@Kitonum 

I tried your method, but unfortunately, it gives wrong results on number of cases. I tried both Splitting and Splitting1. Actually all methods shown in all the answers, do not all work on all cases. This is not an easy problem as it looks.

expr:=-y+4*x^3*y;
result := Splitting(expr,x,y);

It should be y*(4 x^3-1)

And

expr:=exp(x^2)+2*x*y;
r:=Splitting1(expr); 

The above can't be split.

@Carl Love 

" The syntax A:-B searches module A's exports for something named B. On the other hand, the B without a module prefix is just accessing a local "

Ah, but for newbies, it is not clear that A:-B means to search for only the exports of A. I just thought it mean just a naming convention to make it more clear which B is it, weather it is exported or not.  Also did not know if using B with prefix implies local

Good to know. I only been using modules for few days.

But this brings another follow up question.

What if parent has a local proc named foo(), and the child also has a local proc named foo(). Then how will the child call the parent local foo() proc if it wants (in addition to calling its own foo() when it wants) if one is not allowed to use the syntax parent:-foo() to tell Maple which foo() they want to call and without being forced to export the parent foo()? Here is an example

parent_module:=module()
 
 local child_module;
 local foo;

 export parent_entry;

 child_module:= module()
    export child_entry; 
    local foo;

    child_entry :=proc()
     foo(); #How to call the parent foo()??
    end proc;

    foo :=proc()
     print("in child foo");
    end proc;

  end module;

 #local
 foo :=proc()
   print("insider parent foo");
 end proc;

 #public
 parent_entry :=proc()
   child_module:-child_entry();
 end proc;
end module;

I can make the above a seperate question if needed.

thanks

@Preben Alsholm 

Thanks for both answers, Both works.

But I do not like to put the absolute path of child packages inside the $include, since I have to change this later if I move the folder somewhere else or if I want to send this package to someone to use?

So now I put all .mpl files in same folder and use relative path and add $include "child_package.mpl". 

But before loading the main package, I have to set the path, which is OK, since this is done outside the package.

@one man 

on my Maple nops gives an error. How come it worked for you? Which version do you have?

I am using Maple 2018. 

thanks.

@Carl Love 

I have not touched or changed anything. It was OK earlier today. I could do computation. It is only the menu items and help that was causing this error.

It seems something was causing it. The good news, is that I just rebooted the whole PC and now this error is gone !  I can start help OK now.

Something was causing it which I do not know what. But rebooting the PC fixed it.

I could delete this question in this case, or leave if it is of any it could happen for someone else or if it happens again for me. This is first time I've seen this problem with Maple 2018

@acer 

I think I either found a bug in the above function F, or I do not understand what it does when the input is not separable.

I gave it an expression which is not separable. It returns product of two expressions. But this is not separable.   How is one supposed to know if the returned value is the separation/split or may be it it same as the input written in different way? Here is an example

expr := -(y-exp(x))/(x-2);
F (expr, [x,y] );

I also found an expression which it can't separate

expr := 1-x+y^2-x*y^2;
#this can be split to (1-x)*(1+y^2)
F( expr, [x,y] );

And this one

expr:=(1-x^2+y^2-x^2*y^2)/(x^2);
#this can be split to 
fx:=(1-x^2)/x^2;
fy:=(1+y^2);
simplify(expr-(fx*fy));
F( expr, [x,y] );

This is not an easy problem!. I think Maple should have build-in function to do this.

I do not understand your code too well. Too advanced for me to change it. It will help if your function will return a list of the 2 split product, when it can separate them, or 2 empty lists if it can't separate them. This way a user would more easily check, like this

fx,fy := F(expr, [x,y]);

if fx={} or fy={} then
   print("can't separate them");
else
   print("separated them into two functions", fx, fy);
fi;

Thanks

@acer 

I used LibraryTools:-Save as you recommend. I was just asking if there was a different example you refered to. The whole documenation is really confusing in Maple. They do not spell things out very well for newbies. 

For example., in LibraryTools:-Save('name'), it does not say that it will save the contant of "name.mpl" to "name.mla" by default. I found this by trial and error. 

I ended doing this

restart;
read "my_package.mpl"; 
        #note that now Maple currentdir() is automatically 
        #set to where the worksheet is located. This seems to be
        #new in Maple 2018, which is good change.

my_package_name := cat( currentdir(),kernelopts(dirsep),"my_package.mla");

LibraryTools:-Create(my_package_name); 
          #this will create file my_package.mla 
          #in same folder as this worksheet

LibraryTools:-Save('my_package'); 
             #this will save content of "my_package.mpl" 
             #in "my_package.mla" file. This step was a 
             #little confusing. As name passed refers to
             #both the input (.mpl) and the (.mla) with same 

restart;
with(my_package);   #now this will work

#to update the package, after making changes to .mpl, do
restart;
read "my_package.mpl";
LibraryTools:-Save('my_package');

restart;
with(my_package);   #now this load the updated package

@acer 

Use LibraryTools:-Create to make a new .mla file. You only need to do that once.

Use LibraryTools:-Save to store the modules or procedure within that new archive file. You only need to do this after you make any change to the definitions.

(Do Not use the savelib command for the second step, of storing the pieces in the archive. It is too awkward and it's use error-prone for non-experts. Really.)

There's an example of all this in the Programming Manual. 

Would you know where in the programming manual this example is? Do you mean the programming guide?

I am now looking at the programming manual, 2018, downloaded from 

https://www.maplesoft.com/documentation_center/maple2018/ProgrammingGuide.pdf

And it uses savelib. not LibraryTools:-Save (but it uses LibraryTools:-Create) which is strange. On page 394:

Then on next page

@Carl Love 

I need to construct a symbolic expression, inside a local proc(). So one should make all the symbols that are part of the expresion local, so not to use a symbol which could be assigned globally.

Now, then I need to pass this expression to another proc() to do pattern matching. 

But I just found an easy workaround. I will pass the "x" and "C1" symbols in the call. That is all.

restart;
foo:=proc()
  local x,C1;
  boo(exp(C1+x),x,C1);
end proc;

boo:=proc(expr,x,C1)
  local a,b,c,la;
  patmatch(expr,a::anything*exp(b::anything*C1+c::anything),'la');
end proc:

And now it works.

@acer 

thanks. Your new version work. 

I myself do not like the inline `if`. I've seen such constructs used before, but I find it hard to read, since I like to see the explicit IF THEN ELSE, which makes it easier for me to see the logic. So I rewrote yours as follows

restart;
expr  := y(x)^2+diff(y(x),x,x)+g(y(x))+diff(y(x),x)^2+Int(f(x),x)+y(x)+y(x)*y(x)+y;
check := proc(expr::anything)
   local t,__y;
   t := subs(y(x)=__y(x),expr);
   
   if indets(t,{identical(y),'specfunc'(y)}) = {} then
      "OK";
   else
      "NOT OK";
   fi;

end proc:

check(expr);

I also added local __y;  I noticed you replaced the y(x) in the expression by __y(x). But there is always a chance that __y could have global value. Not likely, but possible.

Do you know if this is how dsolve does its parsing, or where to find how dsolve parses the input? I looked at dsolve using  showstat(:-dsolve); but hard for me to see where the parsing/validating of the input is done. I am sure dsolve parsing of input will be hundreds of lines long, as it has to check for so many variations.  I just do not know where it does that. 

But the above now works for me. 

Thanks for help.

@Fabio92 

thanks to all the answers. Yes, this does what I want. This is just for development. I first open the worksheet where the definitions are, then next open the testing worksheet.

Later, when finish development, I can do the library method. I am newbie and do not know how to do library stuff yet. So I work in worksheets now.

@acer 

"Using `has` to attempt this kind of thing, is not The Way, IMO"

You are right. I found that "has" fails on this

expr:=y(x)+g(y(x));
 

restart;
expr:=y(x)+g(y(x));
candidates:=indets(expr);
candidates:=select(has,candidates,y);
reject:=remove(u->[op(u)]=[x], candidates);
if nops(reject)<>0 then
   print("Not ok input");
else
   print("Ok input");
fi;

The above is wrong. The input  g(y(x)) is valid. Your method works on the above 

expr:=y(x)+g(y(x));
candidates:= indets(expr, 'specfunc'(y));
reject:=remove(u->[op(u)]=[x], candidates);
if nops(reject)<>0 then
   print("Not ok input");
else
   print("Ok input");
fi;

Only problem as I mentioned  before with indets is that it fails if input has "y" on its own. 

thanks.

is every one really supposed to know what HPM is? According to wiki

HPM may refer to:

• Harbarian process modeling
• Hargeisa Provincial Museum, in Hargeisa, Somalia
• Haystack Prayer Meeting, held in Williamstown, Massachusetts, in 1806
• High-power microwave, directed-energy weapon using electromagnetic radiation
• high performance mobile, a semiconductor production type parametric variant of TSMC
• Hospice and Palliative Medicine, a medical specialty that focuses on symptom management, relief of suffering and end-of-life care
• Human performance modeling
• Human Potential Movement, self-improvement movement of the 1960s
• Hydrogen-moderated self-regulating nuclear power module, a type of nuclear power generator
• A high proper motion star
•  

Which one of the above are you looking for Maple code for?

@acer 

Thanks. But your solution seems to drop the "y".  If I understand, I will check the removed stuff, and see if it is empty or not. If the removed stuff is empty, then the original input was valid. Correct?  But for example, when I do this

restart;
expr:=y(x)+y;
candidates:=indets(expr, 'specfunc'(y));
reject:=remove(u->[op(u)]=[x], candidates);
if nops(reject)<>0 then
   print("Not ok input");
else
   print("Ok input");
fi;

It prints "Ok input", this is becuase candidates has only y(x) in it. So it missed the "y" there without (x), which is not valid input.

But replacing 

     candidates:=indets(expr, 'specfunc'(y));

by

candidates:=indets(expr);
candidates:=select(has,candidates,y);

fixed things:

restart;
expr:=y(x)+y;
candidates:=indets(expr);
candidates:=select(has,candidates,y);
reject:=remove(u->[op(u)]=[x], candidates);
if nops(reject)<>0 then
   print("Not ok input");
else
   print("Ok input");
fi;

@vv 

But e.g. dsolve cannot solve the simple:

That is not the point actually. This is just a parsing phase. I needed to make sure the input is valid syntax. Not that it can be solved or not.