One good idea is for you to provide code the can be used to reproduce the problem.

You could also attach a worksheet to your Question. Platform details might help.

Form the Help page for Topic solve,ineq

   In general, variables and parameters will automatically be assumed
   to be real-valued in the context of one or more inequalities.

@TechnicalSupport This item is not really about incorporating functionality.

This is about fixing a regression in quite usual functionality. In Maple 2015.2 the following works:

   plots:-matrixplot(<<1,2,3>|<3,4,7>|<4,5,8>>, heights=histogram, color=red);

But in Maple 2016.2 (and on to Maple 2020.1) the data contructed in the COLOR substructure is invalid (the number of entries are wrong, and some of the floats are indexed...)

The primary and central fix should be to restore the proper construction of the COLOR substructure. The fix should not consist primarily of applying options as a final step (eg. using plots:-display, say with _rest, and with or without overrideoption.)

If anyone's interested: the GUI sometimes renders 3D plots as all black when the color data it receives from the kernel is invalid. The CommandLine Interface (CLI) throws an error when running this problematic example in a terminal shell with the default plot-device.

@Anthrazit Your image shows the Help page of the substring command.

Your Post was (only) about the SubString command from the StringTools package. The Help page for that command states,

    All of the StringTools package commands treat strings as (null-terminated)
    sequences of 8-bit (ASCII) characters.  Thus, there is no support for multibyte
    character encodings, such as unicode encodings.

@Anthrazit One of the paragraphs specifically mentions that the command (and the StringTools package) only supports 8-bit ASCII interpretation.

Have you read and understood the Help page for that command?

Is the X-Y grid evenly spaced in both of those directions?

@Ali Hassani Thank you. I'm glad it serves your purpose.

That 3rd procedure is a little hackish, going through strings. It could be done more carefully.

@Joe Riel Incredible.

@Carl Love This is mostly right.

It is not true that libname can only contain references to directories -- it may also contain references to .mla files.

Also you stated, "To make it fully automatic as you describe, you must update the value of libname in your initialization file."  But that is not the only way to do it. A "toolbox" location also works, eg. with "foobar" a name of one's choice,
   cat(kernelopts('homedir'),"/maple/toolbox/foobar/lib")
also works. Or, version specific,
   cat(kernelopts('homedir'),"/maple/toolbox/2020/foobar/lib")
Such locations are automatically added to libname upon launch.

@dharr Yes, it's not clear why the OP wants atomic names for sub/superscripted expressions. My guess it that it might be to avoid any collision with the actual expression, but the OP really ought to clarify.

The third procedure in my Answer automates all generation of the atomic name which renders in 2D Output like the input expression, with minimal concatenation.

I agree with you that it's generally better to automate construction than to build up a complicated piece of MathML (or TypeMK, in undocumented Maple parlance). Natural exceptions are for cases where the 2D rendering doesn't actually match any valid Maple input expression, or when one wants special color/font/size.

Depending on one's view, one drawback here is that numerals are rendered in italics, which makes then look different from the usual expressions.  Eg,   x[5]^7  when normally typeset.

@MapleEnthusiast Alas, you have not actually answered my query as to what further computation you expect to do with this explicitly computed Matrix inverse.

@MapleEnthusiast What is the purpose of computing this inverse explicitly? For what particular further computation will it get used?

(I ask because this approach may be relatively poor -- some tasks like linear solving or least squares solving or eigen-calculation may be done better and generally more stably, numerically, with other, specialized techniques.)

Why don't you demonstrate why you think that the result is meaningless?

Have you noticed that the condition number of your Identity-A is about 8.3e9 ?