Hello Ioan,

Definetly a bug. Seems the problem is in augment vs unary subtraction in matrices vs symbolic optimization



In the example above, for the symbolic evaluation of f(0):
- if k1 is a number, all goes right;
- if k1 is a matrix and k2:=-k1 -> wrong augment
- if k1 is a matrix and k2:=eval(-k1) or k2:=-1*k1 -> ok

No problems if the evaluation is numeric.

In your example the evaluation of RkL:Hing3d(DaStr) is set to symbolic.




Reported as bug 2292

Edited by user 20 May 2016 21:58:54(UTC)  | Reason: marked ad fixed

Smath is reliable. Watch the 32 bit resulting granularity.
XP Home is 64 bits extended, XP Pro is 32 bits. Even if Smath
would be 64 bits extended, in my XP Pro: still 32 bits.
I understand Andrey not bothering with XP's obsolete
Vista obsolete ... nextnextnext obsolete.

Back to the matrix/vector bug; there is no bug, just
wrong user coding wrt matrix/vector conventions.
-k1 does not negate the matrix. To negate a matrix it
must be multipled in the appropriate order by a scalar .
Here your scalar is 1 it is equivalent to the combined
Euler rotation ... you must write k2:= k1*(-1)
............................NOT k2:=(1)*k1
though Smath does accept this non-conventional rule.

Remenber the rule k2:= k1*(-1) to be in convention
with conventional software [typical Mathcad 8.... 11]
Probably convention in Western Avionics.

Jean

Utilities Matrix Bug.sm (13kb) downloaded 13 time(s).

Ioan,

I'm scared "mon ami" that if you doctor something in there
you might kill several birds with one stone. What about
our friend Dragilev animations ? For practical use in the
common context, the rule is the vector/matrix rule. I can
see some geometric applications. As you can see, there is
only one way to rotate the star. Other subuse should go by
the same rules. Can't please every wish. The most lucky is
the farmer : on Sunday he can please his wife [she gets
coquette un brin] and his horse [no hard work scheduled].

I bet the vector/matrix rule dates back to Euler times ?
If not otherwise confirmed, symbolic returns untreated
expression, just like Maple.

jean

Analemma Star.sm (41kb) downloaded 14 time(s).

SS-2269: the problem is in row()... to know what is the cause we are in the hands of Andrey

Can't be more right: Ioan, Martin, Davide as long as the rules are simple algebraic.
Working with "Advanced Algebra" is known as symbolic in scalar maths, that is the great
Maple, Mathematica [two different engines considered ~ equivalent].
Vectors/matrices also go by their own "Advanced Rules". That starts by "baby logic"
I have a matrix, I want it it to be all elements be negative. So the logic comes
what I have I mutltiply. Even if the matrix multiplication is commutative, it's
non sense saying I have (-1) then multiply by something [i.e: matrix].

That's why advanced matrix algebra follow this rule.
In the attached, you can see two applications of advanced vector/matrix algebra.

1. Convolution
2. Cholesky solver.

They go strictly by the rules the "Mathematicians Society" have established to
work universally for all user. To complicate matters, even advanced vector/matrix
algebra are limited in solving. Many decades ago, Souriau [don't recall the decade]
found out a way to transform unusable matrix to usable one : PseudoMatrix by a process
of PseudoInversion [Mathcad "geninv']. What's that doing ? putting wings to stones !

Once in PseudoInverse form, a chemical balance of fractional composition can be
solved generally and in the right proprotion of all "inter actants".

I bet Knovel [essentially Smath] has that implemented... [in visible code ?]

Reverse the order in the "convolution" ... Syntax error or not in correct order.
Cholesky solver is not commutative either.
Think more: think in term of "Eigen stuff" from which "polyroots results".

Follow the rules: don't drive RHS in UK, don't drive LHS in US/Canada.

For an unknown reason to me, in wide routes, horses naturally drove LHS,
English toke it as is. Unfortunately the "church" couldn't burn to stake all
horses to teach them a lesson. But ± imposed human not to be as stupid as the horse.

Working with advanced matrix algebra is like working with Fourier, Laplace...
You convert in the "transform domain", proceed in that domain, reverse from the
"transform domain" back to normal domain and something happened. Isn't great !

Good thing: Smath obeys "advanced Matrix algebra". Just border line in some
applications.

Jean

Convol.sm (100kb) downloaded 20 time(s).

My dear friends, you can't image how much I'm pedalllllling to rotate an image.
If it works, will burn a candle ... maybe the barn too.
0K, it will be graysaclae, but who cares ? Bed time.

Jean