a working notebook
Mathematics,
explained the way it's used -
not the way it's taught.
Mathematics is the study of structure, quantity, and change - a language for describing patterns precisely enough that a machine can act on them. This page is a single-sitting map of what it is, why it still matters after calculators and AI, and how to actually learn it if you're coming at it through code.
What Is Mathematics
At its core, mathematics is the discipline of drawing certain conclusions from clearly stated assumptions. It has no lab, no telescope, no sensor - its raw material is pure logic applied to abstractions like numbers, shapes, and relationships. That abstraction is precisely what makes it portable: the same equation that describes a bouncing ball describes an AC current, a population of rabbits, and a stock price, because all four share the same underlying structure.
In practice it splits into a few families you'll keep running into: algebra (solving for unknowns), geometry & trigonometry(shape and angle), calculus (rates of change), linear algebra (systems and transformations), probability & statistics(uncertainty), and discrete math (the logic, sets, and graphs that computer science is built from).
Why It Matters Now
It would be reasonable to assume calculators and AI made deep math skill optional. The opposite happened - the tools got powerful enough that math literacy became the difference between using a tool and understanding what it just told you.
Every algorithm is math wearing a trench coat
Search, sort, hash, route, rank — each one is a mathematical statement before it's a function. You can't reason about Big-O without a little bit of algebra and limits.
Machine learning is applied linear algebra + calculus
Gradient descent is a derivative. A neural network forward-pass is matrix multiplication. Loss landscapes are multivariable functions. Skip the math, and the model stays a black box.
Security runs on number theory
RSA, Diffie–Hellman, elliptic curves — the internet's locks are built from prime numbers and modular arithmetic invented centuries before computers existed.
Graphics, games, and physics engines are geometry in motion
Rotations, projections, collision detection, easing curves — all trigonometry and vectors, frame by frame, sixty times a second.
Probability is how systems survive uncertainty
A/B tests, recommendation systems, fraud detection, load balancing — all decisions made under uncertainty, which is exactly what probability was built to formalize.
Without It
It's easier to see why math matters by subtracting it. Pull mathematics out of modern life and here's roughly what goes with it:
- ×No GPS — triangulating a position from satellites is trigonometry and relativity-corrected timing.
- ×No encryption — banking, messaging, and passwords all lean on number theory that took mathematicians centuries to develop.
- ×No compression — JPEG, MP3, and ZIP all rely on transforms and probability to shrink data without losing what matters.
- ×No search engines — ranking a trillion pages is a linear algebra problem (eigenvectors, literally what PageRank is).
- ×No structural engineering — bridges and skyscrapers stand because someone solved differential equations correctly.
- ×No medicine at scale — drug trials, dosages, and epidemiology are statistics; get the math wrong and people die.
How It Helps
It builds a transferable mental model
Learn recursion in math (a sequence defined in terms of itself) and recursive functions in code stop being a trick - they're the same idea in a different syntax. Learn what a derivative means and gradient descent reads like a sentence instead of a spell.
It turns debugging into diagnosis
An off-by-one bug, a NaN from dividing by a near-zero number, an infinite loop from a bad recurrence - these stop looking random once you can name the mathematical object misbehaving underneath the code.
It sharpens estimation
Knowing that O(n²) at a million rows means a trillion operations lets you reject a bad design before writing a line of it - no profiler required.
It's the shared vocabulary of every technical field
Physics, economics, biology, and engineering all speak math natively. Learning it once lets you read across fields you'll never formally study.
Learning With AI
AI is a genuinely good study partner for math - patient, always available, and able to re-explain a step as many times as it takes. It's a poor substitute for doing the problem yourself. Used well, it looks like this:
- 01
Ask for the derivation, not just the formula
Instead of "what's the quadratic formula," ask an AI to derive it from ax²+bx+c=0 by completing the square. Watching the steps build the formula sticks far longer than memorizing it.
- 02
Request three explanations of the same idea
Ask for an algebraic explanation, a geometric picture, and a real-world analogy for the same concept. Triangulating a concept from three angles is how intuition actually forms.
- 03
Generate practice problems at your exact edge
Tell the AI what you got wrong and why, then ask for five problems that target that specific gap — not a random worksheet, a targeted one.
- 04
Teach it back, Feynman-style
Explain the concept to the AI in your own words and ask it to point out the first place your explanation breaks down. That crack is exactly what you don't understand yet.
- 05
Verify by hand before you trust it
AI can slip on arithmetic and edge cases. Use it to explain and generate, then re-derive the final two steps yourself on paper. The friction is the learning.
Tech Stack
The tools mathematicians and engineers actually reach for, grouped by job:
Numerical Computing
For crunching real numbers fast — vectors, matrices, solvers.
Symbolic Math
For exact algebra, calculus, and simplification — not decimals, symbols.
Visualization
For seeing a function before you trust it.
Machine Learning Math
Where linear algebra and calculus turn into predictions.
Formal Logic & Proof
For math that has to be provably correct, not just probably correct.
Learning with AI
For step-by-step derivations and on-demand practice sets.
Cheat Sheets
The formulas worth having memorized, organized by topic. Tap a topic to expand it.
- Quadratic formula
- x = (-b ± √(b² - 4ac)) / 2a
- Difference of squares
- a² - b² = (a - b)(a + b)
- Perfect square
- (a + b)² = a² + 2ab + b²
- Log rules
- log(xy) = log x + log y, log(x/y) = log x - log y
- Exponent rules
- aᵐ · aⁿ = aᵐ⁺ⁿ, (aᵐ)ⁿ = aᵐⁿ
Sketchbook
Most math concepts have a picture hiding inside them. Seeing the picture once tends to outlast memorizing the symbols by years.
Interview Puzzles
Classic math puzzles that show up in technical interviews - try each one before revealing the answer.
Formula Sheet
Every formula from this page, laid out line by line in one plain-text file - algebra through discrete math, the exact set that shows up while coding. Keep it open in a side pane while you work.
Keep In Mind
- 01Math is a language, not a trivia set — the goal is fluency in reading and writing it, not memorizing every formula forever.
- 02A formula without its proof is a fact you have to trust blindly. The proof is what lets you rebuild the formula if you forget it.
- 03Mistakes are data. Wrong answers usually point exactly at the misunderstood step — don't just check "right or wrong," check where it diverged.
- 04Every concept has at least three faces: algebraic (symbols), geometric (a picture), and numeric (plug in real numbers). Use whichever face makes it click.
- 05Struggle for real before looking up the answer. The retrieval attempt is what builds the memory — reading a solution passively barely does.
- 06Skipping fundamentals to reach "the interesting part" always costs more time later than it saves now.
- 07In code, an off-by-one error is almost always a math error — index arithmetic is discrete math whether you notice or not.
- 08Consistency beats intensity: twenty focused minutes a day compounds further than one exhausting six-hour session a month.