YOUR NEXT DISCOVERY

Linear Algebra: New Coordinates, Best Fits and Eigenvectors

Find the closest fit, change your measuring directions and uncover the directions a matrix keeps.

Complete the Course to Earn a Certificate and Join the Hall of Fame

An inviting coordinate workshop: two tactile measuring arrows cross a tilted mint grid, with a golden target marker above it. Decorative motion does not represent measured coordinates.
A target arrow and its closest point on a diagonal line. The dashed residual joins the projected tip to the target. The pulse follows this gap without moving either endpoint.

A CLOSER LOOK

Find the missing directions

Fit imperfect measurements, choose useful coordinate systems and predict repeated linear changes.

See it. Try it. Make it yours.

  1. Find the missing directions

  2. Find the closest possible answer

  3. Change the coordinate labels

Learners who can use vectors, matrices and linear equations. · What to know before you start

Bring with you: Add, subtract, multiply and divide signed numbers and fractions. Solve linear equations and use square roots. Polynomial expansion and simple quadratic factoring are taught in this course. Read vectors in two or three coordinates. Add and scale vectors, calculate dot products and lengths, and use span, linear independence, bases and dimension. Multiply matrices and vectors, solve systems by row reduction, and find nullspaces, column spaces, row spaces and rank. Use transpose, inverse and determinant, and recognise a linear transformation.

Fit imperfect measurements, choose useful coordinate systems and predict repeated linear changes.

Construct, calculate, diagnose and justify projections, least-squares fits, coordinate changes, orthonormal bases and real eigenspaces, then combine them in an independent investigation.

By the end, you can…

  • Build orthogonal complements and separate a vector into two perpendicular parts.
  • Construct projections and use least squares when exact equations cannot all be satisfied.
  • Translate coordinates between ordered bases and represent the same transformation in a new basis.
  • Use orthonormal bases and construct them with the Gram–Schmidt process.
  • Find eigenvalues and eigenspaces, decide whether an eigenbasis exists, and use it for repeated transformations.
  • Choose and check a fit, a coordinate system and a repeated-change model on new data.

For another Course

  • Complex-coordinate eigenproblems, infinite-dimensional spaces, singular value decomposition and numerical stability algorithms.

YOUR LEARNING PATH

One idea leads to the next.

0 of 37 Missions completed or proven
Unit 1

Find the missing directions

Build orthogonal complements and separate a vector into two perpendicular parts.

  1. Find every perpendicular directionMission 1 of 37Locked. Test whether a vector is perpendicular to a whole subspace.
  2. Keep every valid perpendicular arrowMission 2 of 37Locked. Explain why an orthogonal complement is a subspace.
  3. Count the directions left overMission 3 of 37Locked. Find the dimension of an orthogonal complement in its stated ambient space.
  4. Split one arrow into two partsMission 4 of 37Locked. Construct and justify the unique perpendicular decomposition of a vector.
  5. Come back to the original spaceMission 5 of 37Locked. Explain the double-complement rule and use row/nullspace complement pairs.
  6. Choose the shortest exact solutionMission 6 of 37Locked. Find the unique rowspace solution of a consistent system and explain its minimum length.
Unit 2

Find the closest possible answer

Construct projections and use least squares when exact equations cannot all be satisfied.

  1. Drop an arrow onto a lineMission 7 of 37Locked. Construct a line projection and test the perpendicular residual.
  2. Land on a tilted planeMission 8 of 37Locked. Project onto a plane through the origin using a normal vector.
  3. Build a projection machineMission 9 of 37Locked. Build and justify a projection matrix from an independent column basis.
  4. Prove the landing point is closestMission 10 of 37Locked. Use a perpendicular split to compare the projection with every other point in the subspace.
  5. Fit readings that disagreeMission 11 of 37Locked. Choose a least-squares objective and interpret residuals when exact equations are inconsistent.
  6. Make every column pass the residual testMission 12 of 37Locked. Derive and solve normal equations, then distinguish model coefficients from predictions.
  7. Build a line from measured pointsMission 13 of 37Locked. Translate a straight-line fitting problem into a design matrix and assess the fitted model.
  8. Separate a unique fit from unique settingsMission 14 of 37Locked. Explain why least-squares predictions can be unique when coefficient vectors are not.
Unit 3

Change the coordinate labels

Translate coordinates between ordered bases and represent the same transformation in a new basis.

  1. Give an arrow new coordinate labelsMission 15 of 37Locked. Find coordinates relative to an ordered basis while keeping the vector itself unchanged.
  2. Translate coordinates in both directionsMission 16 of 37Locked. Build the basis matrix and use its inverse in the correct direction.
  3. Use fewer labels inside a larger spaceMission 17 of 37Locked. Find coordinates in a lower-dimensional subspace without inventing a square inverse.
  4. Describe the same machine in new coordinatesMission 18 of 37Locked. Construct the matrix of a transformation in a chosen basis.
  5. Choose coordinates that make the job easierMission 19 of 37Locked. Recover a standard transformation from its action on a useful basis.
Unit 4

Build perpendicular measuring tools

Use orthonormal bases and construct them with the Gram–Schmidt process.

  1. Build measuring arrows of length oneMission 20 of 37Locked. Distinguish orthogonal from orthonormal bases and normalise nonzero vectors.
  2. Read coordinates with dot productsMission 21 of 37Locked. Find orthonormal coordinates and explain why dot products recover the weights.
  3. Project with perpendicular columnsMission 22 of 37Locked. Build QQᵀ and distinguish rectangular orthonormal-column matrices from square orthogonal matrices.
  4. Keep lengths and angles unchangedMission 23 of 37Locked. Explain orthogonal-matrix length and angle preservation and use an orthogonal change of basis.
  5. Remove an overlap before measuringMission 24 of 37Locked. Perform two-vector Gram–Schmidt while preserving the span.
  6. Remove every earlier overlapMission 25 of 37Locked. Extend Gram–Schmidt to three independent vectors and check the final basis.
  7. Catch a broken basis recipeMission 26 of 37Locked. Diagnose dependent inputs and explain which Gram–Schmidt outputs can change with order.
  8. Build three directions in four coordinatesMission 27 of 37Locked. Carry out Gram–Schmidt in R⁴ and distinguish subspace dimension from coordinate count.
Unit 5

Find directions a matrix keeps

Find eigenvalues and eigenspaces, decide whether an eigenbasis exists, and use it for repeated transformations.

  1. Find an arrow that keeps its lineMission 28 of 37Locked. Test eigenvectors and interpret positive, negative and zero eigenvalues.
  2. Find the numbers that make a product zeroMission 29 of 37Locked. Expand two linear factors and solve simple factored quadratic equations.
  3. Find the possible stretch factorsMission 30 of 37Locked. Derive and solve the characteristic equation for a two-by-two matrix.
  4. Find every arrow for one factorMission 31 of 37Locked. Solve for an eigenspace and distinguish its basis from its nonzero eigenvectors.
  5. Read three-dimensional eigenvalue cluesMission 32 of 37Locked. Find and check eigenvalues and eigenspaces of a three-by-three matrix.
  6. Find hidden directions in a full matrixMission 33 of 37Locked. Find eigenvalues and eigenspaces in a nontriangular three-by-three example.
  7. Turn a matrix into separate stretchesMission 34 of 37Locked. Construct a diagonalisation from an eigenbasis and justify the matrix identities.
  8. Know when diagonal coordinates do not existMission 35 of 37Locked. Distinguish repeated roots, insufficient eigendirections and absence of real eigenvalues.
  9. Predict many steps at onceMission 36 of 37Locked. Use an eigenbasis to compute repeated matrix action and interpret mode growth.
Unit 6

Run a complete investigation

Choose and check a fit, a coordinate system and a repeated-change model on new data.

  1. Choose and defend a complete modelMission 37 of 37Locked. Independently connect fitting, coordinate changes and repeated matrix action, with checks and limits.Mastery Challenge

A Debsie Course · Learners who can use vectors, matrices and linear equations.