@puckie When I execute your worksheet, it works perfectly for me. I didn't make any changes. Try Execute -> Worksheet from the Edit menu.

@puckie When I execute your worksheet, it works perfectly for me. I didn't make any changes. Try Execute -> Worksheet from the Edit menu.

@puckie I didn't see any attached file. Please attach the worksheet itself. Just from your description, I suspect that in the definition of Matrix M, you have F[i](V[i]) instead of the correct F[i](V(j)).

@puckie I didn't see any attached file. Please attach the worksheet itself. Just from your description, I suspect that in the definition of Matrix M, you have F[i](V[i]) instead of the correct F[i](V(j)).

Perhaps this is more clear:

restart:
Digits:= 200:
Order:= 20:
S:= convert(asympt(ln(n!), n), polynom);

evalf[10](evalf(eval(S, n= (1024+83/99)*10^95)/ln(10)));

Perhaps this is more clear:

restart:
Digits:= 200:
Order:= 20:
S:= convert(asympt(ln(n!), n), polynom);

evalf[10](evalf(eval(S, n= (1024+83/99)*10^95)/ln(10)));

Bravo. This is a beautiful and complicated geometric construction done entirely in Maple's geometry package, with step-by-step plots.

@erik10 If you show an example of each situation, I can explain why Maple responded the way that it did.

@erik10 If you show an example of each situation, I can explain why Maple responded the way that it did.

1. Remove all elements of list U from list A:

remove(`in`, A, U);

But searching a list is linear time whereas searching a set is logarithmic time because sets are stored sorted. So the above has time complexity O(|A|*|U|), but a trivial conversion of U to a set lowers the time complexity to O(|A|*ln(|U|)). This makes a huge difference in time when U is large. So, change the above command to

remove(`in`, A, {U[]});

And do likewise for all the other alternatives for removing list U from list A: Replace U with {U[]}.

2. Merge two lists into a list of pairs:

Use zip(`[]`, A, B) instead of zip((a,b)-> [a,b], A, B). This yields about 40% savings because `[]` is a kernel procedure whereas (a,b)-> [a,b] is, of course, user-defined.

@Markiyan Hirnyk Well, if we're having a competition for shortness,

nops([StringTools:-SearchAll("5", sprintf("%q", $1..2013))]);

@Markiyan Hirnyk Well, if we're having a competition for shortness,

nops([StringTools:-SearchAll("5", sprintf("%q", $1..2013))]);

@CarlitosVillalbaGalea I does mean the imaginary unit, sqrt(-1), but in the example above it cancels, leaving only real values. The cancellation can usually not be realized algebraically, but it can sometimes be realized through a conversion to trigonometric functions. I found a slightly simpler example:

Download cubic.mws

@CarlitosVillalbaGalea I does mean the imaginary unit, sqrt(-1), but in the example above it cancels, leaving only real values. The cancellation can usually not be realized algebraically, but it can sometimes be realized through a conversion to trigonometric functions. I found a slightly simpler example:

Download cubic.mws

After reading the help page, which has far too few examples, and also reading the long thread referred to by PatrickT, I would think that you could achieve what you what by using pre(x(t)), that being the previous value of x(t). But I can't make it work. Perhaps pre only works with discrete variables. Specifically, I don't understand why events= [[abs(pre(x(t)) - x(t)) - 1e-8, halt]] does not work.