| Re: vector multiplication of a colum-vectors(1col,3row) and a row-vector(3col,1row), but each vector position[col,row] is a 1000x1400 array [message #52780] |
Wed, 28 February 2007 05:35  |
Paolo Grigis
Messages: 171
Registered: December 2003
|
Senior Member |
|
|
thomas.jagdhuber@dlr.de wrote:
> On 28 Feb., 10:11, Paolo Grigis <pgri...@astro.phys.ethz.ch> wrote:
>> thomas.jagdhu...@dlr.de wrote:
>>> Hi,
>>> I am a rookie in programming IDL. So I try to compute a vector
>>> product out of a colum-vectors(1col,3row) and a row-vector(3col,1row),
>>> with the specialty that each position in the vectors is an 1000x1400
>>> array.
>>> vector1=[[[array1]],[[array2]],[[arry3]]]
>>> vector2=[[[array1]],[[array2]],[[arry3]]]
>>> matrix2=matrix_multiply(vector1,vector2,/btranspose)
>>> But this is not generating a 3x3 Matrix!
>>> Does anyone know anything??
>> Well, most people at least do know something...
>> but maybe you're taking a Socratic stance here ;-)
>>
>> I think that before asking us how to do whatever it is you want
>> done in IDL, you should try to explain better what it is that
>> you are trying to do in the first place (at the level of algebra,
>> not programming language). It seems to me that you are confusing
>> vector (cross) product with scalar product anyway... and why you
>> want to get 9 numbers out of the 4.2 millions you start with?
>>
>> Ciao,
>> Paolo
>>
>>
>>
>>> Thank you very much
>>> Tom
>
> I just have 3 Matrices and I have to calculate the conjugate,
> transpose of this matrices and then multiply each by each so I will
> get 9 matrices
> 11* 12* 13*
> 21* 22* 23*
> 31* 32* 33*
> and in the end I want to store this in one big Matrix of matrices.
> So I can do all this with for-loops but I thought may be there is a
> shorter and more elegant way to compute this.
> Sorry, for my incomplete expalantion.
>
> tom
Ok, then if I understand correctly, if you have just two
3d vectors a and b, you want to compute the outer (or tensor)
product of them in this way:
a=transpose([1,2,3])
b=[10,20,30]
print,a##b
10 20 30
20 40 60
30 60 90
Now, you just happen to have n couples of 3d vectors
and want to compute the n products as above, right?
The you don't have to worry about the loops over the
dimensions (is just 3 by 3 = 9 times), but you want the
multiplication of the n elements to be vectorized.
So if a is nx3 array and b an nx3 array, the result c
should be a nx3x3 matrix given by
FOR i=0,2 DO FOR j=0,2 DO c[*,i,j]=a[*,i]*b[*,j]
where c[m,*,*] is the m-th 3x3 matrix you want.
To see all the 3x3 matrices, you can use
print,transpose(c,[2,1,0])
Ciao,
Paolo
>
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
