I got this solution from a PDE. I normally use unapply on the RHS of the solution to make it a function.

But in this, the PDE solution contain some extra stuff at the end. Which  "Where { .....}"

So the only way for me, was to manually copy the initial part of the solution shown on the screen in order to use it later.

I could not find a way to program this part.

Here is an example

restart;

f:=(r,z)->(r-a)*sin(z/H*Pi);
lap:=VectorCalculus:-Laplacian(u(r, z, t), cylindrical[r, theta,z]);
bc:=u(r,0,t)=0,u(r,H,t)=0, u(a,z,t)=0;
ic:=u(r,z,0) = f(r,z);
sol:=pdsolve([diff(u(r,z,t),t) = k*lap,bc,ic],u(r,z,t)) assuming a>0,k>0;

THis gives

Which I verified to be correct.

In case you are not able to get this solution (it needs Maple 2019.1 and Physics 368), here is the lprint

lprint(sol)

u(r,z,t) = `casesplit/ans`(Sum(-BesselJ(0,lambda[n]/a*r)*sin(z/H*Pi)*exp(-k*t*(
H^2*lambda[n]^2+Pi^2*a^2)/a^2/H^2)*a/lambda[n]^2*Pi*BesselJ(1,lambda[n])*
StruveH(0,lambda[n])/hypergeom([1/2],[1, 2],-lambda[n]^2),n = 1 .. infinity),{
And(lambda[n] = BesselJZeros(0,n),0 < lambda[n])})

Next to use it (plot., evaluate, etc...) changed the sum go to 15 terms (more than enough) and also replaced a->1,H->3,k->1/100 and  also replaced lambda[n] with BesselJ zeros as follows

sol:=subs({infinity=15,a=1,H=3,k=1/100,lambda[n]=BesselJZeros(0, n)},sol);

Before I use unapply, I had to extract the part of the solution up to where it says "where..." since I do not need the rest any more, since I allready replaced lambda[n] with BesselJZeros calls.

This I did by hand using copy/paste from the screen. Now I am able to finish the task:

solFiltered:=Sum(-BesselJ(0, BesselJZeros(0, n)*r)*sin(z*Pi/3)*
     exp(-t*(9*BesselJZeros(0, n)^2 + Pi^2)/900)*Pi*
     BesselJ(1, BesselJZeros(0, n))*
     StruveH(0, BesselJZeros(0, n))/(BesselJZeros(0, n)^2*
     hypergeom([1/2], [1, 2], -BesselJZeros(0, n)^2)), n = 1 .. 15):

mapleSol:=unapply(solFiltered,r,z,t);

value(mapleSol(.5,2,1));

Which prints -0.4107565091 which is the correct value. It matches my hand solution and also match numerical solution.

What would a better way to do the above than what I did? i.e. to obtain the solFitered above using code only? 

I installed Physics package 368 and when I run a test, I found that now the latex() command fails with internal  error which I have never seen before. I think this is first time I see latex() gives an error.

Download bug_june_6_2019.mw

Is this a new bug? Why latex() command now fails? It failed on the above output. When I try latex() on simpler output. no error is generated. So something to do with the above specific output seems to be the issue.

Maple 2019.1 using Physics 368 on windows 10.

This is a new error I have not seen before.  

Trying to verify my solution for the diffusion pde in cylinderical coordinates wth no angle theta dependency.

unassign('z,t,r,u');
lap:=diff(u(r,z,t),r$2)+ 1/r*diff(u(r,z,t),r)+diff(u(r,z,t),z$2);
bc:=u(r,0,t)=0,u(r,1,t)=0, u(1,z,t)=0;
ic:=u(r,z,0) = f(r,z);
pdsolve([diff(u(r,z,t),t) = lap,bc,ic],u(r,z,t)) assuming t>0

The error is

Error, (in assuming) when calling 'BesselJZeros'. Received: '0th zero of BesselJ(0,x) not defined'

Is this a bug or Am I doing something wrong? 

update

fyi, it also fails with same error when trying boundedseries HINT

restart;
unassign('z,t,r,u');
lap:=VectorCalculus:-Laplacian(u(r, z, t), cylindrical[r, theta,z]);
bc:=u(r,0,t)=0,u(r,1,t)=0, u(1,z,t)=0;
ic:=u(r,z,0) = f(r,z);
pdsolve([diff(u(r,z,t),t) = lap,bc,ic],u(r,z,t),HINT=boundedseries(r=0))

I am using Maple 2019.1 with Physics 366 on windows 10

Maple 2019.1 with Physics version 362  gives this strange error on this pde

restart;
pde := x*diff(u(x, y), x) + y*diff(u(x, y), y) = -4*x*y*u(x, y);
ic := u(x, 0) = exp(-x);
sol:=pdsolve([pde,ic],u(x,y));

Error is

Error, (in PDEtools:-casesplit) equation of unknown type integer : 1

Is this a bug? It shows only when using exp(-x). Changing it to exp(-x^2) or exp(x) do not show the error, even though Maple can't solve it.

On windows 10.

I was trying to verify some solution to pde in textbook using Maple. The book gives the Cauchy data for this first order PDE in the form that Maple does not like when I used it as input. Here is an example

When I typed

pde:=u(x,y)*(x+y)*diff(u(x,y),x)+u(x,y)*(x-y)*diff(u(x,y),y)=x^2+y^2;
ic:=u(x,2*x)=0;
pdsolve([pde,ic],u(x,y))

Maple complained

Error, (in PDEtools:-Library:-NormalizeBoundaryConditions) unexpected 
occurrence of the variables {x} in the 2nd operand of u(x, 2*x) in the given initial conditions

In Mathematica it accepts such form of Cauchy data:

pde=u[x,y]*(x+y)*D[u[x,y],x]+u[x,y]*(x-y)*D[u[x,y],y]==x^2+y^2;
ic=u[x,2*x]==0;
DSolve[{pde,ic},u[x,y],{x,y}]

(I have not verified the above answer is correct or not).

Am I doing something wrong in Maple?

Or is there a trick or option or method to allow Maple to accepts such initial conditions? The book I am looking at has many problems where Cauch data is given on such form (i.e. u=0 on specific curve or in 3D on some specific surface). Here is another example

Which I'd like write its initial conditions as u(x,1/x)=0 but can not.

I could ofcourse solve the pde without these initial conditions, and then post process the answer to find the constants of integration from the Cauchy data given. But it will be nice if Maple would accept the IC as is.