@awass It worked fine here with 2015.1 and 2015.2.

@london-algebra Alas, I don't think that solves the problem; I used the wrong expression (the first one).  With the second, is returns FAIL, which is what what verify does. The help page for the greater_equal verification states that it calls signum, which, in turn, calls is.  So the limitation is currently inherent in Maple.

That doesn't sound good.  What's the size of the file?  

@YasH That error indicates Maple cannot find the maplev.mla archive, which is supposed to be installed at

FileTools:-JoinPath([kernelopts(homedir), "maple", "toolbox", "emacs", "lib"])

It's probably safe to ignore the error, but it would be better to install the mla file there.

@Mac Dude Thanks for the help.  As you suggest, the mla is not needed for the basic operation of maplev-mode.  It has some Maple code for displaying Maple library code in a buffer (M-?), though there might be a fallback if the mla is missing (its been a while since I've touched that code).

@sigl1982 The github distribution is the source.  The maplev.info file can be built from the maplev.texi file, assuming you have makeinfo and friends. 

@sigl1982 Where'd you get the maplev from?  I'll have to add the maplev.info file if it is indeed missing. I'm attaching it here: maplev_info.zip.  Did you start maplev-mode with a Maple source file loaded?  Have you added any configuration in your .emacs file to setup maplev-mode (i.e. associate it with particular file extensions, etc.). There should be a README file with the distribution showing a simple setup.

@sigl1982 In Maple, execute kernelopts(homedir); that will tell you where Maple thinks your home directory is.  The mla should go relative to that (with the ~ standing for the home directory). I think Maple will automatically find that if the installed there, however, on Windows I'm never sure.

@Carl Love I've frequently wished that assign would return the value assigned. Possibly it wasn't designed that way because it can make multiple assignments so it isn't clear in that case whether it should return the first, the last, or all the values as a sequence. The latter requires a small bit of memory, to store the sequence in the simplification table.

@taro A simpler method than the custom Frontend suggested in the original post is to use print as the first argument to see what expressions are frozen. Tweak the third argument until the proper expressions are frozen, then replace the print with the procedure of choice.

@Carl Love It seems to be deleting quickly now.

It would be nice if the spam button appeared at the top of a post rather than below it. For long posts it is necessary to scroll down to get to the spam button.

@mnpjf That can be done with a discrete sampler, see Signal Blocks > Discrete.

@Carl Love That seems the ideal solution.  I knew there was such a method in Maple, but didn't think to look in Statistics.  In thinking about how to make this more efficient (pointless, I know) one could avoid generating the intermediate list by using a double loop on the set of sets.  I don't expect it to be significant, or faster, but maybe use a bit less memory (the subsets are already in the simplification table, so accessing them is cheap):

cnt := table('sparse'):
for s in g do
   for i in s do
       cnt[i] := cnt[i]+1;
   end do;
end do:

@Kitonum Thanks.  I first came up with that technique about twenty years ago, replying to a question on usenet.  It's somewhat analogous to using generating functions, in that the results are the coefficients of a polynomial.

Could you explain what you really want to do?  I don't see how the question relates to Maple and Maple worksheets.