What controls of this error 'Inner matrix dimensions must agree'?

Sometimes I use these steps but without error

N=5;
s=linspace(0,2);
C=exp(s*(0:5));

Why sometimes other times this error appears ? Inner matrix dimensions must agree.

[Dennis]

s is a vector of length 100, you want to multiply this vector with a vector of length 6.

What result do you expect?

[Matt J]

You probably set 's' to a scalar value in some situations. Only then will the code run without error.

Answer 1

[Stephan]

When using * in Matlab with non-scalars, matlab follows the calculation rules of linear algebra:

>> A = randi(10,3,2)

A =

       4     4
       2     6
       8     2

>> B = randi(10,2,5)

B =

       7     7     8     1    10
       3     7     5     3     2

>> C = A * B

C =

      40    56    52    16    48
      32    56    46    20    32
      62    70    74    14    84

>> whos A B C
  Name      Size            Bytes  Class     Attributes

    A         3x2                48  double              
    B         2x5                80  double              
    C         3x5               120  double

Multiply a 3x2 Matrix with a 2x5 is possible and results in a 3x5 Matrix. If you try the same "backwards" it will give the error:

>> D = B * A
Error using  * 
Incorrect dimensions for matrix multiplication. Check that the number of columns in the first matrix
matches the number of rows in the second matrix. To perform elementwise multiplication, use '.*'.

Read more in the documentation for Matrix operations, where this behavior is described in detail.

Best regards

Stephan

[Sandi J]

Thanks for your response , but if i add s=s(:),the problem solved .

[Stephan]

yes, but you asked for an explanation why sometimes you get the "inner dimensions error" and sometimes not - that was what i tried to answer.

[Dennis]

This will not return an error either:

N=5;
s=linspace(0,2);
C=exp(s'*(0:5));

The question is if this 'solves the problem'. It might be a way to calculate something that is not result that you want.