The Determinant: Visual Intuition and Comprehensive Guide

A visual understanding of linear transformations is necessary to grasp the concept of the determinant. Linear transformations can be thought of as operations that move space in a way that grid lines remain parallel and evenly spaced, while the origin remains fixed.
Linear transformations often physically change the space they act upon, either by stretching it out or squishing it in.
The determinant is a specific measurement used to understand these transformations. It measures the factor by which a linear transformation changes the area of a given region.
Specifically, the determinant is the scaling factor by which any area (in 2D) or volume (in 3D) increases or decreases under a linear transformation.

Scaling Transformation:
- Consider a matrix with columns $\begin{pmatrix} 3 \\ 0 \end{pmatrix}$ and $\begin{pmatrix} 0 \\ 2 \end{pmatrix}$ .
- This transformation scales the basis vector $\hat{i}$ by a factor of $3$ and the basis vector $\hat{j}$ by a factor of $2$ .
- Focusing on a unit square of area $1$ (with its bottom on $\hat{i}$ and its left side on $\hat{j}$ ), the transformation turns this into a $2 \times 3$ rectangle.
- The area of the region increases from $1$ to $6$ . Therefore, the linear transformation has scaled the area by a factor of $6$ . The determinant of this transformation is $6$ .
Shear Transformation:
- Consider a shear matrix with columns $\begin{pmatrix} 1 \\ 0 \end{pmatrix}$ and $\begin{pmatrix} 1 \\ 1 \end{pmatrix}$ .
- In this transformation, $\hat{i}$ stays in its original place at $(1, 0)$ , while $\hat{j}$ moves to the position $(1, 1)$ .
- The unit square determined by $\hat{i}$ and $\hat{j}$ is transformed into a slanted parallelogram.
- Because the base and height of the parallelogram both remain at length $1$ , the area of the shape remains exactly $1$ .
- Despite the "smushing" of the space, the area is unchanged. The determinant of this transformation is $1$ .

Even though these examples focus on a single unit square, the scaling factor applies to any region in space.
Grid Consistency: Because grid lines remain parallel and evenly spaced after a linear transformation, whatever happens to one square in the grid must happen to any other square, regardless of its size.
Approximation of Arbitrary Shapes: Any non-square shape (such as a generic "blob") can be approximated by grid squares. By using smaller and smaller grid squares, the approximation becomes arbitrarily good.
Because all small component squares are scaled by the same factor (the determinant), the area of the entire shape is scaled by that same single amount.

Determinant of 3: The transformation increases the area of any region by a factor of $3$ .
Determinant of 1/2: The transformation squishes all areas down by a factor of $\frac{1}{2}$ .
Determinant of 0: This occurs if a transformation squishes all of 2D space onto a line or even a single point.
- In these cases, the area of any region becomes zero.
- Checking if a determinant is zero is a crucial computational tool to determine if a transformation squishes everything into a smaller dimension.

The full concept of the determinant allows for negative values, which are used to describe a change in orientation.
Visual Intuition of Negative Areas: A transformation with a negative determinant gives the sensation of flipping space over, similar to turning over a sheet of paper to the other side.
Tracking i-hat and j-hat:
- In the starting orientation, $\hat{j}$ is to the left of $\hat{i}$ .
- If, after the transformation, $\hat{j}$ ends up on the right of $\hat{i}$ , the orientation has been inverted.
- Whenever orientation is inverted, the determinant is negative.
Scaling Properties: The absolute value of the determinant still provides the factor by which areas have been scaled. For example, a matrix with columns $\begin{pmatrix} 1 \\ 2 \end{pmatrix}$ and $\begin{pmatrix} 1 \\ -1 \end{pmatrix}$ has a determinant of $-3$ . This means the area is scaled by a factor of $3$ and the orientation is flipped.
Transition to Negative Values: As $\hat{i}$ moves closer to $\hat{j}$ , the areas in space are squished more, and the determinant approaches $0$ . When the vectors align perfectly, the determinant is $0$ . If $\hat{i}$ continues past $\hat{j}$ , it is natural for the determinant to continue decreasing into the negative numbers.

In 3D, the determinant represents the factor by which volumes are scaled.
To visualize this, focus on a $1 \times 1 \times 1$ unit cube whose edges rest on the basis vectors $\hat{i}$ , $\hat{j}$ , and $\hat{k}$ .
Parallelepiped: After the transformation, the unit cube is warped into a shape known as a parallelepiped.
Because the initial cube has a volume of $1$ , the determinant is simply the volume of the resulting parallelepiped.
Determinant of 0 in 3D: This implies that all of space is squished onto a shape with zero volume, such as a 2D flat plane, a 1D line, or a 0D point.
- From a column perspective, a determinant of zero indicates that the columns of the matrix are linearly dependent.
Orientation in 3D (Right Hand Rule):
- Orientation is determined using the Right Hand Rule: Point the forefinger of the right hand in the direction of $\hat{i}$ , the middle finger in the direction of $\hat{j}$ , and the thumb in the direction of $\hat{k}$ .
- If the transformed basis vectors still allow the use of the right hand in this manner, the orientation is unchanged (positive determinant).
- If the transformed configuration can only be represented using the left hand, the orientation is flipped (negative determinant).

2x2 Matrix Formula: For a matrix $\begin{pmatrix} a & b \\ c & d \end{pmatrix}$ , the formula is: $ad - bc$
2x2 Intuition:
- If $b$ and $c$ are $0$ , the term $a$ represents how $\hat{i}$ is stretched in the x-direction, and $d$ represents how $\hat{j}$ is stretched in the y-direction. The area is a rectangle of $a \times d$ .
- If only one of $b$ or $c$ is $0$ , the shape is a parallelogram with base $a$ and height $d$ , so the area remains $a \times d$ .
- The $bc$ term generally represents how much the parallelogram is stretched or squished in the diagonal direction.
3D Formulas: There is a formula for 3D determinants, but manual computation is often considered less essential than understanding the underlying representation of scaling and orientation.

Composition of Transformations: If you multiply two matrices together, the determinant of the resulting matrix equals the product of the determinants of the original two matrices: $\det(M_1 M_2) = \det(M_1) \det(M_2)$
Conceptual Explanation: This makes sense because if the first transformation scales space by one factor, and the second transformation scales that already-scaled space by another factor, the total scaling factor is simply the product of those individual scaling steps.