No grades. No wrong moves.
The playground
Nineteen stations from across the curriculum — calculus, linear algebra, statistics, optimization, signals — each one idea made touchable. Drag things. Break nothing. Intuition built here makes every lesson land harder.
Station 01
Ride the wave
Slope is a living thing: positive on climbs, zero at crests, negative on descents.
Station 02
Shrink the gap
The derivative being born: averages over a closing gap settling on the instant.
Station 03
Slice and stack
Riemann sums in your hands — rectangles melting into exact area.
Station 04
Function vs. derivative
Two graphs, one truth: the bottom one's height is the top one's slope.
The function
Its derivative — the slope, graphed
Station 05
The moving cart
Position, velocity, acceleration — one slider, three linked stories.
Position s(t) — where it is
Velocity v(t) = s′(t) — how fast
Acceleration a(t) = v′(t) — how the speed changes
Station 06
Find the peak
Optimization by feel: walk uphill until the slope reads zero.
Station 07
The +C family
Infinitely many originals, one shared rate — why antiderivatives come with a constant.
Station 08
The grand connection
Watch accumulation grow at exactly the rate of the curve above — the fundamental theorem, live.
A rate f(t) — area shading in
Total so far A(x) — the area, graphed
Station 09
The circle that makes waves
Walk the unit circle and watch your height trace a sine wave — trigonometry's one essential picture.
Station 10
Move all of space
Rotate, stretch, squash, shear: a matrix is a verb, and the grid is its sentence.
Station 11
The bell curve
Slide the mean, stretch the spread, and watch the 68–95–99.7 rule count its share.
Station 12
Euler's bridges
The 1736 puzzle that founded network theory — toggle bridges and let degree-counting decide the walk.
Station 13
Square waves from circles
Fourier's scandalous claim, performed: stack smooth harmonics until corners appear.
Station 14
Catch the impostor
Sample a wave too slowly and a phantom frequency appears — the aliasing that haunts all digital audio.
Station 15
Compress with SVD
Rebuild a data matrix from its strongest rank-1 layers — the low-rank approximation behind PCA, compression, and recommendations.
Rank-2 approximation
Full data (rank 8)
singular values σ₁ … σ₈ — kept layers are solid
Station 16
Roll downhill
Tune the learning rate and watch gradient descent converge, crawl, or diverge — exactly how models learn.
Station 17
Count without listing
The fundamental counting principle as a branching tree, and permutations vs combinations as filling slots.
Station 18
The bell from chaos
Average samples from any lopsided population and watch the sample mean go normal — the central limit theorem, live.
distribution of the sample mean
Station 19
3D Loss Landscapes & Gradients
Rotate surfaces in 3D and watch the gradient vector ∇f point along the path of steepest climb or descent.