W3. Orthogonal Vector Spaces

1. Theory

1.1 Vector Spaces

A vector space over a field (typically or ) is a fundamental mathematical structure consisting of a set equipped with two operations: vector addition and scalar multiplication. These operations must satisfy eight axioms that ensure the space is “well-behaved” and closed under these operations.

1.1.1 Definition and Axioms

Vector spaces must satisfy the following axioms for all vectors and scalars :

Addition Axioms:

• Commutativity:
• Associativity:
• Identity: There exists a zero vector such that for all
• Inverses: For every , there exists such that

Scalar Multiplication Axioms:

• Associativity:
• Identity:
• Distributivity (over vectors):
• Distributivity (over scalars):

1.1.2 Common Examples of Vector Spaces

• Euclidean spaces : Collections of ordered -tuples of real numbers with component-wise addition and scalar multiplication
• Matrix spaces : All matrices under matrix addition and scalar multiplication
• Polynomial spaces : All polynomials of degree at most with polynomial addition and scalar multiplication
• Function spaces : All continuous functions on interval with pointwise addition and scalar multiplication

1.2 Subspaces

A subspace is a subset of a vector space that is itself a vector space under the same operations. This is a key concept because subspaces inherit the structure of the larger space while potentially having lower dimension.

1.2.1 Subspace Test

Rather than checking all eight axioms, we can use a simpler criterion. A non-empty subset of a vector space is a subspace if and only if:

• Closed under addition: For all , we have
• Closed under scalar multiplication: For all and , we have

These two conditions automatically guarantee closure gives us the zero vector (set ), additive inverses (set ), and all other axioms are inherited from the parent space.

1.2.2 Examples of Subspaces

Subspaces of :

• The origin:
• Any line through the origin
• Any plane through the origin
• The entire space

Not Subspaces of :

• A line NOT through the origin (fails the zero vector test)
• A plane NOT through the origin (fails the zero vector test)
• The first octant (closed under addition but NOT under scalar multiplication by negative numbers)

1.2.3 Span of Vectors

The span of a set of vectors is the set of all possible linear combinations:

The span of any set of vectors is always a subspace of the ambient vector space.

1.3 Column Space and Null Space

For a matrix , two fundamental subspaces are essential in linear algebra.

1.3.1 Column Space

The column space (or range) of an matrix , denoted or , is the span of its column vectors. It is a subspace of .

The column space answers the question: “For which vectors does the equation have a solution?”

1.3.2 Null Space

The null space (or kernel) of an matrix , denoted or , is the set of all solutions to the homogeneous equation :

It is a subspace of .

1.3.3 Complete Solution Structure

If has a solution, then the complete solution is:

where is a particular solution (any solution to ) and represents all solutions to the homogeneous equation (the null space).

1.4 Inner Products and Orthogonality

An inner product generalizes the dot product to abstract vector spaces, allowing us to measure angles and distances even in high-dimensional spaces.

1.4.1 Definition of Inner Product

An inner product on a vector space is a function satisfying for all and :

• Conjugate symmetry: (equals for real spaces)
• Linearity in first argument:
• Positive-definiteness: , with equality if and only if

1.4.2 Standard Inner Products

• In : (dot product)
• In :
• For continuous functions:

1.4.3 Orthogonality

Two vectors and are orthogonal (perpendicular) if their inner product is zero:

Orthogonal vectors have zero angle between them (90°).

1.4.4 Orthonormal Sets

A set of vectors is orthonormal if:

The vectors are mutually orthogonal and each has unit length (magnitude 1).

1.4.5 Projection onto a Vector

The projection of vector onto vector (for ) is:

This formula gives the component of in the direction of . The remaining component is orthogonal to .

1.5 The Gram-Schmidt Orthogonalization Process

The Gram-Schmidt process is an algorithm that takes a set of linearly independent vectors and produces an orthonormal basis for the same subspace. This is invaluable because orthonormal bases simplify computations and improve numerical stability.

1.5.1 Why Orthogonalization?

Orthogonal bases are superior to arbitrary bases for several reasons:

• Simplified computations: Inner products become straightforward
• Numerical stability: Reduces round-off errors in calculations
• QR decomposition: Enables efficient matrix factorizations
• Signal processing: Removes correlations in data
• Least squares: Provides efficient solutions to overdetermined systems

1.5.2 The Algorithm

Given linearly independent vectors , the Gram-Schmidt process constructs orthonormal vectors spanning the same subspace.

Step 1: First vector

Step 2: Second vector (subtract projection of onto )

Step 3: Third vector and beyond

For :

The key insight is that we subtract from each new vector all its projections onto the previously orthonormalized vectors, ensuring orthogonality.

1.6 QR Decomposition

QR decomposition is a matrix factorization that writes any matrix with linearly independent columns as a product of an orthogonal matrix and an upper triangular matrix.

1.6.1 The QR Theorem

Every matrix with linearly independent columns can be factorized as:

where:

• is an matrix with orthonormal columns
• is an upper triangular matrix with positive diagonal entries

1.6.2 Construction from Gram-Schmidt

The QR decomposition emerges naturally from the Gram-Schmidt process. If , the Gram-Schmidt process gives:

where the coefficients form the matrix :

1.6.3 Computing R from Q and A

Once is obtained, the matrix is computed as:

This works because (orthonormal columns).

1.7 Numerical Stability: Modified Gram-Schmidt

The classical Gram-Schmidt process can suffer from numerical instability when vectors are nearly linearly dependent. The modified Gram-Schmidt algorithm performs the same calculations in a different order that is more resistant to round-off errors:

Instead of projecting onto all previous vectors simultaneously, modified Gram-Schmidt subtracts projections one at a time, updating the remaining vector after each subtraction. This preserves orthogonality better in finite precision arithmetic.

---

2. Definitions

• Vector Space: A set with operations of addition and scalar multiplication satisfying eight axioms (closure, associativity, commutativity, identities, and inverses).
• Subspace: A non-empty subset of a vector space that is closed under addition and scalar multiplication.
• Column Space (): The set of all linear combinations of the columns of matrix ; equivalently, the set of all such that has a solution.
• Null Space (): The set of all vectors satisfying .
• Span: The set of all linear combinations of a given set of vectors; always forms a subspace.
• Basis: A linearly independent set of vectors that spans a subspace.
• Orthogonal Vectors: Two vectors whose inner product is zero; they are perpendicular.
• Orthonormal Set: A set of vectors that are mutually orthogonal and each have unit length.
• Inner Product: A function that assigns a scalar to each pair of vectors, generalizing the dot product.
• Norm (Magnitude): The length of a vector, computed as .
• Projection: The component of one vector in the direction of another, computed as .
• Gram-Schmidt Process: An algorithm that converts a linearly independent set of vectors into an orthonormal set spanning the same subspace.
• QR Decomposition: A matrix factorization where has orthonormal columns and is upper triangular.

---

3. Formulas

• Dot Product (Inner Product in ):
• Vector Magnitude (Norm):
• Unit Vector: (for )
• Orthogonality Condition: if and only if
• Projection of onto :
• Orthogonal Component: is orthogonal to
• Gram-Schmidt (General Step): For :
• QR Decomposition: where and is upper triangular
• Computing R from Q and A:

---

4. Practice

4.1. Which Pairs Are Orthogonal (Lab 3, Task 1)

Which pairs are orthogonal among the vectors ?

Click to see the solution

Key Concept: Two vectors are orthogonal if their dot product equals zero: .

We need to compute all pairwise dot products:

1. Check : Not orthogonal.
2. Check : Orthogonal.
3. Check : Not orthogonal.
4. Check : Orthogonal.
5. Check : Not orthogonal.
6. Check : Not orthogonal.

Answer: The orthogonal pairs are and .

4.2. Project Vector onto Line (Lab 3, Task 2)

Project the vector onto the line through . Check that is perpendicular to :

where .

Click to see the solution

Key Concept: The projection of onto is .

1. Compute :
2. Compute :
3. Calculate the projection:
4. Find the error vector :
5. Verify : ✓ Indeed perpendicular.

Note: Since , the vector lies exactly on the line through (they are parallel).

Answer: and .

4.3. Apply Gram-Schmidt (Basic Case) (Lab 3, Task 3a)

Apply the Gram-Schmidt process to:

Click to see the solution

Key Concept: The Gram-Schmidt process converts linearly independent vectors into orthonormal vectors.

1. Orthogonalize the first vector:
2. Orthogonalize the second vector:
   • Calculate
   • Calculate
   • Thus:
3. Normalize to get orthonormal vectors:

Answer: Orthogonal vectors: ,

Orthonormal vectors: ,

4.4. Find QR Decomposition of a 3×2 Matrix (Lab 3, Task 3b)

Find the QR decomposition of:

Click to see the solution

Key Concept: QR decomposition factors a matrix with linearly independent columns as where has orthonormal columns and is upper triangular.

1. Apply Gram-Schmidt to the columns of :

   • ,

2. First orthonormal vector:

3. Orthogonalize second column:

4. Normalize second vector:

5. Construct and :

   From the Gram-Schmidt process, we have:

   • , so first column of is
   • , so second column of is

Answer:

4.5. Orthogonalize Vectors in ℝ⁴ and Find QR Decomposition (Lab 3, Task 3c)

Orthogonalize the following vectors in and compute the QR decomposition of :

Click to see the solution

Key Concept: Gram-Schmidt orthogonalization systematically removes projections to create orthogonal vectors.

1. First vector:
2. Second vector:
3. Third vector:
4. Construct by normalizing:
5. Construct from dot products:

Answer: The orthogonal vectors are as computed above, and the QR decomposition is with the matrices shown.

4.6. Express Gram-Schmidt Orthogonalization as QR (Lab 3, Task 4)

Express the Gram-Schmidt orthogonalization of as :

Given vectors with components, what are the shapes of , , and ?

Click to see the solution

Key Concept: QR decomposition naturally emerges from Gram-Schmidt orthogonalization. The shapes are determined by the input dimensions.

1. Apply Gram-Schmidt:
2. Orthogonalize second vector:
3. Normalize for :
4. Construct matrices:
5. Matrix shapes: Given vectors with components:
   • is
   • is (with orthonormal columns)
   • is (upper triangular)

Answer: ,

For vectors with components: is , is , is .

4.7. Find Orthonormal Vectors from Nonorthogonal Vectors (Lab 3, Task 5)

From the nonorthogonal vectors , find orthonormal vectors :

Click to see the solution

Key Concept: Apply Gram-Schmidt and normalize each orthogonal vector to unit length.

1. First orthonormal vector:
2. Second orthonormal vector:
3. Third orthonormal vector:

Answer:

4.8. Project Vector onto Two Orthogonal Lines (Lab 3, Task 6)

Project the vector onto the lines through and . Draw the projections and and add . The projections do not add to because the ’s are not orthogonal.

Click to see the solution

Key Concept: When basis vectors are not orthogonal, projections don’t simply add. This motivates orthogonalization.

1. Project onto :

2. Project onto :

3. Add the projections:

4. Compare with original vector:

   The sum of projections does not equal because and are not orthogonal. Check: .

Answer: , ,

4.9. Identify Which Subsets Are Subspaces (Lab 3, Task 7)

Which of the following subsets of are actually subspaces?

1. The plane of vectors with first component .
2. The plane of vectors with .
3. The vectors with (the union of two subspaces: the plane and the plane ).
4. All combinations of two given vectors and .
5. The plane of vectors that satisfy .

Click to see the solution

Key Concept: A subset is a subspace if it contains the zero vector and is closed under addition and scalar multiplication.

(a) Vectors with : IS a subspace * Contains zero vector ✓ * Closed under addition: if and , then ✓ * Closed under scalar multiplication: if , then ✓

(b) Vectors with : NOT a subspace * Does not contain the zero vector (zero vector has ) ✗

(c) Union of planes or : NOT a subspace * Contains zero vector ✓ * However, not closed under addition: example: , which has and ✗

(d) All combinations of and : IS a subspace * This is the span of two vectors, which is always a subspace * Contains zero vector (take both coefficients as 0) ✓ * Closed under addition (linear combinations remain linear combinations) ✓ * Closed under scalar multiplication ✓

(e) Vectors satisfying : IS a subspace * The zero vector satisfies ✓ * Closed under addition: if and , then ✓ * Closed under scalar multiplication: if , then ✓

Answer: Subspaces are (a), (d), and (e). Not subspaces are (b) and (c).

4.10. True or False for Matrix Subspaces (Lab 3, Task 8)

For all matrices, mark each statement true or false (check addition using an example):

1. The skew-symmetric matrices in (with ) form a subspace.
2. The unsymmetric matrices in (with ) form a subspace.
3. The matrices that have in their nullspace form a subspace.

Click to see the solution

Key Concept: Check if subsets contain the zero matrix and are closed under matrix addition and scalar multiplication.

(a) Skew-symmetric matrices (): TRUE * Zero matrix satisfies ✓ * Closed under addition: if and , then ✓ * Closed under scalar multiplication: if , then ✓

(b) Unsymmetric matrices (): FALSE * The zero matrix satisfies , so it is symmetric, not unsymmetric ✗ * Counterexample for closure: Two unsymmetric matrices can add to a symmetric matrix. For instance: Both are unsymmetric, but is also unsymmetric. However, the failure to contain the zero matrix already disqualifies it.

(c) Matrices with in nullspace: TRUE * These are matrices such that * The zero matrix satisfies this (any vector in its nullspace) ✓ * Closed under addition: if and , then ✓ * Closed under scalar multiplication: if , then ✓

Answer: (a) TRUE, (b) FALSE, (c) TRUE

4.11. Show Vector in Subspace Spanned by Vectors (Assignment 3, Task 1)

Show that is in the subspace of spanned by :

Click to see the solution

Key Concept: To show is in the span of , we need to find scalars such that .

1. Set up the linear system:
2. This gives us the augmented matrix:
3. Row reduce to find if a solution exists: After row reduction (detailed steps omitted for brevity), we can verify the system is consistent.
4. Check by finding coefficients (one particular solution): After row reduction, a solution exists. One particular solution is (confirmed by consistent RREF), confirming is in the span.

Answer: is in (solution exists).

4.12. Determine if Vector is in Column Space (Assignment 3, Task 2)

Determine if is in the subspace of spanned by the columns of :

Click to see the solution

Key Concept: is in if and only if the system is consistent.

1. Set up the augmented matrix :

2. Row reduce to echelon form:

   • , ,

3. Continue reduction: After complete row reduction, check if there is a row of the form with .

   If no such row exists, the system is consistent and . If such a row exists, the system is inconsistent and .

4. Result: Based on the reduced form, determine consistency.

Answer: Check row-reduced form: if consistent, then ; if inconsistent, then .

4.13. Find for Null and Column Spaces (Assignment 3, Task 3)

For the matrices in Exercises 17–20, (a) find such that is a subspace of , and (b) find such that is a subspace of .

Matrix 17:

Matrix 18:

Matrix 19:

Matrix 20:

Click to see the solution

Key Concept: For an matrix :

• is a subspace of (based on number of columns)
• is a subspace of (based on number of rows)

Matrix 17 ():

• 1. is a subspace of (2 columns)
• 2. is a subspace of (4 rows)

Matrix 18 ():

• 1. is a subspace of (3 columns)
• 2. is a subspace of (4 rows)

Matrix 19 ():

• 1. is a subspace of (5 columns)
• 2. is a subspace of (2 rows)

Matrix 20 ():

• 1. is a subspace of (5 columns)
• 2. is a subspace of (1 row)

Answer:

• Matrix 17: (a) , (b)
• Matrix 18: (a) , (b)
• Matrix 19: (a) , (b)
• Matrix 20: (a) , (b)

4.14. True or False: Null Space and Column Space (Assignment 3, Task 4)

For an matrix , mark each statement True or False and justify:

1. The null space of is the solution set of the equation .
2. The null space of an matrix is in .
3. The column space of is the range of the mapping .
4. If the equation is consistent, then is .
5. The kernel of a linear transformation is a vector space.
6. Col is the set of all vectors that can be written as for some .

Click to see the solution

Key Concept: Understand the definitions of null space, column space, and their properties.

(a) TRUE By definition, , which is exactly the solution set of .

(b) FALSE The null space is in (the domain of ), not . Since has columns, solutions have components. Justification: .

(c) TRUE The range (or image) of the mapping is exactly , which is the column space of .

(d) FALSE If is consistent for ONE , it does not mean . For example, if (size ), then is solvable, but (all reals), not . Justification: We would need the equation to be consistent for ALL .

(e) TRUE The kernel of a linear transformation is , which is a subspace of . It contains the zero vector, is closed under addition, and is closed under scalar multiplication.

(f) TRUE This is the definition of column space: .

Answer: (a) TRUE, (b) FALSE, (c) TRUE, (d) FALSE, (e) TRUE, (f) TRUE

4.15. True or False: Subspaces and Dimension (Assignment 3, Task 5)

For an matrix , mark each statement True or False and justify:

1. A null space is a vector space.
2. The column space of an matrix is in .
3. Col is the set of all solutions of .
4. is the kernel of the mapping .
5. The range of a linear transformation is a vector space.
6. The set of all solutions of a homogeneous linear differential equation is the kernel of a linear transformation.

Click to see the solution

Key Concept: Apply subspace criteria and definitions of fundamental matrix spaces.

(a) TRUE A null space is a subspace by definition. It contains the zero vector, is closed under addition, and is closed under scalar multiplication.

(b) TRUE The column space consists of all linear combinations of the -component column vectors of , so .

(c) FALSE The column space is the set of all vectors for which HAS A SOLUTION, not the set of all solutions themselves. Justification: Solutions to are vectors in , while .

(d) TRUE By definition, the kernel of is .

(e) TRUE The range of a linear transformation is , which is a subspace of . It is closed under addition and scalar multiplication.

(f) TRUE Consider a homogeneous linear differential equation like . The solution set is the kernel of the differential operator . More generally, this mapping is a linear transformation between function spaces, and the solutions form its kernel, which is always a vector space.

Answer: (a) TRUE, (b) TRUE, (c) FALSE, (d) TRUE, (e) TRUE, (f) TRUE

4.16. True or False: Column Space Properties (Tutorial 3, Task 1)

True or false (with a counterexample if false)?

1. The vectors that are not in the column space form a subspace.
2. If contains only the zero vector, then is the zero matrix.
3. The column space of equals the column space of .
4. The column space of equals the column space of .

Click to see the solution

Key Concept: Understand when sets are closed under operations and form subspaces.

(a) FALSE Counterexample: Let . Then , so the complement is empty. For a non-empty case where Col: Let (a matrix with entry 1). Then . But if , then , and vectors not in Col would be all non-zero reals, which don’t form a subspace (no zero vector).

More generally: If , their sum might be in Col or might not, so closure is not guaranteed. Example: In with , Col. Then and , but .

(b) TRUE If contains only the zero vector, then the only possible output of is . This means every column of must be the zero vector. Therefore must be the zero matrix.

(c) TRUE

Since scalar multiplication by 2 is invertible (we can divide by 2), we have: (scalar multiples of elements in Col are still in Col, and vice versa).

More rigorously: where , so (since the set is unchanged by scalar multiplication).

(d) FALSE Counterexample: Let (the identity matrix). Then (the zero matrix). - -

These are not equal, so the statement is false.

Answer: (a) FALSE, (b) TRUE, (c) TRUE, (d) FALSE

4.17. Examples of Matrices with Varying Solution Counts (Tutorial 3, Task 2)

Give examples of matrices for which the number of solutions to is

1. 0 or 1, depending on .
2. , regardless of .
3. 0 or , depending on .
4. 1, regardless of .

Click to see the solution

Key Concept: The number of solutions depends on the rank and dimensions of the matrix.

(a) 0 or 1 depending on : Example: (a matrix with full column rank)

For :

• If , then the system has exactly 1 solution:
• If , there is no solution (inconsistent)

(b) regardless of : This requires a non-square matrix with more rows than columns, but the system must always be solvable (Col). This is impossible.

However, if we allow underdetermined systems: Example: (a matrix)

For any , the equation has infinitely many solutions (one free variable).

(c) 0 or depending on : Example: (a matrix with rank 1)

For :

• If (consistent with the dependence of rows), there are infinitely many solutions
• If , there is no solution

(d) 1 solution regardless of : Example: (a identity matrix)

For any , the unique solution is .

More generally, any invertible matrix works.

Answer: (a)

4. (any invertible matrix)

4.18. Find RREF and Special Solutions for Parameter-Dependent Matrix (Tutorial 3, Task 3)

For every , find (reduced row echelon form) and the special solutions to :

Click to see the solution

Key Concept: Row reduction and special solutions depend on parameter values.

First Matrix:

1. Row reduce:

2. Case 1:

   Swap rows 2 and 3, then divide row 2 by :

   Free variables:

   Special solution for :

   Special solution for :

3. Case 2:

   Free variables:

   Special solutions: , ,

Second Matrix:

1. Case 1: and

   Invertible diagonal matrix:

   Special solution: None (only the trivial solution )

2. Case 2:

   Free variable:

   Special solution:

3. Case 3:

   Free variable:

   Special solution:

Answer: Results depend on parameter as shown above.

4.19. Find Solutions to Ax = 0 and Ax = b from Given Solution (Tutorial 3, Task 4)

1. If has two solutions and , find two solutions to .
2. Then find another solution to .

Click to see the solution

Key Concept: The difference of two solutions to is a solution to .

(a) Find two solutions to :

From and , we get:

So is a solution to .

Similarly, is another solution to .

Since both and satisfy , they are both solutions. Note that , so they are not independent unless (which would mean , a unique solution).

Alternative answer: More generally, any scalar multiple for is a solution to .

(b) Find another solution to :

The complete solution to has the form:

where is a particular solution and are solutions to .

We can use and add any solution from the null space. For instance:

Verify: ✓

Answer: (a) Two solutions to are: and

2. Another solution to is: (or for any scalar )

4.20. Conditions for Solvability and Solution Structure (Tutorial 3, Task 5)

What conditions on make each system solvable? Solve for :

Click to see the solution

Key Concept: A system is solvable if and only if is in the column space of . We find this by row reducing the augmented matrix.

First System:

1. Row reduce the augmented matrix:

   After row operations:

   Further reducing:

2. Solvability conditions: For consistency:

3. Solution (when solvable): ,

Second System:

1. Row reduce the augmented matrix:

   After row reduction (details omitted):

2. Solvability conditions:

3. Solution (when solvable): With free variable :

Answer: First system is solvable iff and . Solution:

Second system is solvable iff and . Solution:

4.21. Write Complete Solutions as (Tutorial 3, Task 6)

Write the complete solutions to these systems:

Click to see the solution

Key Concept: Complete solution = particular solution + null space solution: .

First System:

1. Row reduce the augmented matrix:

2. Find particular solution: From row 2: From row 1:

   Choose :

   Particular solution:

3. Find null space (special solution): The null space comes from with free variable .

   Set : . Special solution:

4. Complete solution:

Second System:

1. Row reduce the augmented matrix:

2. Check consistency: Row 2 says , which is impossible.

   This system has no solution.

Answer: First system:

Second system: No solution (inconsistent)

4.22. Find Unsolvable System with Infinite Solutions in Null Space (Tutorial 3, Task 7)

Write a system with many solutions in the null space but no solution to . (Therefore the system has no solution.) Which ’s allow a solution?

Click to see the solution

Key Concept: For a system to be unsolvable yet have a non-trivial null space, we need a rank-deficient matrix with not in the column space.

1. Construct the system: Let (rank 1, second row is 2 times the first)

   The column space is (the line through )

   The null space is (infinite solutions to )

2. Choose not in Col: Let (not a multiple of )

   Check: gives:

   From the first equation: Substitute into the second: ✗

   No solution, but the null space has infinite vectors.

3. Which ’s allow a solution? Only for any scalar .

   These are vectors in the column space of .

Answer: Example system:

This system has no solution, but contains infinitely many vectors:

For the system to be solvable, must satisfy: for some scalar .

4.23. Find Matrix from Complete Solution (Tutorial 3, Task 8)

Find the matrix such that the complete solution to is:

Click to see the solution

Key Concept: The complete solution tells us the particular solution and the null space. From these, we can reconstruct .

1. Interpret the solution:

   • Particular solution:
   • Null space direction:

2. Find constraints on from the null space: If is in , then:

   If , then:

   So and :

3. Use the particular solution: :

   So and :

4. Verify: Complete solution to :

   Particular solution: ✓

   Null space: , with free , so ✓

Answer:

4.24. Find Echelon Form and Complete Solution (Tutorial 3, Task 9)

Find the echelon form , the free variables, and the special solutions:

is consistent (has a solution) when satisfies ______. Find the complete solution in the form .

Click to see the solution

Key Concept: Echelon form reveals pivots, free variables, and solvability constraints.

1. Row reduce to echelon form:

   Note: Column 1 has no pivot (skip it).

   :

2. Echelon form :

3. Solvability condition: For consistency: , i.e.,

4. Pivot and free variables:

   • Pivot columns: column 2 (variable )
   • Free variables:

5. Particular solution (set free variables to 0): From row 1:

6. Special solutions (null space): From row 1:

   • Set : , giving
   • Set : , giving
   • Set : , giving

7. Complete solution:

Answer: Echelon form:

Solvability condition:

Free variables:

Complete solution:

4.25. Find Invertible Submatrix from Rank (Tutorial 3, Task 10)

If has rank , then it has an submatrix that is invertible. Find that submatrix from the pivot rows and pivot columns of each :

Click to see the solution

Key Concept: The submatrix formed by the intersection of pivot rows and pivot columns is invertible.

Matrix 1:

1. Row reduce:

2. Identify pivots:

   • Row 1, Column 1: pivot
   • Row 2, Column 3: pivot Rank = 2

3. Submatrix : Take rows 1 and 2, columns 1 and 3:

   Verify: ✓

Matrix 2:

1. Row reduce:

2. Identify pivots:

   • Row 1, Column 1: pivot Rank = 1

3. Submatrix : Take row 1, column 1:

   Verify: ✓

Matrix 3:

1. Already in echelon form:

   • Row 1, Column 2: pivot
   • Row 3, Column 3: pivot Rank = 2

2. Submatrix : Take rows 1 and 3, columns 2 and 3:

   Verify: ✓

Answer:

• Matrix 1: (rank 2)
• Matrix 2: (rank 1)
• Matrix 3: (rank 2)