Help needed please

edit savecancel

transpose ?

Some remarks

• Multiplying Ht and Xt works just fine. That is column and row multiplication. This will however yield a [4x4 matrix], not a [4x1] matrix as you are writing.
• This will ensure that subtracting a [4x1] matrix with a [4x4] matrix will not work also due to dimensions not corresponding.
• And thus all the following steps will fail also...

@StevenPuttemans yes I'm wrong ([1 row x 4columns] [4 rows x 1coloumn] =[1 row 1 column]Ht.Xt is a scalar so transpose scalar give a scalar).

@Conor100 your equation are not good

@LBerger, multiplying a 1x4 matrix with a 4x1 matrix is NOT a Scalar. This is the principle behind seperable filters, where you split up a 2D filter by a row and a column based filter. It will yield a 4x4 matrix...

@StevenPuttemans I just write ([1 row x 4columns] [4 rows x 1coloumn] =[1 row 1 column] Ht.Xt is a scalar is it wrong?

I am wondering if I could make Ht a [4x1] and Xt a [1x4] thus Ht.Xt would be a [4x4]. Can I now take the diagonals of this [4x4] to form a [4x1] to represent Ht.Xt and then proceed to calc Yt - Ht.Xt

In my opinion, you should first check your formulas (the theory ?) and then implement it. Not trying to fix the formulas to make it works.

By the way, what is the name of the method you want to use (predictive control ?, ...) ?