Summary of "Pertemuan 15 Aljabar Linear"

Main ideas & concepts

1) General inverse (generalized inverse / pseudo-inverse concept)

• The course reviews matrix inverse, then extends it to matrices that are not necessarily square.
• Matrix inverse (regular inverse):
  • Works when the matrix is square (e.g., (2\times2), (3\times3)).
  • Requires the determinant is not zero.
• General inverse:
  • Can be used for non-square matrices (examples mentioned: (2\times3), (3\times2)).
  • Also addresses cases where the system may have infinite solutions.
• Motivation in linear algebra:
  • Used to help solve SPL (systems of linear equations) in matrix form: [ AX = B ]

2) Definition of general inverse via a consistency condition

For a matrix (A), search for a matrix (G) such that: [ AGA = A ]

• If this holds, (G) is called the general inverse of (A).
• Intuition:
  • Regular inverse “perfectly flips” the matrix.
  • General inverse is the “best possible” usable inverse when usual inverse conditions fail.

3) Penrose conditions (hierarchy of generalized inverses)

The video outlines four Penrose conditions:

1. [ AGA = A ]

2. [ GAG = G ]

3. [ (A^T G)^T = A^T G \quad \text{(transpose condition involving } A^T G\text{)} ]

4. [ (G^T A)^T = G^T A \quad \text{(transpose condition involving } G^T A\text{)} ]

Different generalized inverses satisfy different subsets of these conditions.

4) Types of generalized inverse mentioned

1. (G) inverse (basic form)

• Defined mainly by: [ AGA = A ]

• No extra Penrose constraints → the result may be non-unique.

2. Reflexive (J) inverse

• Extends the basic (J)-inverse by also requiring: [ G G A G = G ]

• The added condition makes the structure more consistent than the ordinary (J) inverse.

3. Moore–Penrose inverse (called “more penos” / written (A^+))

• Emphasized as satisfying all four Penrose conditions.
• Computation formulas are highlighted depending on rank structure.

---

Method / instruction lists

A) Computing Moore–Penrose inverse (A^+) (two cases)

1) Full column rank case (independent columns)

• Assumption: rank equals the number of columns.

• Formula: [ A^+ = (A^T A)^{-1} A^T ]

• Called a left inverse because it satisfies: [ A^+ A = I ]

2) Full row rank case (independent rows)

• Assumption: rank equals the number of rows.

• Formula: [ A^+ = A^T (A A^T)^{-1} ]

• Called a right inverse because it satisfies: [ A A^+ = I ]

B) Solving SPL with (A^+) (closest / least-norm solution idea)

• Given: [ AX = B ]

• Using the generalized inverse: [ X = A^+ B ]

• Interpretation from the worked example:

  • When there are infinitely many solutions, (A^+B) gives the closest point to the origin (described as the “smallest non-solution” / closest solution point).

C) Eigenvalues and eigenvectors (“agent value” & “agent factor”)

For a square matrix (A) and a nonzero vector (x): [ Ax = \lambda x ]

• (\lambda) = eigenvalue (“agent value”)
• (x) = eigenvector (“agent factor”)

Steps to compute eigenvalues

1. Form the characteristic equation: [ \det(\lambda I - A)=0 ]

2. Here, (I) is the identity matrix of the same size as (A).

Steps to compute eigenvectors (once (\lambda) is known)

1. Substitute (\lambda) into: [ (\lambda I - A)x = 0 ]

2. Solve the resulting homogeneous linear system.

3. Take non-trivial solutions as eigenvectors.

---

Key worked examples (what they demonstrate)

1) Moore–Penrose inverse via full row rank formula

• Setup: find (A^+) for a full row rank matrix.

• Process:

  1. Compute (A A^T)
  2. Invert the resulting (smaller) matrix
  3. Use: [ A^+ = A^T (A A^T)^{-1} ]

• Result shown in the video: [ A^+ = [0.5 \ \ 0.5] ]

2) SPL example: closest solution using (A^+)

• Problem statement (interpreted): “Use (A^+) to find the smallest non-solution of (X+Y=2)”, i.e., the closest solution point among infinitely many.
• Given:
  • (A = [1 \ \ 1]), (B = [2])

• Use: [ X = A^+ B ]

• Output discussed:

  • solution point ((1,1)), described as the one closest to the origin ((0,0)).

3) Eigenvalues for a (2\times2) matrix

• Example matrix mentioned: [ A=\begin{bmatrix}1&3\2&0\end{bmatrix} ]

• Steps shown:

  1. Build (\lambda I - A)
  2. Compute the determinant (using the (2\times2) rule (AD-BC))

  3. Get the characteristic polynomial: [ \lambda^2 - \lambda - 6 ]

  4. Factor: [ (\lambda-3)(\lambda+2)=0 ]

  5. Eigenvalues:

     • (\lambda_1=3)
     • (\lambda_2=-2)

4) Eigenvectors for eigenvalue (\lambda=3)

• Solve: [ (\lambda I - A)x=0 ]

• The video arrives at an eigenvector proportional to: [ x = [1,1]^T ]

5) Higher-difficulty eigenvalues/eigenvectors (another matrix)

• Another example mentioned: determine eigenvalues and eigenvectors for a matrix stated as: [ A= \begin{bmatrix}2&1&2\ \cdot & \cdot & \cdot\end{bmatrix} ] (the full matrix entries are not clearly listed in the provided text).

• Eigenvalues stated: [ \lambda = 2,\ 7,\ -1 ]

• Eigenvector computation shown by substitution into: [ (\lambda I - A)x=0 ]

• For (\lambda=2), one eigenvector result reported: [ x = [1,0,0]^T \quad (\text{represented as }100) ]

• Similar substitution steps for (\lambda=7) and (\lambda=-1) are mentioned, though final vectors are not clearly enumerated in the summary.

---

Speakers / sources featured

• Course instructor / lecturer (narrator): the person teaching the linear algebra course content.
• No other named speakers or external sources are clearly identified in the subtitles.

Category

Educational

---

Share this summary

Share on X/Twitter Share on Facebook Share on LinkedIn Share on Reddit Share on WhatsApp Share on Telegram

---

Is the summary off?

If you think the summary is inaccurate, you can reprocess it with the latest model.

Reprocess summary

Video