Nonsquare matrices as transformations between dimensions | 3Blue1Brown

1. Formula
   Observation

   The screen shows the matrix[[1,3],[2,1]].

[[1, 3], [2, 1]]

A 2×2 matrix representing a linear map from two-dimensional vectors to two-dimensional vectors.

2×2 real matrices

1. Formula
   Observation

   The screen shows the matrix[[0,1,2],[3,4,5],[6,7,8]].

[[0, 1, 2], [3, 4, 5], [6, 7, 8]]

A 3×3 matrix representing a linear map from three-dimensional vectors to three-dimensional vectors.

3×3 real matrices

1. Formula
   Observation

   The screen shows the illustrative matrix[[3,1],[4,1],[5,9]].

[[3, 1], [4, 1], [5, 9]]

An illustrative3×2 matrix with two columns and three rows.

3×2 real matrices

1. Formula
   Observation

   A green label associates the first column with the image of i-hat.

\hat{i}

The first basis vector of the two-dimensional input space.

Two-dimensional vectors

1. Formula
   Observation

   A red label associates the second column with the image of j-hat.

\hat{j}

The second basis vector of the two-dimensional input space.

Two-dimensional vectors

1. Formula
   Observation

   The dimensional illustration displays the input column[2,7].

\begin{bmatrix} 2 \\ 7 \end{bmatrix}

A two-dimensional input vector in a separate dimensional illustration.

Two-dimensional vectors

1. Formula
   Observation

   The dimensional illustration displays the output column[1,8,2]. This is not a computed product of the preceding illustrative matrix and input.

\begin{bmatrix} 1 \\ 8 \\ 2 \end{bmatrix}

A three-dimensional output vector in the separate dimensional illustration.

Three-dimensional vectors

1. Formula
   Observation

   The notation L(v) connects the input and output coordinate columns.

L(\vec{v})

The image of vector v under the linear map L.

Linear transformations

1. Formula
   Observation

   The first column is labeled "Where î lands" / "变换后的 i".

2. Audio
   Observation

   The narration identifies the first basis image as(2,-1,-2).

\hat{\imath}

First basis vector of the input space; its transformed image forms the first column of the matrix.

Input basis vector in a 2-dimensional space

1. Formula
   Observation

   The second column is labeled "Where ĵ lands" / "变换后的 j".

2. Audio
   Observation

   The narration identifies the second basis image as(0,1,1).

\hat{\jmath}

Second basis vector of the input space; its transformed image forms the second column of the matrix.

Input basis vector in a 2-dimensional space

1. Formula
   Observation

   The displayed matrix is $\begin{bmatrix} 2 & 0 \\ -1 & 1 \\ -2 & 1 \end{bmatrix}$.

2. Audio
   Observation

   The narration associates each matrix column with one basis image and identifies the shape as3×2.

\begin{bmatrix} 2 & 0 \\ -1 & 1 \\ -2 & 1 \end{bmatrix}

Matrix encoding a linear transformation from 2D input space to 3D output space, with columns equal to the images of the two basis vectors.

3×2 real matrix

1. Formula
   Observation

   A brace labels the vertical extent as "3 rows" / "3行".

2. Audio
   Observation

   The narration counts three matrix rows.

3 \text{ rows}

Number of coordinates used to describe each landing spot in the output space.

Output dimension count

1. Audio
   Observation

   The narration recalls the earlier2D-to2D and3D-to3D examples.

2. Formula
   Observation

   The animations show2×2 and3×3 matrices beside their grid transformations.

With chosen bases, a linear map between spaces of the same dimension has a square matrix representation.

1. Input and output spaces have the same dimension

1. Audio
   Observation

   The narration introduces the geometric meaning of nonsquare matrices.

2. Formula
   Observation

   The screen separately illustrates a 3×2 matrix and a 2D-input/3D-output map.

A nonsquare matrix represents a linear map whose input and output coordinate spaces have different dimensions. A 3×2 matrix maps two-dimensional inputs to three-dimensional outputs.

1. The number of rows is the output dimension; the number of columns is the input dimension

1. Square matrices and equal-dimensional linear maps

1. Audio
   Observation

   The narration describes preserved grid structure and the origin condition.

2. Animation
   Observation

   The split-screen animation maps a planar grid to a plane in 3D.

The example preserves a parallel, evenly spaced grid and maps the input origin to the output origin. This is a geometric illustration of linearity. Degenerate linear maps may collapse entire grid directions.

1. A linear map; directions may collapse in degenerate cases

1. Audio
   Observation

   The narration explains why it uses separate input and output views.

2. Animation
   Observation

   The two coordinate spaces are displayed side by side.

The input and output vectors have different dimensions and coordinate descriptions. Separate views emphasize this distinction; they are a visualization choice, not a prohibition on other representations.

1. The input and output spaces have different dimensions

1. Nonsquare matrices represent maps between dimensions

1. Audio
   Observation

   The narration connects the nonsquare construction to the earlier matrix method.

2. Formula
   Observation

   The screen begins constructing a matrix by examining where the basis vectors land.

With chosen input and output bases, the coordinates of the image of each input basis vector form one column of the matrix.

1. The map is linear and its action on each input basis vector is known

1. Nonsquare matrices represent maps between dimensions

1. Audio
   Observation

   The narration constructs matrix columns from the coordinates of the basis images.

2. Formula
   Observation

   Columns are explicitly labeled as the landing spots of \hat{\imath} and \hat{\jmath}.

For a linear transformation, the matrix that encodes it is built by placing the coordinates of each transformed basis vector into a column. The first column corresponds to the image of the first basis vector, the second column to the image of the second, and so on.

1. The transformation is linear.

2. The input space has a chosen ordered basis.

3. Coordinates of the output vectors are taken with respect to the output space's standard basis.

1. Audio
   Observation

   The narration counts three rows and two columns to name the matrix shape.

2. Formula
   Observation

   Braces label "3 rows" and "2 columns" around the example matrix.

The number of columns of the matrix equals the number of basis vectors in the input space (input dimension). The number of rows equals the number of coordinates needed to describe each landing spot in the output space (output dimension). A matrix with m rows and n columns is called an m×n matrix.

1. The matrix represents a linear transformation between finite-dimensional spaces.

2. Rows and columns are counted in the standard matrix layout.

1. Matrix columns as transformed basis vectors

1. Audio
   Observation

   The narration describes this example’s column space as a plane through the 3D origin.

2. Formula
   Observation

   Text equates "Span of columns" with "Column space".

The column space of a matrix is the span of its columns. Geometrically, for a linear transformation, it is the set of all possible output vectors—the place where every input vector lands after the transformation.

1. The matrix has columns \mathbf{c}_1, \dots, \mathbf{c}_n.

2. The transformation is linear.

1. Matrix columns as transformed basis vectors

1. Audio
   Observation

   The narration compares the dimension of the displayed image plane with the 2D input to identify full rank in this example.

Full column rank means that all columns are linearly independent, so the rank equals the number of columns. This is possible when there are at least as many rows as columns. In the displayed 3×2 example, the two columns span a 2D plane and the matrix has full column rank. For a general m×n matrix, full rank means rank=min(m,n); full row rank and full column rank should be distinguished.

1. A is an m×n matrix with m≥n when full column rank is possible.

2. The displayed 3×2 example has two independent columns.

1. Column space as the landing region
2. Dimensions of a transformation matrix

1. Audio
   Observation

   The narration interprets3×2 as a map with two input coordinates and three output coordinates.

2. Formula
   Observation

   The matrix $\begin{bmatrix} 3 & 1 \\ 4 & 1 \\ 5 & 9 \end{bmatrix}$ is shown with annotations about columns and rows.

A 3×2 matrix represents a linear transformation from 2D space to 3D space. The two columns indicate two input basis vectors; the three rows indicate that each transformed basis vector is described by three coordinates in the output space.

1. The matrix has exactly 3 rows and 2 columns.

2. Standard basis conventions are used for both input and output spaces.

1. Dimensions of a transformation matrix
2. Matrix columns as transformed basis vectors

1. Audio
   Observation

   The narration explains that the three columns correspond to input basis vectors while the two rows supply output coordinates.

2. Formula
   Observation

   The matrix $\begin{bmatrix} 3 & 1 & 4 \\ 1 & 5 & 9 \end{bmatrix}$ is annotated with "3 basis vectors" and "2 coordinates for each landing spots".

A 2×3 matrix represents a linear transformation from 3D space to 2D space. The three columns correspond to three input basis vectors; the two rows mean each transformed basis vector is described by only two coordinates in the output plane.

1. The matrix has exactly 2 rows and 3 columns.

2. Standard basis conventions are used for both input and output spaces.

1. Dimensions of a transformation matrix
2. Matrix columns as transformed basis vectors

1. Audio
   Observation

   The explanation uses basis images to describe a map from spatial input to a planar output.

2. Diagram
   Observation

   Left side is the three-dimensional input space, right side is the two-dimensional output space, with three basis vectors mapping to the two-dimensional plane.

This segment first understands the linear transformation as a mapping that compresses the entire 3D space onto a 2D plane: the input space is three-dimensional, and the output space is two-dimensional. The screen uses a split view to show this dimension reduction process, and uses the images of the three basis vectors to illustrate that the transformation is determined by where the basis vectors land.

1. Input space is three-dimensional space

2. Output space is two-dimensional plane

3. Subject of discussion is linear transformation

1. Audio
   Observation

   The narration describes the displayed matrix’s image as a plane through the origin in 3D.

2. Animation
   Observation

   A translucent plane is shown passing through the origin in the 3D grid, containing the transformed basis vectors.

For the specific 3×2 matrix $\begin{bmatrix} 2 & 0 \\ -1 & 1 \\ -2 & 1 \end{bmatrix}$, the column space is a 2-dimensional plane passing through the origin of \mathbb{R}^3.

1. The matrix is the one displayed in the video.

2. The transformation is linear.

3. The output space is \mathbb{R}^3 with standard coordinates.

This claim applies to the specific example matrix shown, not to all 3×2 matrices.

1. Audio
   Observation

   The narration establishes full rank for this specific matrix by comparing its image dimension with its input dimension.

The 3×2 example matrix has full rank because dim(Col(A)) = dim(input space) = 2.

1. A is the displayed 3×2 matrix.

2. The input space is 2-dimensional.

3. The column space is 2-dimensional.

Applies to the specific example but reflects the general definition of full column rank.

1. Audio
   Observation

   The narration invites viewers to imagine the effect of flattening spatial structure.

If you imagine yourself being swept up in a transformation from 3D space to a 2D plane, the experience would be very uncomfortable.

1. Imagine experiencing the transformation

Statement of intuitive feeling regarding this specific 3D→2D transformation.

1. Audio
   Observation

   The lesson introduces a linear map from a plane to a number line.

Besides 3D→2D, one can also consider linear transformations from two-dimensional space to one-dimensional space.

1. Discussing mappings between different dimensions within the framework of linear transformations

Existential statement.

1. Audio
   Observation

   The explanation models the output coordinate space as the real number line.

One-dimensional space can be directly understood as the number line.

1. Adopting the video's geometric intuitive expression

Interpretive equivalence for “one-dimensional space”.

1. Audio
   Observation

   The narration follows an equally spaced point sequence and explains that its output spacing stays uniform.

A linear map sends an equally spaced sequence on a line to an equally spaced sequence; the common output spacing may be zero. Equal spacing by itself is not sufficient to establish linearity.

1. The map is linear.

2. The input points form an equally spaced sequence on one line; the output spacing may be zero.

Holds for this class of point sequences.

1. Audio
   Observation

   The explanation constructs the row matrix by writing the scalar image of each input basis vector into its own column.

This type of linear transformation from two-dimensional to one-dimensional can be represented by a 1×2 matrix, and its two columns each have only one element.

1. Transformation is a linear map from R^2 to R^1

Holds for the type shown in this segment.

1. Audio
   Observation

   The explanation constructs the row matrix by writing the scalar image of each input basis vector into its own column.

The two columns of this 1×2 matrix represent the landing spots of the two basis vectors on the number line, and each column requires only one number.

1. Using basis vectors î, ĵ

2. Output space is the number line

Holds for each of the two columns of the matrix.

1. Audio
   Observation

   The ending previews the relationship with dot products, leaving its full explanation to the next lesson.

This type of transformation from two-dimensional to one-dimensional has close ties to the dot product, which will be discussed further in the next episode.

1. Refers to the 2D→1D transformation type just introduced in this segment

Makes a connective judgment about “this type of transformation”.

1. Audio
   Observation

   The narration moves from the 3×2 example to the 2×3 example while retaining the same input/column and output/row interpretation.

2. Formula
   Observation

   Both matrices are annotated with matching explanations of columns→input basis count and rows→output coordinate count.

1. Expression

   $$3\times 2 \text{ matrix}: 2 \text{ cols} \Rightarrow 2 \text{ input basis vectors};\; 3 \text{ rows} \Rightarrow 3 \text{ output coordinates}$$
   Explanation

   Established rule from the preceding segment: columns encode input dimension, rows encode output dimension.

   Justification

   Directly stated and visually annotated for the 3×2 example at 02:18–02:37.

   Shown in the video
2. Expression

   $$2\times 3 \text{ matrix}: 3 \text{ cols} \Rightarrow 3 \text{ input basis vectors};\; 2 \text{ rows} \Rightarrow 2 \text{ output coordinates}$$
   Explanation

   Apply the same column/input and row/output roles to the new dimensional counts.

   Justification

   The narrator explicitly uses "Likewise" and provides identical annotation structure for the 2×3 matrix at 02:38–02:58.

   Shown in the video
3. Expression

   $$T: \mathbb{R}^3 \to \mathbb{R}^2$$
   Explanation

   Conclusion: a 2×3 matrix maps 3D space to 2D space.

   Justification

   Follows directly from step 2 by the definition established in ki-matrix-dimensions.

   Derived from the video

A 2×3 matrix represents a linear transformation from 3-dimensional input space to 2-dimensional output space, by the same column/row dimensional logic used for 3×2 matrices.

1. Audio
   Observation

   The explanation constructs the row matrix by writing the scalar image of each input basis vector into its own column.

2. Formula
   Observation

   Screen first shows an empty matrix, then fills in the first column with 1 and the second column with 2, labeling î lands on 1, ĵ lands on 2.

1. Expression

   $$$$
   Explanation

   First determine that we are discussing a linear transformation from two-dimensional to one-dimensional, so the output is a coordinate on the one-dimensional number line.

   Justification

   The earlier part of the video defined this type of transformation as 2D→1D.

   Shown in the video
2. Expression

   $$\text{Transformation matrix: }[\ \ \ ]$$
   Explanation

   Write out an empty 1×2 matrix framework, indicating that the transformation will consist of two columns.

   Justification

   The narration identifies a 1×2 matrix for this type of linear map.

   Shown in the video
3. Expression

   $$\hat{i}\text{ lands on }1$$
   Explanation

   Observe that the landing spot of basis vector î on the number line is 1, so put 1 into the first column.

   Justification

   Narrator says the two columns represent basis vector landing spots, and the screen synchronously fills the first column with 1.

   Shown in the video
4. Expression

   $$\hat{j}\text{ lands on }2$$
   Explanation

   Observe that the landing spot of basis vector ĵ on the number line is 2, so put 2 into the second column.

   Justification

   Narrator continues to explain that each column needs only one number, i.e., the number that basis vector landed on, and the screen synchronously fills the second column with 2.

   Shown in the video
5. Expression

   $$\begin{bmatrix}1&2\end{bmatrix}$$
   Explanation

   Obtain the complete 1×2 transformation matrix.

   Justification

   Formed by combining the first and second columns corresponding to the landing spots of î and ĵ respectively.

   Shown in the video

For a 2D→1D linear transformation, one can directly read the matrix according to “first column = landing spot of î, second column = landing spot of ĵ”; this example yields [1 2].

1. Formula
   Observation

   Matrix $\begin{bmatrix} 2 & 0 \\ -1 & 1 \\ -2 & 1 \end{bmatrix}$ is displayed throughout.

2. Audio
   Observation

   The narration gives the two basis images(2,-1,-2) and(0,1,1).

3. Animation
   Observation

   3D grid shows transformed basis vectors and a plane representing their span.

Given a linear transformation that sends \hat{\imath} to (2,-1,-2) and \hat{\jmath} to (0,1,1), construct its matrix and interpret its geometry.

1. \hat{\imath} \mapsto (2, -1, -2)

2. \hat{\jmath} \mapsto (0, 1, 1)

3. Input space is 2D; output space is 3D.

Write the transformation matrix and identify its column space and rank status.

1. Expression

   $$A = \begin{bmatrix} 2 & 0 \\ -1 & 1 \\ -2 & 1 \end{bmatrix}$$
   Explanation

   Place the coordinates of T(\hat{\imath}) as column 1 and T(\hat{\jmath}) as column 2.

   Justification

   Definition of matrix representation of a linear transformation (ki-basis-columns).

   Shown in the video
2. Expression

   $$A \in \mathbb{R}^{3 \times 2}$$
   Explanation

   The matrix has 3 rows and 2 columns.

   Justification

   Counting rows and columns of the constructed matrix; confirmed by on-screen labels at 01:48–01:57.

   Shown in the video
3. Expression

   $$\operatorname{Col}(A) = \operatorname{span}\left\{ \begin{pmatrix}2\\-1\\-2\end{pmatrix}, \begin{pmatrix}0\\1\\1\end{pmatrix} \right\}$$
   Explanation

   The column space is the span of the two columns.

   Justification

   Definition of column space (ki-column-space); stated verbally and shown as a plane in the animation.

   Derived from the video
4. Expression

   $$\dim(\operatorname{Col}(A)) = 2 = \dim(\text{input space})$$
   Explanation

   The two columns are linearly independent, so they span a 2D plane; this matches the input dimension.

   Justification

   Stated by narrator at 02:07–02:17 as the reason the matrix is full rank.

   Shown in the video

The matrix is $\begin{bmatrix} 2 & 0 \\ -1 & 1 \\ -2 & 1 \end{bmatrix}$, a 3×2 matrix whose column space is a 2D plane through the origin in \mathbb{R}^3, and which has full rank.

Linear independence of the two columns can be checked directly: neither is a scalar multiple of the other, confirming the column space is 2-dimensional and the matrix is full rank. This matches the narrator's statement and the visual plane in the animation.

1. Formula
   Observation

   Matrix $\begin{bmatrix} 3 & 1 & 4 \\ 1 & 5 & 9 \end{bmatrix}$ is displayed with annotations.

2. Audio
   Observation

   The narration assigns three input basis vectors and two output coordinates to the 2×3 matrix.

Interpret the geometric meaning of the 2×3 matrix $\begin{bmatrix} 3 & 1 & 4 \\ 1 & 5 & 9 \end{bmatrix}$.

1. Matrix has 2 rows and 3 columns.

2. Entries are 3,1,4 in row 1 and 1,5,9 in row 2.

Determine input/output dimensions and describe what rows and columns represent.

1. Expression

   $$3 \text{ columns} \Rightarrow 3 \text{ input basis vectors} \Rightarrow \text{input space is } \mathbb{R}^3$$
   Explanation

   Each column corresponds to one transformed basis vector, so three columns mean three input basis vectors.

   Justification

   Rule established for 3×2 case and reapplied via "Likewise" at 02:38; annotated on screen as "3 basis vectors".

   Shown in the video
2. Expression

   $$2 \text{ rows} \Rightarrow 2 \text{ coordinates per landing spot} \Rightarrow \text{output space is } \mathbb{R}^2$$
   Explanation

   Each row adds one coordinate to every output vector, so two rows mean outputs live in 2D.

   Justification

   Annotated on screen as "2 coordinates for each landing spots" at 02:50–02:58.

   Shown in the video
3. Expression

   $$T: \mathbb{R}^3 \to \mathbb{R}^2$$
   Explanation

   The transformation maps 3D space to 2D space.

   Justification

   Combines steps 1 and 2 using the dimension rule (ki-matrix-dimensions).

   Derived from the video

The 2×3 matrix represents a linear transformation from 3-dimensional space to 2-dimensional space. Its three columns are the images of three input basis vectors, each described by two output coordinates.

Consistent with the general rule A∈ℝ^{m×n} ⇔ T:ℝⁿ→ℝᵐ stated earlier in the clip. No numerical computation is performed in the video for this example; verification is by structural match to the defined rule.

1. Diagram
   Observation

   Left three-dimensional grid has three basis vectors, right two-dimensional grid shows three corresponding image vectors L(î), L(ĵ), L(k̂).

2. Audio
   Observation

   The explanation uses basis images to describe a map from spatial input to a planar output.

Use a split-screen animation to illustrate what a linear transformation from three-dimensional space to a two-dimensional plane looks like.

1. Input space is 3D

2. Output space is 2D

3. Three input basis vectors î, ĵ, k̂ are mapped to the two-dimensional plane

Show how a dimension-reducing linear transformation can be understood via the images of basis vectors.

1. Expression

   $$$$
   Explanation

   First establish the three-dimensional input space on the left and mark the three basis vectors.

   Justification

   Screen title is “3d input”, presenting the domain with a three-axis grid.

   Shown in the video
2. Expression

   $$$$
   Explanation

   Then establish the two-dimensional output space on the right and draw the three image vectors.

   Justification

   Screen title is “Output in 2d”, with three landing vectors appearing on the right.

   Shown in the video
3. Expression

   $$L(\hat{i}),\ L(\hat{j}),\ L(\hat{k})$$
   Explanation

   Label the three images as L(î), L(ĵ), L(k̂) respectively, to represent the action of the entire transformation on the basis vectors.

   Justification

   These symbols are directly labeled next to the vectors on the right.

   Shown in the video

This example shows: a 3D→2D linear transformation can be intuitively grasped through the images of the three input basis vectors in the two-dimensional plane.

The three basis-image arrows correspond to the input basis vectors. This schematic conveys dimensions without giving numerical output coordinates; it is not a computation with the preceding matrix.

1. Formula
   Observation

   Screen shows [2;7] → L(\vec{v}) → [1.8].

2. Audio
   Observation

   The lesson introduces a linear map from a plane to a number line.

Give a specific two-dimensional input vector and show its result in the one-dimensional output.

1. Input vector is $\begin{bmatrix}2\\7\end{bmatrix}$

2. Transformation is denoted L(\vec{v})

3. Output is a one-dimensional vector

Explain that a 2D→1D transformation turns a two-dimensional vector into a number.

1. Expression

   $$\begin{bmatrix}2\\7\end{bmatrix}$$
   Explanation

   First write down the two-dimensional input vector.

   Justification

   The yellow brackets on the left clearly display components 2 and 7.

   Shown in the video
2. Expression

   $$L(\vec{v})$$
   Explanation

   Apply the linear transformation L to this input.

   Justification

   Center of screen labels L(\vec{v}).

   Shown in the video
3. Expression

   $$\begin{bmatrix}1.8\end{bmatrix}$$
   Explanation

   Obtain the one-dimensional output, with value 1.8.

   Justification

   Pink brackets on the right display 1.8, labeled “1d output”.

   Shown in the video

The output of this two-dimensional vector under this transformation is $\begin{bmatrix}1.8\end{bmatrix}$.

The actual dimensional diagram shows [2;7] mapped to [1.8], with a scalar output. It does not reveal the map coefficients. This is a separate illustration from the later row matrix [1,2]; that matrix is not used to compute this diagram.

1. Animation
   Observation

   A string of evenly spaced yellow dots appears on the two-dimensional grid, then is compressed onto the number line while remaining evenly spaced.

2. Audio
   Observation

   The narration follows an equally spaced point sequence and explains that its output spacing stays uniform.

Use a string of dots to demonstrate that linear transformations preserve uniform spacing.

1. There is a line on the two-dimensional plane

2. There is a string of evenly spaced yellow dots on the line

3. The transformation compresses the plane onto the number line

Intuitively explain the geometric manifestation of linearity in dimension-reducing cases.

1. Expression

   $$$$
   Explanation

   First draw a line and several evenly spaced dots on the two-dimensional grid.

   Justification

   In the animation, yellow dots are arranged uniformly along the same line.

   Shown in the video
2. Expression

   $$$$
   Explanation

   Then apply the transformation compressing the entire plane onto the number line.

   Justification

   Screen shows the two-dimensional grid being squeezed onto a one-dimensional line.

   Shown in the video
3. Expression

   $$$$
   Explanation

   Observe the spacing of the mapped dots on the number line.

   Justification

   The narration directs attention to the uniform spacing of the mapped points.

   Shown in the video

These originally evenly spaced dots remain evenly spaced on the number line, thereby reflecting that linearity preserves uniform intervals.

Before-and-after comparison in the animation shows that the relative uniformity of the dots is not broken.

1. Formula
   Observation

   Screen writes Transformation matrix and fills in [1 2].

2. Audio
   Observation

   The explanation constructs the row matrix by writing the scalar image of each input basis vector into its own column.

Write the corresponding transformation matrix based on the landing spots of î and ĵ on the number line.

1. Transformation is 2D→1D

2. î lands on 1

3. ĵ lands on 2

Translate geometric landing spots into matrix representation.

1. Expression

   $$\text{Transformation matrix: }[\ \ \ ]$$
   Explanation

   First establish the 1×2 matrix framework.

   Justification

   Narrator says this type of transformation is encoded by a 1×2 matrix.

   Shown in the video
2. Expression

   $$1$$
   Explanation

   Write the landing spot 1 of î into the first column.

   Justification

   Screen labels “î lands on 1”.

   Shown in the video
3. Expression

   $$2$$
   Explanation

   Write the landing spot 2 of ĵ into the second column.

   Justification

   Screen labels “ĵ lands on 2”.

   Shown in the video
4. Expression

   $$\begin{bmatrix}1&2\end{bmatrix}$$
   Explanation

   Synthesize the complete matrix.

   Justification

   The two columns correspond to the landing spots of the two basis vectors.

   Shown in the video

The corresponding matrix is $\begin{bmatrix}1&2\end{bmatrix}$.

The two columns of the matrix correspond one-to-one with the landing spots of the two basis vectors on the number line.

1. Diagram
   Observation

   Top reads “Next video: Dot products and duality”, top-left shows [3;1]·[2;-1].

2. Formula
   Observation

   Diagonal line shows “Scale by ||\vec{v}||” and “Length of scaled projection”.

1. This segment only gives preview illustrations, without expanding on the complete derivation of dot product and duality.

Use a new set of graphics to hint at the relationship between 2D→1D transformations and the dot product.

1. Dot product notation for vectors [3;1] and [2;-1] appears

2. A diagonal line and projection scaling labels appear

Preview that subsequent episodes will link this type of transformation with the dot product.

1. Expression

   $$\begin{bmatrix}3\\1\end{bmatrix}\cdot\begin{bmatrix}2\\-1\end{bmatrix}$$
   Explanation

   First show a dot product expression.

   Justification

   Top-left corner clearly writes this dot product.

   Shown in the video
2. Expression

   $$\text{Scale by }\|\vec{v}\|$$
   Explanation

   Then show the idea of scaling by vector length.

   Justification

   This text is labeled above the diagonal line.

   Shown in the video
3. Expression

   $$\text{Length of scaled projection}$$
   Explanation

   Finally direct attention to the length of the scaled projection.

   Justification

   This label subsequently appears above the diagonal line.

   Shown in the video

This segment previews the next episode's topic with a dot product expression and projection scaling illustration.

The top title “Next video: Dot products and duality” together with these illustrations constitutes a clear hint of the subsequent topic.

1. Animation
   Observation

   A 2D grid and its basis vectors are transformed beside the 2×2 matrix.

1. A planar grid

2. Basis vectors i-hat and j-hat

3. A 2×2 matrix

1. The grid directions tilt and scale

2. The basis vectors move to their images

1. The illustrated grid remains parallel and evenly spaced

2. The origin maps to the origin

The animation shows a square matrix acting within a space of unchanged dimension.

1. Animation
   Observation

   A 3D grid is transformed beside the 3×3 matrix.

1. A spatial grid

2. Basis vectors

3. A 3×3 matrix

1. Grid directions tilt and scale

2. The basis vectors move

1. Grid directions remain parallel and evenly spaced, including possible collapsed directions

2. The origin maps to the origin

The animation shows a square matrix acting between three-dimensional spaces.

1. Animation
   Observation

   The left view shows the 2D input; the right view shows its image plane in 3D.

1. A 2D input grid

2. A 3D output space

3. The image plane

1. The input grid is mapped to the output plane

2. The output coordinates lie in 3D

1. The illustrated grid remains parallel and evenly spaced

2. The input origin maps to the output origin

Separate views make the different dimensions of input and output explicit.

1. Animation
   Observation

   Right half of screen shows a 3D coordinate grid with yellow axes, green/red transformed vectors, and a translucent plane.

2. Audio
   Observation

   Narrator describes the output of the transformation taking î and ĵ to specified coordinates.

1. 3D Cartesian grid with yellow axis arrows

2. Green vector representing T(\hat{\imath})=(2,-1,-2)

3. Red vector representing T(\hat{\jmath})=(0,1,1)

4. Translucent gray plane through the origin

1. Camera rotates slightly around the origin to show depth

2. Transformed basis vectors remain fixed once placed

3. Plane becomes visible as the span of the two vectors

1. Origin remains at center of grid

2. Vectors emanate from origin

3. Plane always contains both transformed vectors

Demonstrates that a 2D input space is mapped into a 2D subspace (plane) within 3D output space, making the column space geometrically concrete.

1. Animation
   Observation

   Braces appear sequentially labeling rows and columns, followed by "3×2 matrix" text.

2. Formula
   Observation

   Labels read "3 rows / 3行", "2 columns / 2列", "3×2 matrix / 3×2矩阵".

1. Matrix $\begin{bmatrix} 2 & 0 \\ -1 & 1 \\ -2 & 1 \end{bmatrix}$

2. Vertical brace labeled "3 rows"

3. Horizontal brace labeled "2 columns"

4. Title text "3×2 matrix"

1. Row brace appears first

2. Column brace appears second

3. Dimension title fades in last

1. Matrix entries do not change

2. Color coding of columns (green/red) persists

Visually reinforces that row count = output coordinates and column count = input basis vectors, establishing the m×n naming convention.

1. Animation
   Observation

   Text "Span of columns ⇔ Column space" appears beneath the matrix while the 3D plane remains visible on the right.

2. Formula
   Observation

   Bilingual labels: "列张成的空间 ⇔ 列空间".

1. Matrix with bracket underneath

2. Text "Span of columns" and "Column space" connected by ⇔

3. 3D plane in adjacent panel

1. Bracket and text fade in together

2. Plane continues rotating subtly

1. Matrix values unchanged

2. Plane still contains the two column vectors

Links the algebraic notion of column span to the geometric landing region shown in 3D, defining column space operationally.

1. Animation
   Observation

   Pi creature appears beside matrix $\begin{bmatrix} 3 & 1 \\ 4 & 1 \\ 5 & 9 \end{bmatrix}$ with thought bubble showing 3D plane.

2. Audio
   Observation

   Narrator generalizes from the specific example to any 3×2 matrix.

1. Pink Pi creature

2. Generic 3×2 matrix

3. Thought bubble with 3D grid and plane

1. Previous specific matrix replaced by generic entries

2. Thought bubble animates the 2D→3D mapping conceptually

1. Column/row dimensional logic preserved

2. Pi creature serves as consistent visual anchor

Transitions from worked example to general principle, using character + thought bubble to signal conceptual abstraction rather than computation.

1. Animation
   Observation

   Matrix $\begin{bmatrix} 3 & 1 & 4 \\ 1 & 5 & 9 \end{bmatrix}$ receives braces and labels "3 basis vectors" and "2 coordinates for each landing spots".

2. Audio
   Observation

   The narration applies the same dimensional reading to the new matrix.

1. 2×3 matrix

2. Top brace over three columns labeled "3 basis vectors"

3. Side brace over two rows labeled "2 coordinates for each landing spots"

4. Pi creature observing

1. Braces and labels appear sequentially as narrator speaks

2. Colors distinguish the three columns (green, red, blue)

1. Matrix entries static

2. Same annotation style as 3×2 case

The new 2×3 example uses the same column/input and row/output convention. It is a different matrix with different counts; it is not the transpose of the earlier numerical matrix.

1. Diagram
   Observation

   Split screen: left “3d input”, right “Output in 2d”.

2. Animation
   Observation

   Right side sequentially shows image vectors L(î), L(ĵ), L(k̂).

1. three-dimensional grid

2. two-dimensional grid

3. Three input basis vectors

4. Three output image vectors L(î), L(ĵ), L(k̂)

1. Image vectors on the right appear one by one

2. three-dimensional grid briefly fades out, leaving only basis vectors for observation

1. Left side is always the three-dimensional input space

2. Right side is always the two-dimensional output space

3. Three basis vectors maintain correspondence with three image vectors

This animation uses left-right contrast to show dimension-reducing linear transformations: basis vectors in three-dimensional space are mapped to three vectors in the two-dimensional plane, turning abstract transformations into visible landing relationships.

1. Formula
   Observation

   [2;7] → L(\vec{v}) → [1.8].

2. Diagram
   Observation

   Input labeled “2d input”, output labeled “1d output”.

1. two-dimensional input column vector [2;7]

2. Transformation notation L(\vec{v})

3. one-dimensional output column vector [1.8]

1. Arrows connect input, transformation, and output from left to right sequentially

1. Input is always a column of two numbers

2. Output is always a column of one number

This concise flowchart compresses the 2D→1D transformation into a single mapping of “two-dimensional vector enters, one-dimensional number exits”, highlighting dimension reduction.

1. Diagram
   Observation

   Appears “1d space (number line)” and a horizontal line with tick marks.

2. Animation
   Observation

   The planar grid collapses onto the number line; the green basis image lands at 1 and the red image at 2.

1. Number line

2. two-dimensional grid

3. Red basis vector image

4. Green basis vector image

1. two-dimensional plane gradually collapses into a one-dimensional line

2. Basis vector images land on number line ticks 1 and 2

1. Number line as one-dimensional space remains unchanged

2. Tick direction and origin remain consistent

The animation concretizes the abstract statement “from plane to number line”: the entire two-dimensional structure is squeezed onto a one-dimensional line, so the landing spots of basis vectors can be represented by single numbers.

1. Animation
   Observation

   A string of yellow dots is first arranged on a two-dimensional line, then mapped to the number line.

2. Audio
   Observation

   The narration follows an equally spaced point sequence and explains that its output spacing stays uniform.

1. two-dimensional line

2. String of evenly spaced yellow dots

3. Number line

1. Yellow dots move from two-dimensional line to number line

2. Visual positions of dots change but spacing remains uniform

1. Yellow dots were originally evenly spaced

2. Remain evenly spaced after mapping

This set of animations visualizes the linear property using the invariance of uniform dot spacing, explaining that linearity is not just “lines become lines”, but also reflected in preserving evenly spaced structures.

1. Audio
   Observation

   The narration explicitly introduces maps between different dimensions as a meaningful interpretation.

Only square matrices can represent linear transformations; a 3×2 matrix has no geometric meaning.

Nonsquare matrices represent linear maps between coordinate spaces with different dimensions.

1. Audio
   Observation

   The narration explicitly names the nonsquare example as a matrix with three rows and two columns.

2. Formula
   Observation

   On-screen labels explicitly mark unequal row and column counts.

Learners may assume only square matrices encode linear transformations, since introductory examples often use 2×2 or 3×3 matrices.

The video demonstrates that nonsquare matrices like 3×2 and 2×3 also encode valid linear transformations between spaces of different dimensions. Row and column counts independently track output and input dimensions.

1. Audio
   Observation

   The narration identifies full rank even though the image is a plane inside3D.

2. Animation
   Observation

   Plane is shown as a proper subspace, yet rank is affirmed as full.

Students might think that because a 3×2 matrix maps into a 2D plane inside 3D space, it must be rank-deficient or degenerate.

For the displayed 3×2 matrix, the largest possible rank is2. Its column space has that dimension, giving full column rank even though it does not fill the 3D codomain. In general, full rank means rank=min(number of rows,number of columns).

1. Audio
   Observation

   The same column/input and row/output reading is applied to both matrix shapes.

2. Formula
   Observation

   Annotations maintain columns→input, rows→output consistently across both cases.

Learners may incorrectly believe that 'rows always mean input' or that the dimensional rule flips when output dimension is smaller than input dimension.

The video shows the rule is invariant: columns always correspond to input basis vectors and rows always to output coordinates, regardless of whether m>n or m<n. The 2×3 example confirms this by applying the identical logic with swapped magnitudes.

1. Audio
   Observation

   Collapsed grid directions motivate following an equally spaced sequence of points instead.

When understanding 2D→1D linear transformations, still trying to force-fit the intuition of “grid lines remaining parallel and evenly spaced”.

Flattening can make grid directions collapse, so the video follows equally spaced points instead. This remains a visual illustration of an already linear map, not a sufficient test: affine maps can preserve spacing too.

1. Formula
   Observation

   Output written as [1.8], labeled “1d output”.

2. Audio
   Observation

   The lesson introduces a linear map from a plane to a number line.

Seeing a linear transformation and assuming the output is still a vector in the plane.

The plane-to-number-line example produces a scalar, written as [1.8]. It is also a vector with one real component, rather than a vector with two coordinates in the input plane.

1. Audio
   Observation

   The narration moves from square matrices to nonsquare matrices while retaining the linear-map interpretation.

Maps between spaces of different dimensions extend the same matrix-representation method used for maps between equal-dimensional spaces.

1. Audio
   Observation

   The example retains the illustrated grid structure and maps the input origin to the output origin.

The geometric illustration of linearity also applies to the example represented by a nonsquare matrix.

1. Audio
   Observation

   Narrator builds matrix from basis images then immediately counts rows/columns to name dimensions.

2. Formula
   Observation

   Column labels feed directly into row/column brace annotations.

The definition of matrix dimensions (m×n) is derived directly from counting how many basis vectors produce columns (n) and how many coordinates describe each output (m). The former is prerequisite to understanding the latter.

1. Audio
   Observation

   After defining 3×2, narrator immediately interprets it as 2D→3D mapping.

2. Animation
   Observation

   Dimension labels precede the Pi character's generalized explanation.

The abstract dimension rule (columns=input, rows=output) is applied concretely to give geometric meaning to the 3×2 format. The general definition enables the specific interpretation.

1. Audio
   Observation

   Narrator defines column space then uses its dimension to justify full rank in the same breath.

2. Animation
   Observation

   Plane visualization supports both concepts simultaneously.

For this3×2 example, comparing the column-space dimension with the input dimension establishes full column rank. A general nonsquare matrix may instead have full row rank.

1. Audio
   Observation

   The narration compares the two matrix shapes using the same dimensional interpretation.

2. Formula
   Observation

   Identical annotation scheme applied to both matrices with swapped counts.

The two knowledge items illustrate the same underlying rule applied to opposite dimensional inequalities (m>n vs m<n). They are contrastive examples that together demonstrate the rule's generality beyond square cases.

1. Audio
   Observation

   Columns are introduced as landing spots, then those same columns are spanned to define column space.

2. Formula
   Observation

   The matrix built from basis images is the exact object whose columns are spanned.

The column space is constructed from the very columns defined by the transformed basis vectors. The basis-column definition provides the building blocks that the column-space definition operates on.

1. Audio
   Observation

   The lesson introduces a linear map from a plane to a number line.

The video first establishes the intuition of “cross-dimensional linear transformations” with 3D→2D, then generalizes it to 2D→1D, showing that dimension-reducing transformations are not limited to one dimensional combination.

1. Formula
   Observation

   The two columns of the matrix are filled by the landing spots of î and ĵ.

2. Audio
   Observation

   The explanation constructs the row matrix by writing the scalar image of each input basis vector into its own column.

The previous method of “characterizing transformations using images of basis vectors” is specifically applied here to the matrix reading rule: matrix columns are basis vector landing spots.

1. Audio
   Observation

   The explanation models the output coordinate space as the real number line.

2. Diagram
   Observation

   Appears “1d space (number line)” and shows the process of compressing onto the number line.

To understand the output of 2D→1D transformations, one must first accept the video's geometric interpretation of one-dimensional space: it is a number line.

1. Audio
   Observation

   The narration follows an equally spaced point sequence and explains that its output spacing stays uniform.

The lesson illustrates a geometric property of linear maps with an equally spaced sequence. Spacing alone is not a complete characterization of linearity; a direction may also collapse to one point.

1. Formula
   Observation

   Matrix [1 2] appears synchronously with “î lands on 1” and “ĵ lands on 2”.

The representation of the 1×2 matrix internally contains a core reading rule: the two columns record the landing spots of the two basis vectors respectively.

1. Audio
   Observation

   The lesson introduces nonsquare matrices as maps between dimensions.

1. Nonsquare matrices represent maps between dimensions

1. Audio
   Observation

   The narration and animation distinguish a 2D input from a 3D output.

1. Nonsquare matrices represent maps between dimensions
2. Geometric features of linear maps

1. Audio
   Observation

   The narration explains the coordinate-space distinction behind the split-screen presentation.

1. Distinguishing input and output spaces

1. Audio
   Observation

   Opening instruction: "write the coordinates of the landing spots as the columns of a matrix."

2. Formula
   Observation

   Explicit column construction shown for î and ĵ.

1. Matrix columns as transformed basis vectors
2. Specific 3×2 transformation example

1. Audio
   Observation

   Narrator defines 3×2 terminology then gives geometric interpretation.

2. Formula
   Observation

   Dimension labels and generic matrix both present.

1. Dimensions of a transformation matrix
2. Geometric meaning of a 3×2 matrix

1. Audio
   Observation

   Explicit statement: "the matrix is still full rank, since the number of dimensions in this column space is the same as the number of dimensions of the input space."

2. Animation
   Observation

   2D plane in 3D space shown while affirming full rank.

1. Full column rank in the 3×2 example
2. Full rank criterion for the example matrix
3. Misconception: mapping to lower-dimensional subspace implies rank deficiency

1. Audio
   Observation

   "the column space of this matrix, the place where all the vectors land"

2. Formula
   Observation

   "Span of columns ⇔ Column space" equivalence displayed.

1. Column space as the landing region
2. Equating span of columns with column space

1. Audio
   Observation

   Narrator asks and answers what a 2×3 matrix means using the same rule.

2. Formula
   Observation

   Annotated 2×3 matrix with explicit basis-vector and coordinate labels.

1. Geometric meaning of a 2×3 matrix
2. Deriving the geometric meaning of a 2×3 matrix from the 3×2 rule
3. s89-cr-3x2-vs-2x3

1. Audio
   Observation

   Same phrasing pattern used for both 3×2 and 2×3 explanations.

2. Formula
   Observation

   Consistent annotation style across both matrix types.

1. Dimensions of a transformation matrix
2. Geometric meaning of a 3×2 matrix
3. Geometric meaning of a 2×3 matrix
4. Misconception: row/column semantics depend on which is larger

1. Audio
   Observation

   The explanation uses basis images to describe a map from spatial input to a planar output.

1. Linear transformation from three-dimensional space to two-dimensional plane
2. Characterizing cross-dimensional linear transformations using images of basis vectors

1. Formula
   Observation

   Right two-dimensional diagram marks L(î), L(ĵ), L(k̂).

1. Characterizing cross-dimensional linear transformations using images of basis vectors

1. Formula
   Observation

   Screen shows [2;7] → L(\vec{v}) → [1.8].

1. Linear transformation from two-dimensional to one-dimensional