I just had enough with odetest hanging (even when using with timelimit). I wait hours and hours each time (even though I have 30 second timelimit, which Maple ignores) and I do not think Maplesoft is going to fix this in my lifetime.  

So I am attempting to make my own very simple and basic odetest.

I give it an ode and explicit solution in for form sol:= y(x)=... and the function uses algsubs(sol,ode) and checks it is gets zero or not (it will also do simplify if needed)

But there is a BIG problem.   Even though algsubs(z=0,z/z);  gives back 1 as expected,  but

restart;
ode:=diff(y(x), x)/y(x);
sol:=y(x)=0;

algsubs(sol,ode);   #this gives ZERO. It should be 1

If we do algsubs on each term one by one

algsubs(sol,numer(ode));
algsubs(sol,denom(ode));

            0
            0

So why did algsubs give zero in the first case, since  the result of the algsubs should be 0/0 which algsubs knows in the limit it is 1?   How did it come up with zero?

Clearly my simple method of replacing odetest with algsubs is not working. I need a more robust way to handle this.

subs does not work. Since subs does not know how substitute y(x)=f(x) into derivatives involved in an ODE.

My question is: Is there a way to teach algsubs to give 1 for the above example? or better function to use?

I tried applyrule instead of algsubs, but that does not work.

applyrule(sol, ode)

Error, (in rec) numeric exception: division by zero
 

Is there a better method to use? I am trying to do simple version of odetest that does not hang. Even if not perfect. Will only use it for explicit solutions, not implicit since implicit is much harder.

Here is a more full example

ode:=diff(y(x),x)^2+2*x*diff(y(x),x)/y(x)-1 = 0;
sol:=y(x)=0;
odetest(sol,ode)

                            0

But when I use my simple method

ode:=diff(y(x),x)^2+2*x*diff(y(x),x)/y(x)-1 = 0;
sol:=y(x)=0;
algsubs(sol,ode);

                          -1 = 0

The reason it failed, because algsubs replaced the second term by 0 instead of 1. The second term in the ode is diff(y(x),x)/y(x)

Which is 0/0 but this is 1 in the limit. But algsubs used 0 instead for some reason.

So I need a little bit smarter way to replace my solution into the ode than just using algsubs. May need to use some of the tricks I've seen used here before using freez/thaw/frontend, etc.. which I still do not understand.

Any recommendation?  This is meant to work for any single ode and any explicit solution of the form y(x)=....

This algsubs method works actually pretty well on many ode's. I've tested it on 2,000 ode's. It just fails so far on subtle ones like the above. Here is just random example where it works

restart;
ode:=diff(y(x),x)^2 = (-x+1)/x;
sol:=y(x)=_C1+arcsin(2*x-1)/2+sqrt(x-x^2);

evalb(simplify(algsubs(sol,ode)))

                     true

So I just need a way to handle the cases where it gives 0/0 I think. I have 20 ODE's which now fail out of about 2,000 using this basic method compared to using odetest.

Thank you

I was trying to simplify long expression of the form    f(x)=g(x), where f(x) should simplify to be the same as g(x) and get 0=0. But Maple simplify just return the input back.

Then I wrote simplify(f(x)-g(x)) and now Maple returned 0 back.

What is the difference between simplify (f(x)=g(x)) and simplify(f(x)-g(x))? And why the second worked but not the first?

restart;
f:=(9*(x^(-2/3*a))^2*exp(6/a*(x^(-2/3*a))^(1/2))^2*_C0^2-6*(x^(-2/3*a))^(3/2)*exp(
6/a*(x^(-2/3*a))^(1/2))^2*_C0^2*a+x^(-2/3*a)*exp(6/a*(x^(-2/3*a))^(1/2))^2*_C0^
2*a^2+18*(x^(-2/3*a))^2*exp(6/a*(x^(-2/3*a))^(1/2))*_C0-2*x^(-2/3*a)*exp(6/a*(x
^(-2/3*a))^(1/2))*_C0*a^2+6*(x^(-2/3*a))^(3/2)*a+x^(-2/3*a)*a^2+9*(x^(-2/3*a))^
2)/(3*_C0*exp(6/a*(x^(-2/3*a))^(1/2))*(x^(-2/3*a))^(1/2)-exp(6/a*(x^(-2/3*a))^(
1/2))*_C0*a+3*(x^(-2/3*a))^(1/2)+a)^2:

g:=x^(-2/3*a):

And now

simplify(f=g);

But

simplify(f-g);

                                                           0

And

simplify(f-g=0);

                 0 = 0 

Why Maple behaves like this? I did not know it makes difference if one writes f=g vs. f-g in terms of simplification. 

I copied the code to Mathematica to see how it behaves, and Mathematica Simplify worked on both f=g and also on f-g as one would have expected:

Maple 2020.2 on windows

For the last hr, I've been trying to find out why Maple solve was hanging inside one module in my .mla when called.

I would try the same code in new worksheet, no hang. I had timelimit on it. Same exact code as in the module.

I go back to the other worksheet, do restart, call the module (after loading the library by setting libname) which calls solve(), and it hangs. Each time.

I tried may be 10 times. No difference between the code in the mla and the other worksheet where I had copy of the solve command. All other settings are the same. 

ONLY after I closed Maple, and restarted Maple, and opened the worksheet where the call used to hang, and called the module again, now it did not hang and solve completed.

This is not the first time I've seen such a strange thing. Sometimes Maple will hang for hrs (I start  a test, go to sleep,. wake up hrs later, and see maple hanged for hrs from the time stamp).

Closes Maple, start the test from where it was hanging, and now no hang. It runs OK from then on.

This tells me that doing just restart from the worksheet do not clean everything. (I use the command restart, not by pressing the restart button in the menu, not sure if these are different, they should be the same  I would think).

I've seen this problem before. sometimes the test just hangs. No matter how many time I do restart from the worksheet, it just hangs. When I restart Maple, it works from same place it was hanging.

My question is, what could possibly make Maple not hang when starting fresh. vs. using the restart command?  I am using the same worksheet, and I have "create new engine for each document" selected in options.

The hang used to happen in solve, in here

#this is inside a function inside a module inside .mla library
#

eq:=x = 1/2/(p^2+2)^(1/8)*2^(1/2)*((1+(-4*p^2+1)^(1/2))/(1/(4*p^2-1)*p^2)^(1/2)/(-4
*p^2+1)^(1/2))^(3/4)/(((-4*p^2+1)^(1/2)+3)/(p^2+2)^(1/2))^(1/4)*_C1*p^(1/4);

        try               
            sol_p:=timelimit(20,[:-solve(x=rhs(eq),p)]);
        catch:
           return [];
        end try;
#hangs here. Never completes. Only when starting Maple the hang is gone. 

#trace below
......
Main: polynomial system split into 1 parts under preprocessing
Main: using RegularChains based methods
SolverVariableOrder: using the variable order  _S000008 > _S000011 > _S000009 > _S000006 > _S000005
TriangularDecomposition: using deterministic algorithm for decomposition
<<<< HANGS HERE EACH TIME >>>>

Again, only when I restarted Maple, did the hang go away. it is one equation solving for one variable.

So now, I get into the habit of closing all of Maple and starting it again when I get a hang. I do not trust restart command any more to clear everything.

Any idea why this happens? Memory not cleared somewhere? Cache problem? Does the frontend itself stores something that could cause this? does restart use new mserver.exe each time or same mserver.exe process as last time?

if restart does not actually terminate the mserver.exe that the worksheet was talking to, but only sends request to mserver.exe to clear its own memory, this could explain things.

It is possible that mserver.exe which is attached to that worksheet does not fully clear all its own memory and something remain there which affects why it hangs in same place each time, even though restart was issued each time, and only when restarting Maple itself will clear this and the hangs goes away.

When asking dsolve to use specific ode type to solve an ode, in particular, dAlembert type, which typically generate complicated solutions, sometimes dsolve solution shows up using parameter T.

But odetest gets confused by this expression it seems. I am not able to figure if I am doing something wrong in using odetest, or may be odetest does not know how to handle such form of a solution.

Here is an example. This ode

is of these types:

restart;
ode:=diff(y(x), x) = (y(x)^3 + 2*x*y(x)^2 + x^2*y(x) + x^3)/(x*(y(x) + x)^2);
DEtools:-odeadvisor(ode);

By default, dsolve was smart to use homogeneous type to solve the ODE, as this gives the simplest solution. 

One can force dsolve to use the other types. When using dAlembert, odetest gives an error trying to verify any one of the solutions returned from dsolve due to the way the solution is returned. Here is the result

restart;
ode:=diff(y(x), x) = (y(x)^3 + 2*x*y(x)^2 + x^2*y(x) + x^3)/(x*(y(x) + x)^2);
sol:=[dsolve(ode,y(x),[homogeneous])];

No problem here for odetest. it can verify any of the above 3 solutions with no error generated.

odetest(sol[1],ode)

          0

Lets compare using dAlembert type

restart;
ode:=diff(y(x), x) = (y(x)^3 + 2*x*y(x)^2 + x^2*y(x) + x^3)/(x*(y(x) + x)^2);
sol:=[dsolve(ode,y(x),[dAlembert])]: #solution too complicated to show here
odetest(sol[1],ode)

It looks like odetest does not know how to handle the form of the solution as returned by dsolve for this case. The problem is that each solution is actually made up of two parts, not just y(x) as normally is the case. One part defines something called X(T_) and the next part which is the solution y(x) uses this X(T) in it. 

Did I do something wrong, or is there a way around this, or is this by design?

Maple 2020.2

restart;
f:=(x,y)-> x*y^3-x^2/y^3;
eq:= subs([x=alpha*x,y=alpha^p*y],f(x,y));
lis:=[]:
for tmp in op(eq) do
    if has(tmp,alpha) then
       lis:=[op(lis),op(2,simplify(select(has,tmp,alpha)))];
    fi;
od;
lis;
PDEtools:-Solve(lis[1]=lis[2],p)

This method happend to work for this example, but it does not work in general. This is because the function can be anything. Here is a second example

restart;
f:=(x,y)-> -2/3*x+1/3*(x^2+3*y)^(1/2);
eq:= subs([x=alpha*x,y=alpha^p*y],f(x,y));

We see by inspection if p=2 then we can write the above as, by setting p=2 as 

But the Maple code I had above does not work on this

restart;
f:=(x,y)-> -2/3*x+1/3*(x^2+3*y)^(1/2);
eq:= subs([x=alpha*x,y=alpha^p*y],f(x,y));
lis:=[]:
for tmp in op(eq) do
    if has(tmp,alpha) then
       lis:=[op(lis),simplify(select(has,tmp,alpha))];
    fi;
od;
lis;

The problem is finding all powers of alpha in each term and setting up an equation to find p such that all terms have same numerical value.

So I abandoned this method as too messy to program (it works well for hand solution, this is an example where solving something by hand is easier than on the computer) and then tried solve directly, like this (on the first example)

restart;
f:=(x,y)-> x*y^3-x^2/y^3;
eq:= subs([x=alpha*x,y=alpha^p*y],f(x,y))- alpha^r*f(x,y)=0;

And here is where I am stuck. I need to ask Maple if it can find alpha, r that are rational numbers, such that the above equation is solved.

PDEtools:-Solve(eq,[p,r])

But we know from above that p=1/6 and r=3/2 is solution

simplify(subs([p=1/6,r=3/2],eq))

             0=0

The question is: Can Maple solve such an equation for  p,r? I remember now reading something about parameteric solver in Maple. I need to look that up to see if it helps. I tried SolveTools:-SemiAlgebraic on this, but it did not help.

Are there better methods to determine in Maple if function is isobaric and to find the index p and r?

Reference: book

Notice, some places define isobaric function as one in which we can find p such that

But this is not correct. I've seen it on 2 pages on the net. The correct definition is 

Also note: When p=1 isobaric function becomes a homogeneous function which is special case of isobaric and r is now called the degree of homogeneity 

But for isobaric, p do not have to be 1. This is the main difference.

f(x,y) will always be a function of x,y.  I will add more examples of isobaric functions to test against.

edit. Simplified question is restated below

After looking more into this. I found I actually wanted to solve the simpler problem, which is

So please ignore the general case of isobaric where r<>p-1 which is harder. 

I only need to look at this case where r=p-1 which is actually simpler. It turned out this is what I need for my solver and not the general case.

I will now give my solution to this and show where the problem I am having.

I am using Maple solve to find p. So no r any more. Only solving for one variable. But Maple solve fails sometimes. And this is the problem.

The problem is that solve can find p for many cases, but fails on some, where there is clearly a solution. I will show 5 examples below.

#example 1. WORKS
restart;
f:=(x,y)-> 3*sqrt(x*y);
eq:= subs([x=alpha*x,y=alpha^p*y],f(x,y))- alpha^(p-1)*f(x,y)=0;
PDEtools:-Solve(eq,p) assuming alpha>0;

#example 2. WORKS
restart;
f:=(x,y)-> 4*(x*y)^(1/3);
eq:= subs([x=alpha*x,y=alpha^p*y],f(x,y))- alpha^(p-1)*f(x,y)=0;
PDEtools:-Solve(eq,p) assuming alpha>0;

#example 3. WORKS
restart;
f:=(x,y)-> (-3*x^2*y-y^2)/(2*x^3+3*x*y);
eq:= subs([x=alpha*x,y=alpha^p*y],f(x,y))- alpha^(p-1)*f(x,y)=0;
PDEtools:-Solve(eq,p) assuming alpha>0;

#example 4 WORKS
restart;
f:=(x,y)-> (-(x*y)/2+sqrt(x^2*y^2-4*y)/2)*y;
eq:= subs([x=alpha*x,y=alpha^p*y],f(x,y))- alpha^(p-1)*f(x,y)=0;
PDEtools:-Solve(eq,p) assuming alpha>0;

And here is the one that does not work

#Example 5. Does not work. How to make it work?
restart;
f:=(x,y)-> -2*x/3+sqrt(x^2+3*y)/3;
eq:= subs([x=alpha*x,y=alpha^p*y],f(x,y))- alpha^(p-1)*f(x,y)=0;
PDEtools:-Solve(eq,p) assuming alpha>0;

There is a solution for this. p=2 works. Why solve did not find it? I mean, it gives RootOf, which does not help me decide if this is isobaric or not. Why it did not find the solution p=2?

simplify(subs(p=2,eq)) assuming alpha>0

                 0=0

Using Mathemtica, it found p=2 solution. How to make Maple find this solution is my question.

Sorry for the long question. I first thought I needed the general isobaric case, but found later i needed the simpler one.