Getting Maple on the Raspberry Pi would require a port of the Maple kernel to the ARM CPU architecture. If such a port were accomplished then IMNSHO there'd be much more interesting platforms than just the Pi.

I don't understand how your comment about contracts makes any sense.

Why do you fail to provide your code that reproduces the problem?

@Mac Dude I think that Christian's suggestion is the same as what I showed by example. That is, custom tickmarks of the integers value converted to strings.

Have you tried  Cmd =

By that I mean Cmd and = together, with no Enter or Return.

Do you have a so-called international keyboard layout, for which the usual keyboard shortcuts don't work? In that case I'd ask whether Alt Enter itself worked.

Your comment makes it sound as if you believe that the local solver found the better point by virtue of its doing a gradient search. That is not what happens, as far as I've seen (and demonstrated in part). What happens is that the local Optimization gradient method fails outright due to the extreme flatness at the supplied initial point. It just happens to then return a better value found at a boundary point for this particular problem -- but for a similar problem it might well return the inferior result.

The GlobalSolve procedure does call the local Optimization solver at some point. Indeed it can do so in several ways (including a gradient search from whatever value the global solver returns). It happens that it does not utilize the approach of passing up front the original initial point to the local solver and utilizing whatever it returns. That would be an improvement that would manifest itself for some examples, but then so also would be the rather ad hoc fashion of checking any arbitary number of local solver calls with various boundary or other initital values. (The number of such possible attempts is indeed ad hoc for other problems in general.)

It happens that the default global search options are slightly too conservative for this particular problem. The use of GlobalSolve options which allows this problem's better value to be found is hardly a major stumbling block. They may be set to low for a wide variety of problems. Changing that would be a much better general change, in my opinion.

Sorry that I don't have an OS X 10.14.x machine to test on. Perhaps someone else can check on Mojave. But I obtain this using an earlier OS X (and Linux).

Here's what I obtain on Linux, using Maple 2019.1,

kernelopts(version);

     Maple 2019.1, X86 64 LINUX, May 21 2019, Build ID 1399874

assume(a>-1,b>0);
additionally(a<=1);
about(a);
Originally a, renamed a~:
  is assumed to be: RealRange(Open(-1),1)

restart;
assume(tau<1,tau>0,s<1,s>0):
a_e1:=tau*s*(1+tau)<tau*s+tau+s-1:
b_e2:=expand(lhs(a_e1)-rhs(a_e1))<0:
b_e3:=collect(b_e2,s,factor):
solve(b_e2,s) assuming tau<1;

                       1
                  [{-------- < s~}]
                    1 + tau~

I get the same results using Maple 2019.1 on an OS X machine with the following specs (which I believe is an OS X Yosemite release),

$ sw_vers
ProductName:	Mac OS X
ProductVersion:	10.10.3
BuildVersion:	14D136

Are you trying to obtain this plot?

   plot(x-g(x)*exp(-mu*x), x = 0 .. 20);

If not then please explain what you mean by stating that this is a Matrix:  x - u = x - M.exp(-mu*x) 

Those kind of wild values in the Matrix look similar to what can happen when the Maple kernel thinks that the installed Maple license is not valid.

There have been a small number of reports in which a valid license (eg. Maple 2019.1 or below) may get incorrectly recognized. It's possible that yours is such a case.

You could contact Maplesoft's Technical Support for advice.

Have you considered uploading a worksheet that reproduces your input to solve, as an attachment on your original or a Reply/Comment?

ps. I changed your Post to a Question.

@Carl Love Thanks for the reminder. I've re-attached it as a .zip file.

@Rouben Rostamian  

Was not 2008 the last year for the US penny reverse image that you've used? From 2010 to date the "Union Shield" design is used.

Fwiw, you can also use the Import command to pull in the images.

(Sorry, my exported .gif doesn't look as sharp as it does in the GUI. I didn't have patience to wait on the Mapleprimes uploader.)

Download penny_2013_S.mw

[edit] Image files could also be attached to a Maple workbook, so that they can be re-used programmatically without need for auxiliary files. penny_2013.zip  (I have zipped the .maple file since Mapleprimes is broken with respect to downloading them as attachments. Thanks for the reminder, Carl.)

I see that Carl had already come up with the same thing, while I was fiddling.

If I recall correctly then, no, the only way to export a dualaxisplot to jpeg in Maple 12 is through the right-click context-menu. I don't think that it can be done programmatically (ie. via plotsetup and plotdevice).

Are you looking for the lprint command?

If that is not the case then could you reformulate your query in the form of a sentence?

If you want each re-execution to be preserved automatically then you could use a DataTable embedded component. That has the benefit that it stores data across a restart.

You could even write a procedure that computed the final result in question, augmented the embedded component, and then printed/returned  the current result. Or any other bells and whistles you want.

Perhaps you could be much more specific about how you'd like it to behave.