Hi,

While playing with this, I got:

> sum(Beta(k,1/2)/(2*k+1)^2,k=1..infinity)
Error, (in SumTools:-DefiniteSum:-ClosedForm) summand is singular in the interval of summation

I don't see how this can be singular, as Beta(k,1/2) is trivially bounded for k>=1, and sum(1/(2*k+1)^2) is absolutely convergent.

Maple is still able to evaluate numerically the inert sum (I get 0.3361376233). If I replace Beta with the expression with GAMMA, I get a result with MeijerG, that doesn't seem to help much for numerical evaluation, as it takes a long time and gives the wrong result -1.966769953*10^10 - (6.886827095*10^8)*I.

I have no reason to expect a closed form for this sum, but the error is surprising. And I don't know MeijerG well enough to explain the numerical error either. It fails with Maple 2020, 2021 and 2022.

The following code completes apparently without error on Maple 2020.2, but with Maple 2021.1 I get a deconnection from the kernel:

restart:
with(LinearAlgebra):
A:=Statistics:-Sample(Normal(0,1),[2500,2500]):
U,S,Vt:=SingularValues(A,output=['U','S','Vt']):
Norm(U.DiagonalMatrix(S).Vt-A);

The crash occurs when running the last line.

The code works on both versions with a smaller matrix (256x256).

Is there something obviously wrong with this piece of code, or a change in Maple 2021 that could explain this?

Note: I'm running Maple on Windows 10, the machine has 16 GB RAM, and the memory usage stays low.

Hi,

In Maple 2020, the following line works fine:

Statistics:-HeatMap(Matrix(128,(i,j)->modp(binomial(i,j),2)),color=["White","Black"]);

With Maple 2021, I get

Interestingly, the exported PNG is better with Maple 2021 (not blurry as in Maple 2020). However, there is an alignment problem with the axes. And it appears in Maple, before exporting. Bug?

(Maple 2020/2021 on Windows 10)

Here is what I'm trying to do. Say I have a Digraph G1 defined by:

with(GraphTheory):
G1:=Digraph([a,b,c],{[[a,b],2],[[b,c],3],[[c,a],4]});

I would like to produce the undirected graph G2, with the same weights:

G2:=Graph([a,b,c],{[{a,b},2],[{b,c},3],[{c,a},4]});

After looking in the GraphTheory package, I found UnderlyingGraph, which seems to do what I want.

Namely,

G3:=UnderlyingGraph(G1,weighted=true);

I had a first problem: there is a bug in the documentation, as the option is 'weights' in the documentation, whereas the source code shows it must be 'weighted'.

But then I had another problem, but maybe I didn't understand the purpose of UnderlyingGraph: apparently, I don't get G2. For instance:

DijkstrasAlgorithm(G1,a);
            [[[a], 0], [[a, b], 2], [[a, b, c], 5]]

DijkstrasAlgorithm(G2,a);
              [[[a], 0], [[a, b], 2], [[a, c], 4]]

DijkstrasAlgorithm(G3,a);
            [[[a], 0], [[a, c, b], 0], [[a, c], 0]]

The problem seems to come from the weight matrix, which is not symmetric (it is for G2):

WeightMatrix(G3);
                           [0  2  0]
                           [       ]
                           [0  0  3]
                           [       ]
                           [4  0  0]

Edges(G3,weights=true);

{[{a, b}, 0], [{a, b}, 2], [{a, c}, 0], [{a, c}, 4], [{b, c}, 0], [{b, c}, 3]}

However, G3 is undirected:

IsDirected(G3);

          false

So, the graph is undirected, but it has different weights for a-b and b-a. Weird.

Now, I am wondering what UnderlyingGraph is supposed to return. After looking at the source code, it seems the statement EW := EW0 + LinearAlgebra:-Transpose(EW0) builds a symmetric weight matrix, but for some reason it's not what is returned.

Is this a bug in the function? Or did I do something wrong? Is there a better way to achieve what I wanted?

Is there a way to tell Maple to compute an integral over a domain defined by an implicit condition?

A trivial example would be: integrate 1 over the 2D domain defined by x^2+y^2<=1.

I know it's possible to write the explicit double integral

int(int(1,y=-sqrt(1-x^2)..sqrt(1-x^2)),x=-1..1);

Or even as a single integral:

integrate(2*sqrt(1-x^2),x=-1..1);

However, it's not always possible to do this explicitly, depending on the form of the relation f(x,y)=0 defining the domain.

My idea was:

int(Heaviside(1-x^2-y^2),[x=-1..1,y=-1..1]);

But Maple returns 0. However, evalf/Int returns the correct numerical value of Pi.