How to separate a vector into sub-vectors?

Assume, I have the following vector Zs (9x1):

0

0

1

0

1

0

0

1

0

I want to create sub-vectors such that each of them will include three numbers in the Zs, consecutively.

For example,

Zs1 = [0; 0; 1]

Zs2 = [0; 1; 0]

Zs3 = [0; 1; 0]

Thanks,

[James Tursa]

How large is your real problem?

[Taner Cokyasar]

My real Zs is 95x1 and I want to create an even larger one. This is just a conditional situation to decide for creation of a constraint (I am trying to construct an MILP model by intlinprog).

[Taner Cokyasar]

But, in the 95x1 Zs, I need to separate the vector into 95/5 sub-vectors. Then, I will multiply those column vectors by a row vector and compare all of them with each other.

[James Tursa]

How are you doing the multiply? Inner product? If so, then you really really do not want to do what you are proposing. Instead you should be reshaping your original vector into a matrix and then doing a simple matrix multiply. Please post a small example using your 9x1 above to show the exact "multiply ... by a row vector" calculation you would like to do and the expected result. Then we can point you towards a better way of doing it.

[Taner Cokyasar]

Assume, m = 3

k = 1:m; %k is a row vector and every Zs is a column vector.

k*Zs(1) = scalar

k*Zs(2) = scalar2

k*Zs(3) = scalar3

If scalar>scalar2, do something... Otherwise, don't do anything.

If scalar>scalar3, do something... Otherwise, don't do anything.

If scalar2>scalar3, do something... Otherwise, don't do anything.

This is basically, what I am trying to do. "do something" means to create a constraint to be used in intlinprog. Intlinprog uses A, Aeq, b, beq... matrices to solve the problem. I will need to compare "scalar" values and either put a number into "matrix b" (which is a required matrix for intlinprog) or not.

I am planning to calculate "scalar" values first. Then, maybe I can create a for loop to compare them. According to the result of whichever scalar is SMALLER, its corresponding f value ( f values are in the data) will be subtracted from the LARGER scalar's f value.

[Stephen23]

@Taner Cokyasar: Don't create lots of variables like that! You will only make your code much slower, more complicated, and buggy. Oh, it it will also be much harder to debug! Read thsi carefully to know why:

http://www.mathworks.com/matlabcentral/answers/196952-how-to-create-new-variables-in-batches-with-strcat

The best solution: Keep your data in one variable, and learn to use indices effectively.

Answer 1

[James Tursa]

Don't do that. Use cell arrays or some other method instead. E.g., see this link:

http://www.mathworks.com/matlabcentral/answers/143-how-do-i-make-a-series-of-variables-a1-a2-a3-a10

EDIT 7/6/2016:

OK, based on what you have recently posted, try this:

k = your row vector
Zs = your large column vector
n = numel(k);
scalars = k * reshape(Zs,n,[]);

The elements of scalars will be the values you want.

[Taner Cokyasar]

Thank you so much, James!