Linear Algebra Study Notes

Goal of the Study

  • Understanding of linear transformations, their matrix representations, determinants, eigenvalues/eigenvectors, diagonalization, matrix exponentials, orthogonal projections, least squares, and applications.

Prerequisites

  • Familiarity with basic vector space concepts:

    • Vector addition

    • Scalar multiplication

    • Basis

    • Dimension

    • Span

    • Linear independence

  • Matrix operations:

    • Addition

    • Multiplication

    • Transpose

    • Inverse

1. Linear Transformations: The Core Concept

  • Definition: Linear transformations are functions between vector spaces that preserve structure, meaning:
    T(av+bw)=aT(v)+bT(w)T(av + bw) = aT(v) + bT(w)
    for any vectors $v, w$ in vector space $V$ and scalars $a, b$ in $R$.

  • Key Property: Maps zero vector of input to zero vector of output: T(0<em>V)=0</em>WT(0<em>V) = 0</em>W. Non-linear example: $g(x) = 2x - 2$, since $g(0) ≠ 0.</p></li></ul><h4id="0c9d2225341d45149b55461fa96ba6c5"datatocid="0c9d2225341d45149b55461fa96ba6c5"collapsed="false"seolevelmigrated="true">ExamplesofLinearTransformations</h4><ol><li><p><strong>Example1</strong>:.</p></li></ul><h4 id="0c9d2225-341d-4514-9b55-461fa96ba6c5" data-toc-id="0c9d2225-341d-4514-9b55-461fa96ba6c5" collapsed="false" seolevelmigrated="true">Examples of Linear Transformations</h4><ol><li><p><strong>Example 1</strong>:T(x) = 3x</p><ul><li><p></p><ul><li><p>T(ax + by) = 3(ax + by) = aT(x) + bT(y)

  • Example 2: If $A$ is an $m imes n$ matrix,

    • T(v) = Av definesalineartransformationfromdefines a linear transformation fromR^n$ to $R^m.</p></li></ul></li><li><p><strong>Example3</strong>:Differentiationmap.</p></li></ul></li><li><p><strong>Example 3</strong>: Differentiation mapD: Pn o P{n-1}, D(p(t)) = p'(t).</p></li><li><p><strong>Theorem49</strong>:Alineartransformationisdeterminedbyitseffectonbasisvectorsofinputspace.</p></li></ol><h3id="19131f138b494fa68bf0dd13f8a16480"datatocid="19131f138b494fa68bf0dd13f8a16480"collapsed="false"seolevelmigrated="true">2.CoordinateMatrices:RepresentingLinearTransformations</h3><ul><li><p><strong>StandardBasis</strong>:For.</p></li><li><p><strong>Theorem 49</strong>: A linear transformation is determined by its effect on basis vectors of input space.</p></li></ol><h3 id="19131f13-8b49-4fa6-8bf0-dd13f8a16480" data-toc-id="19131f13-8b49-4fa6-8bf0-dd13f8a16480" collapsed="false" seolevelmigrated="true">2. Coordinate Matrices: Representing Linear Transformations</h3><ul><li><p><strong>Standard Basis</strong>: ForT: R^n o R^m, standard matrix $A$ (often written $T_{E,E}$ with $E$ as basis) is formed from:

      • Columns are given by T(e1), T(e2), …, T(e_n)

    • General Bases:

      • For bases B in $V$ and C in $W$:

      • Coordinate Matrix T{C,B} is an $m imes n$ matrix from [T(b1)]C, …, [T(bn)]_C.</p></li></ul></li></ul><h3id="8cbde86465724863a2ae29f3d145ef0e"datatocid="8cbde86465724863a2ae29f3d145ef0e"collapsed="false"seolevelmigrated="true">3.Determinants:ASpecialNumberAssociatedwithSquareMatrices</h3><ul><li><p><strong>Definition</strong>:Determinantsprovideascalarvaluecalculatedfrommatrixentries.</p></li><li><p><strong>KeyProperty</strong>:</p><ul><li><p>.</p></li></ul></li></ul><h3 id="8cbde864-6572-4863-a2ae-29f3d145ef0e" data-toc-id="8cbde864-6572-4863-a2ae-29f3d145ef0e" collapsed="false" seolevelmigrated="true">3. Determinants: A Special Number Associated with Square Matrices</h3><ul><li><p><strong>Definition</strong>: Determinants provide a scalar value calculated from matrix entries.</p></li><li><p><strong>Key Property</strong>:</p><ul><li><p>det(A)
        eq 0 indicates that matrix $A$ is invertible (Theorem 54).

    • Properties:

      • det(I_n) = 1</p></li><li><p>Rowreplacementleavesdeterminantunchanged</p></li><li><p>Rowinterchangemultipliesdeterminantby1</p></li><li><p>Rowscalingby</p></li><li><p>Row replacement leaves determinant unchanged</p></li><li><p>Row interchange multiplies determinant by -1</p></li><li><p>Row scaling bysmultipliesdeterminantbymultiplies determinant bys.

    Calculation

    • Use row operations to obtain triangular form. The determinant of a triangular matrix is the product of diagonal entries.

    • Use cofactor expansion for calculations with small matrices.

    4. Eigenvalues and Eigenvectors

    • Definition: An eigenvector of matrix $A$ is a non-zero vector $v$ such that Av = 4 vforsomescalarfor some scalar4. The vector $v$ only changes in magnitude.

    Finding Eigenvalues

    • Rewrite Av - 4 I v = 0leadstononzerosolutions.</p></li><li><p>Condition:leads to non-zero solutions.</p></li><li><p>Condition:det(A - 4 I) = 0(Theorem58).</p></li></ul><h3id="26aaf89fbcbf49c39159508a075ec232"datatocid="26aaf89fbcbf49c39159508a075ec232"collapsed="false"seolevelmigrated="true">5.Diagonalization</h3><ul><li><p><strong>Definition</strong>:AmatrixAisdiagonalizableif(Theorem 58).</p></li></ul><h3 id="26aaf89f-bcbf-49c3-9159-508a075ec232" data-toc-id="26aaf89f-bcbf-49c3-9159-508a075ec232" collapsed="false" seolevelmigrated="true">5. Diagonalization</h3><ul><li><p><strong>Definition</strong>: A matrix A is diagonalizable ifA = PDP^{-1}whereDisdiagonal(Theorem66).</p><ul><li><p>Existenceofaneigenbasis(collectionofeigenvectors)crucial.</p></li><li><p>Todiagonalize:</p></li><li><p>Findeigenvalues,determineeigenvectors,constructmatricesPandD.</p></li></ul></li></ul><h3id="f7589d708b6d4667baeb19abce4c7107"datatocid="f7589d708b6d4667baeb19abce4c7107"collapsed="false"seolevelmigrated="true">6.LinearDifferentialEquations</h3><ul><li><p>Solvewhere D is diagonal (Theorem 66).</p><ul><li><p>Existence of an eigenbasis (collection of eigenvectors) crucial.</p></li><li><p>To diagonalize:</p></li><li><p>Find eigenvalues, determine eigenvectors, construct matrices P and D.</p></li></ul></li></ul><h3 id="f7589d70-8b6d-4667-baeb-19abce4c7107" data-toc-id="f7589d70-8b6d-4667-baeb-19abce4c7107" collapsed="false" seolevelmigrated="true">6. Linear Differential Equations</h3><ul><li><p>Solve rac{du}{dt} = AuwithgiveninitialconditionsusingtheeigenvectorsofA.</p></li><li><p><strong>Solution</strong>:Theuniquesolutionwith given initial conditions using the eigenvectors of A.</p></li><li><p><strong>Solution</strong>: The unique solutionu(t) = e^{At}v, is simplified by discoveries of eigenvectors/eigenvalues, yielding exponential growth/decay behavior.

    7. Orthogonality and Projection

    • Orthogonal Projection: Projects vectors onto subspaces preserving linear structure.

    • Understand $projw(v)$ and the projection matrix $Pw$.

    • Calculation: P_w = rac{1}{w ullet w} ww^T.</p></li></ul><h3id="9f07b667d03042b28ab4bd95b170716d"datatocid="9f07b667d03042b28ab4bd95b170716d"collapsed="false"seolevelmigrated="true">8.LeastSquaresSolutions</h3><ul><li><p>Thegoalistominimizedistanceforsystemswheresolutionsdontexist.</p></li><li><p><strong>NormalEquations</strong>:.</p></li></ul><h3 id="9f07b667-d030-42b2-8ab4-bd95b170716d" data-toc-id="9f07b667-d030-42b2-8ab4-bd95b170716d" collapsed="false" seolevelmigrated="true">8. Least Squares Solutions</h3><ul><li><p>The goal is to minimize distance for systems where solutions don’t exist.</p></li><li><p><strong>Normal Equations</strong>:A^T(b - Ax) = 0leadstoleads toA^TAx = A^T b.</p></li><li><p>Solveforleastsquaressolutionwhen.</p></li><li><p>Solve for least-squares solution whenA$$ has full column rank.

    9. Linear Regression

    • Fit linear models to data points using least squares, forming appropriate matrices to calculate coefficients.

    Markov Matrices

    • Inform about the transitions with the properties of columns summing to one. This aids in analysis pertaining to long-term behavior.

    Summary

    • Linear operations define, analyze, and simplify complex systems through structured approaches,

      • Applications in solving equations, data fitting, response modeling, and understanding dynamic systems. Good luck with the quiz!