I expect, that 'limit' should use provided assumptions. But it is not so much work to process it before.

For evalf it should do better (without assumptions).

But for students that is a good example: why to study if hitting an enter would be enough? Do they even 'learn' how to make simple tests as cross checks - for example looking at specific values, plotting some stuff or remember a bit of Math lectures? Ahem ...

1. Using MultiSeries gives it

MultiSeries:-limit(exp(n)/(-1)^n, n = infinity) assuming n::posint; evalf(%);

                              undefined
                           Float(undefined)

2. Unsatisfactory, yes. But doing it in steps gives it

sin(Pi/2+2*Pi*n); % assuming n::posint;

                                      1

Andriy, 'simplify' combines it into one exp ...

For your other (?) example - series - we can have a look at it in a new thread, if needed.

In html it looks like an accent, ASCII 180, but it is ASCII 146

It would be better to say which book, what is c etc For your code: it seems to have non-printable (=hidden) characters (copy paste error or what ever)

Hm ... it is an affine linear subspace, intersected with a hypercube (over Z). Is there no other way than exhaustive search?

PS: if you avoid saving results and compile (without much care, since integers are small in that task) that  will be much faster, roughly 5 sec to count that ~ 2 Mio solutions.

You better use the variables, not 'indets'.

For the matrix: just use 'simplify' (which I de-activated in my posting).

The strange result RootOf(Im(_Z)) in (1,2) and (2,1) comes in through the
equation 2*I*Im(m12)=0; solve(%); and that is Maple's way to say Im(m12=0).
Or in plain words: m12 has to real.

Compare

indets(M);
             /                                 ___\ 
            { m11, m12, m22, Re(m11), Re(m22), m12 }
             \                                    / 

vars:=indets(M, symbol);
                        {m11, m12, m22}

Then

sols:=solve(convert(P . M - M . P,set), vars);

eval(M, sols); 
# map(simplify,%);  # just in case ...
type(%,'Matrix'(hermitian));

       true

To sum up what already was said (Maple does not know that a and b are Reals
and will not modify non-consistent indicator functions) here a varaition:

tmp1:=Matrix(3, (i,j) -> a||i||j + I*b||i||j,shape = hermitian);
tmp2:=Matrix(3, (i,j) -> a||i||j +     0,    shape = diagonal);

M:=tmp1+tmp2;
map(evalc, M); # evalc always assumes real variables

             [    a11        a12 + b12 I    a13 + b13 I]
             [                                         ]
             [a12 - b12 I        a22        a23 + b23 I]
             [                                         ]
             [a13 - b13 I    a23 - b23 I        a33    ]

map(evalc@Re, M);
map(evalc@Im, M);

                        [ 0      b12     b13]
                        [                   ]
                        [-b12     0      b23]
                        [                   ]
                        [-b13    -b23     0 ]

Without going into your code: Searching for Collatz Conjecture maple will give you suggestions.

Ok (guess that some feel insecure using broken English). And sorry for attacking.

I understand that.

Though I have a political problem with the attitude of the original question, no need to discuss it at Maple: it reminds me too much to "Am xxx Wesen soll die Welt genesen".

Is it a bug if posts are in languages other than English? That is actually an incredible attitude.

My real reaction would be much more drastic.

Enough said to that off topic. I hope.

Though it is true that all are born as Americans and speek the very language
due to HIS will and it is obvious that THY will has do be done those stubborn
fuckwitts from Google and NSA have entered Maple and about 0.01% of the web
to silently modify reality.

They even created an atlas asserting that countries like Ukraine or Germany
exist. Or Mexico or Argentina. Please do not believe that.

Even in Canada (a northern province or so) they all speak English, ab initio.

Almost for sure it is that referer (for FF?) and a bug for that board SW.

Ok, will send you one (though it will not be true or illuminating).

At least it seems that I located one source for issues (using Firefox):

It seems that this board (or its trackers) expect the http referer in the header is filled (usually I do not send referers)