@Ratch 

Insisting in that is like "to kick against the prick" (or the goat, sounds less vulgarly today) ...

PS: it was kind of fun to locate that biblical phrase in a dictionary from 1783 ...

@Ratch 

Insisting in that is like "to kick against the prick" (or the goat, sounds less vulgarly today) ...

PS: it was kind of fun to locate that biblical phrase in a dictionary from 1783 ...

@John May 

Hm ... of course I have not looked into that in detail, but consider the following example
(Maple 12 on Win XP and do not use an evalf on the fcts below before using them):

_Envsignum0:=0;

f := proc(x) if x < 0         then x^2+1; else cos(x); end if; end proc;
g := proc(x) if signum(x) < 0 then x^2+1; else cos(x); end if; end proc;
h := proc(x) if is(x < 0)     then x^2+1; else cos(x); end if; end proc;

t:=18624599/134217728*2^(1/2)+14929833/33554432-
  2^(-18624599/134217728*2^(1/2)-14929833/33554432); 
evalf(%); # ~ - 1e-8

f(t) fails, g(t) returns almost immediately, while h(t) does not and I kill it after 10 min.

PS: If I search for that thread via 'your contributions', then I can not find it.
It seems it silently searches only the unread contributions, what a nonsense
with that board.

@John May 

Hm ... of course I have not looked into that in detail, but consider the following example
(Maple 12 on Win XP and do not use an evalf on the fcts below before using them):

_Envsignum0:=0;

f := proc(x) if x < 0         then x^2+1; else cos(x); end if; end proc;
g := proc(x) if signum(x) < 0 then x^2+1; else cos(x); end if; end proc;
h := proc(x) if is(x < 0)     then x^2+1; else cos(x); end if; end proc;

t:=18624599/134217728*2^(1/2)+14929833/33554432-
  2^(-18624599/134217728*2^(1/2)-14929833/33554432); 
evalf(%); # ~ - 1e-8

f(t) fails, g(t) returns almost immediately, while h(t) does not and I kill it after 10 min.

PS: If I search for that thread via 'your contributions', then I can not find it.
It seems it silently searches only the unread contributions, what a nonsense
with that board.

Using Maple 12 with WIN XP on an intermediate PC (2.2. GHz AMD with 1 MB)

this is much slower than shown in the sheet, roughly by a factor of 20.

Thus I reduced to 10^5 loop for the scalar case and then both timings are
almost the same, somewhat above 2 seconds.

For the matrix case I used 10^4 loops and got ~ 10 seconds for both cases.

@Will feel to do what you want, the logic of that board and of a common conversation need not to be the same

@Will feel to do what you want, the logic of that board and of a common conversation need not to be the same

Ok, the following works as well instead of Dirac

restart;
y(n)-3*y(n-1)-4*y(n-2)=delta(n)+2*delta(n-1); 
rec1:=eval(%, delta= charfcn[0]);
rsolve({rec1 ,y(0) = 1,y(1) = 5}, y(k));

But look at 

rsolve({rec1 }, y(k));

which is like your original problem

For that you may re-read Robert's answer pondering on

eval(rec1, n=0);

                     y(0) - 3 y(-1) - 4 y(-2) = 1

eval(rec1, n=1);

                     y(1) - 3 y(0) - 4 y(-1) = 2

You do not have a definite value in 0 and 1

Ok, the following works as well instead of Dirac

restart;
y(n)-3*y(n-1)-4*y(n-2)=delta(n)+2*delta(n-1); 
rec1:=eval(%, delta= charfcn[0]);
rsolve({rec1 ,y(0) = 1,y(1) = 5}, y(k));

But look at 

rsolve({rec1 }, y(k));

which is like your original problem

For that you may re-read Robert's answer pondering on

eval(rec1, n=0);

                     y(0) - 3 y(-1) - 4 y(-2) = 1

eval(rec1, n=1);

                     y(1) - 3 y(0) - 4 y(-1) = 2

You do not have a definite value in 0 and 1

@Christopher2222 : it acts as printer and works like these (no editor functionality)

First: Thx for making that a post of its own.

Well, "in a few weeks ..." ... happiness is in the soul of the beholder or similar

The only thing I remember as positive change since that board started was,
that Will made possible a thing (trivial expectation of users): print a thread
to a pdf (beyond only the 1st page).

The board started 3 month ago, no?

Cheers

that sounds a bit lame to me: already the first reply by John May has it and Google's 6 entry shows "Maple Code"

You have to look into the links and 'study' it a bit.

At a second thought: what do you really want and know? Scalar multiplication (this is just multiple addition here, no?) or a fast coding?

PS: this is another example for "no updates shown for 'recent contributions', so the thread will die"
I certainly do not want to subscribe to my answers and stumbled over it again just by accident

that sounds a bit lame to me: already the first reply by John May has it and Google's 6 entry shows "Maple Code"

You have to look into the links and 'study' it a bit.

At a second thought: what do you really want and know? Scalar multiplication (this is just multiple addition here, no?) or a fast coding?

PS: this is another example for "no updates shown for 'recent contributions', so the thread will die"
I certainly do not want to subscribe to my answers and stumbled over it again just by accident

a good online dictionary: Milchmädchenrechnung or a reasonable, free off-line dictionary

certainly much could be said ... so just 1 thing: in the long run it is an existential for Maple

And another one: quite far from having a perfect product and even if there are missing a lot of applications beyond the standard I like to use Maple (already MMA syntax would drive me crazy).

Enough said.

But a bit off topic towards "Milchmädchenrechnung" (though only in German), it is a funny and interesting  expression:

http://de.wikipedia.org/wiki/Milchm%C3%A4dchenrechnung

http://de.wikipedia.org/wiki/Fingerrechnen

No problem (may be just some "edited to do ..." already helps).

My ranting was about that board, which kills discussions by construction (thus I reply as 'answer', otherwise one can not see the change).

BTW already a 'thumb up' for my last reply complete crashes the logical and temporal structure in the thread. This was also criticised for a long time.

The feed page neither shows all threads nor time stamps nor any structure, quite useless.

A Brave New Board.