The experiment notebook

Change something.
See what follows.

Guided experiments connect a picture to a mathematical idea. Open a lesson to explore the controls, work an example, and test your prediction.

Guided experiment

Numbers that tell a story

Fractions, powers, roots, and logarithms are different ways to describe scale.

Guided experiment

Functions are machines

Follow an input through simple rules, then connect the rules together.

Guided experiment

Give a point an address

Build a coordinate from directions you can see, then turn the view in three dimensions.

Guided experiment

Read symbols as sentences

Turn sets, quantifiers, sums, and products into instructions you can follow.

Guided experiment

Change, one small step at a time

See derivatives as local slopes and integrals as accumulated contributions.

Guided experiment

How do you know it is true?

Separate a definition, an example, and a proof—and learn to break a false claim.

Guided experiment

The math behind a prediction

Meet the three toolkits—and learn what this course and its source actually cover.

Guided experiment

A field guide to the symbols

Translate every entry in the source’s notation table into words and a small example.

Guided experiment

Know what kind of thing you have

Scalars, vectors, matrices, and random variables play different roles—even when all of them involve numbers.

Guided experiment

Let dimensions catch mistakes

Before calculating, check whether the pieces can fit together.

Guided experiment

What can you build from two directions?

Explain how span, independence, and a basis fit together.

Guided experiment

Vectors as coordinates and displacements

Connect componentwise addition and scaling to a geometric displacement.

Guided experiment

Which sets keep every linear combination?

Test subspaces and explain when a sum has a unique decomposition.

Guided experiment

What must a linear map preserve?

Recognize a linear map and construct an isomorphism from bases.

Guided experiment

Why do matrix columns describe a map?

Recover a linear map from basis images and transpose products correctly.

Guided experiment

What disappears, and what can you reach?

Distinguish nullspace, range, row space, column space, and rank.

Guided experiment

How can distance define a limit?

Use the metric axioms and the epsilon–N definition of convergence.

Guided experiment

Does changing length change convergence?

Compare p-norms and explain the finite-dimensional norm equivalence statement.

Guided experiment

How does a product measure alignment?

Connect inner products, orthogonality, induced norms, and distance.

Guided experiment

How large can an inner product be?

Prove the orthogonal length identity and identify Cauchy–Schwarz equality cases.

Guided experiment

Why is the perpendicular foot the closest point?

Prove unique orthogonal decomposition and all eight projection properties.

Guided experiment

Which directions stay on the same line?

Identify eigenvectors and track eigenvalues under shifts, inverses, and powers.

Guided experiment

Which changes leave the diagonal sum alone?

Use trace linearity and cyclic invariance without assuming arbitrary reordering.

Guided experiment

When does a transformation flatten area?

Use the five determinant identities and distinguish area scale from length scale.

Guided experiment

Which matrices preserve length and angle?

Derive orthogonal matrix identities and prove preservation of inner products.

Guided experiment

Can a symmetric map be split into three simple steps?

Derive the spectral decomposition from an orthonormal eigenbasis.

Guided experiment

Which direction gives the largest quadratic value?

Prove the Rayleigh bounds and their equality conditions, including repeated eigenvalues.

Guided experiment

When is every nonzero direction uphill?

Prove the eigenvalue criteria and explain why a positive diagonal shift removes singularity.

Guided experiment

Why does a larger eigenvalue make a shorter ellipse axis?

Derive quadratic level sets from a sphere and identify their eigenvector directions.

Guided experiment

What does a matrix stretch?

Follow orthogonal changes of coordinates and singular-value scaling, including rectangular matrices.

Guided experiment

Which part of an input survives?

Decompose inputs and outputs into the four fundamental subspaces.

Guided experiment

How much can a matrix amplify a vector?

Compare induced norms, Frobenius size, and orthogonal invariance.

Guided experiment

How much can you discard?

Use truncated SVD and prove its spectral-norm optimality.

Guided experiment

Can you undo a map that loses information?

Construct the Moore–Penrose pseudoinverse and verify its four conditions.

Guided experiment

How do columns and outer products build a matrix?

Translate matrix products and quadratic forms into sums of entries.

Guided experiment

Where are you allowed to look for a minimum?

Distinguish feasible, local, global, and strict minima.

Guided experiment

Which direction changes the value fastest?

Read the gradient as a column of partial derivatives and a local direction of steepest change.

Guided experiment

What linear map describes a nonlinear change nearby?

Match Jacobian rows to outputs and columns to inputs.

Guided experiment

How does curvature change a step?

Use the Hessian to measure directional curvature and compare gradient and Newton steps.

Guided experiment

Why does a quadratic gradient contain a transpose?

Derive linear and quadratic gradients, including nonsymmetric matrices.

Guided experiment

How does a change travel through a composition?

Multiply Jacobians in the correct order and propagate a scalar gradient.

Guided experiment

How accurate is the local picture?

Distinguish first-order, integral-gradient, and second-order Taylor formulas.

Guided experiment

When does a stationary point really minimize?

Prove necessary and sufficient local-minimum conditions and test their limits.

Module 2 · coordinate scaling

Stretch a vector

Experiment 01 / Scalar multiplicationℝ²
A vector scaled by a real numberThe original vector is (2, 1). With scale 1, the result is (2.0, 1.0).x₁x₂0v = (2, 1)αv
αv = (2.0, 1.0)

A positive scale preserves the direction.

Read the full explanation →