperla]

I think that most jupyterhub installations are moving quickly to jupyterlab, so it might make sense to test there as the primary interface. But, if this is a jupyter vs. jupyterhub issue, I think it also points out that there is only so much you can do to deal with different versions of jupyterhub

[bayeswhu]

@mmcky Is there any method to enlarge the labelled equations? As you mentioned, they are a little bit smaller than the equations without labels. Especially when I tried to present the notebook in slides (with RISE), I found it is hard to read.

[arnavs]

Hi @mmcky, just wanted to chime in that I can reproduce this issue on https://compare-lectures.quantecon.org/jl/jupinx_test.html

<table width=100%><tr style='background-color: #FFFFFF !important;'>
<td width=10%></td>
<td width=80%>
$$
v = f(v)
$$
</td><td width=10% style='text-align:center !important;'>

</td></tr></table><p>And want to link to it in text <a href="julia_by_example.html#equation-fixed-point-map">(1)</a></p>

[mmcky]

we could wrap the tables in a class and adopt a local mathjax resize in our theme for now.

https://stackoverflow.com/questions/19086152/changing-mathjaxs-font-size

[mmcky]

A fix for the current solution has been implemented in https://github.com/QuantEcon/sphinxcontrib-jupyter/pull/152

[jlperla]

Was this solution used in the latest generation of the notebooks? I just downloaded a notebook geenrated from the mccall lecture and in jupyterlab it looks like

On the other hand, to look at the julia examples file, the HTML is a little odd in that the numbered equations are larger than the other ones:

and the of numbered equations are still an issue with weird splitting

[mmcky]

these issues should be fixed by https://github.com/QuantEcon/sphinxcontrib-jupyter/pull/154 when merged.

[mmcky]

154 is merged so closing this issue.