Markdown parsing occurs before katex parsing and causes error

[PaulEcoffet]

Hello, I want to use the markdown parsing with katex, and I noticed that for some formula involving underscores, the markdown parsing occurs before the katex transformation and alter the result.

I think it would be clearer with an example

Markdown:

$\sum_{j=0}^{M} \sum_{i=0}^{N} (\hat{k}_{i}^{j} - k_{i}^{j})$

output from ngx-markdown with katex activated:

$\sum_{j=0}^{M} \sum_{i=0}^{N} (\hat{k}<em>{i}^{j} - k</em>{i}^{j})$

Notice the <em> tags that appears. If I remove the curly braces which are optional in this context, the formula is rendered correctly. It seems that some underscores are considered as at the beginning or at the end of words and thus the markdown parser replace them with emphases.

Thank you for you help