# How to | Update Parts of a Matrix

The Wolfram Language has many matrix operations that support operations such as building, computing, and visualizing matrices. It also has a rich language for picking out parts of matrices and assigning new values to them.

Define the following matrix:

 In[1]:=
 Out[1]//MatrixForm=

Use [[...]] (the short form of Part) on the left-hand side of an assignment to set an element:

 In[2]:=
 Out[2]=

This shows that the element at position (1, 2) has been updated:

 In[3]:=
 Out[3]//MatrixForm=

To set an entire row, use one index to specify the row and assign it to the new row:

 In[4]:=
 Out[4]//MatrixForm=

To set an entire column, select all rows with All and specify the column:

 In[5]:=
 Out[5]//MatrixForm=

To set a submatrix you can use the short form of Span (;;).

First set up a 5×5 matrix of random integers between 0 and 10:

 In[1]:=
 Out[1]//MatrixForm=

The top-left 3×4 matrix highlighted here corresponds to rows 1 through 3 and columns 1 through 4:

 Out[2]=

Update the highlighted submatrix by using the short form of Span (;;) to specify the relevant span of rows and columns:

 In[3]:=
 Out[3]//MatrixForm=

Update all elements except the outermost rows and columns (negative indices count from the end):

 In[4]:=
 Out[4]//MatrixForm=

When you update a large matrix you should try to avoid doing this in a loop. If you can use one of the updating techniques to update all elements in one command, this will typically be much faster:

 In[1]:=

This is a slow way to update every element in a row:

 In[2]:=
 Out[2]=

This is much faster:

 In[3]:=
 Out[3]=

If you cannot avoid updating a matrix in a loop, you need to take care to avoid extra references to the matrix. Otherwise the matrix will be copied and the loop will not run very fast:

 In[4]:=

This takes care to avoid extra references. The loop runs quite fast:

 In[5]:=
 Out[5]=

This makes a copy of the matrix at every step, and the loop does not run very fast:

 In[6]:=
 Out[6]=