Blog

How to define “no worse anywhere” and recognize incomparable alternatives.

How to compare algorithm settings using validation, and why the test must not become an unlimited hint.

How a squared-norm penalty changes an objective and its optimizer.

Why fitting accidental details of the training sample can improve familiar answers while hurting new ones.

What happens when the chosen model cannot express a useful pattern even on familiar examples.

How nearby labelled examples vote on a new object, and what changes when the neighbour count changes.

What separates regions with different predicted labels, and why a boundary need not pass through training points.

How feature weights and a bias define a linear boundary, and how the sign of one score selects a class.

Why fitting parameters, choosing a model, and final evaluation need different data roles.

How error indicators become a misclassification rate, and how a sample result differs from error on new data.

How an object becomes features, what an observed label means, and what a classifier predicts for a new object.

How averaging products of deviations describes joint linear variation.

Why dividing by n and by n−1 answers different questions.

How to measure spread without positive and negative deviations cancelling.

How to average possible values using their probabilities as weights.

How a rule assigning numbers differs from an observed value.

When learning one event does not change the probability of another.

How new information changes which outcomes remain under consideration.

How an event inherits probability from its outcomes.

How observed values form an average, and why that differs from a distribution’s expectation.

How comparing the middle element removes half a sorted array without losing a possible answer.

How two constant-factor bounds establish an order without numerical equality.

How to show a function eventually stays above a chosen growth scale.

How a constant factor and a threshold define an asymptotic upper bound.

Why a ratio tending to zero is stronger than an ordinary upper bound.

When two functions differ by a vanishing relative fraction.

How sums over small pieces lead to a definite integral.

Why a descent direction still needs a suitable step length.

How partial derivatives combine into a local linear model.

How to measure one variable’s local effect while holding the others fixed.

How a limit of difference quotients becomes a local linear approximation.

Why a limit concerns nearby values rather than the value at the point itself.

What it means for every sufficiently late term to be close enough.

Matching shapes, multiplying a row against a column, and adding contributions.

How rows become columns while entries are preserved and shape changes.

Rows, columns, shape, and reading an entry through two indices.

Multiplying corresponding coordinates while keeping a separate result in every position.

The length axioms, choosing a ruler, and three norms of the same vector.

Pairwise coordinate products are added into one signed scalar result.

Why coordinate order and dimension are part of a vector’s meaning.

How to identify an upper edge without requiring attainment.

How to describe a lower edge even when no value attains it.

How to return the indices of top scores while preserving ties.

How to identify every input achieving the smallest function value.

How an upper bound becomes a maximum only when attained.

Why a lower bound must also be a value at an admissible input.

Turning checks into zeros and ones to count successes and calculate their proportion.

Finding every input that reaches a target set, even without an inverse function.

When an output uniquely recovers its input, and why an inverse function is not a reciprocal.

Feeding one function’s output into another, and why order changes the answer.

The domain, codomain, and actual outputs of a single-valued rule.

Open and closed endpoints, fractional interior points, and infinite boundaries.

Every combination of two sets, with distinct roles for the first and second positions.

How to partition a union into “only A,” “only B,” and the shared part without losing or double-counting elements.

How to distribute union and intersection across parentheses, and why both identities hold.

Why negation exchanges union and intersection, and how element membership proves both identities.

Why “everything outside a set” is meaningful only after its universe has been fixed.

How to remove the second set’s elements from the first, and why reversing the arguments changes the result.

How to keep only shared elements, and why a family intersection requires membership in every set.

Why membership in any one set is enough for a union, including a whole indexed family.

How a candidate domain and a predicate define a set of qualifying elements.

Counting distinct elements, ignoring duplicates, and distinguishing two kinds of emptiness.

How to check that every element of one set is in another, and why inclusion differs from membership.

How to prove set equality element by element or through two inclusions, and why equal size is not enough.

What membership states, why order and repetition do not change a set, and how an empty set differs from a set containing it.

An arbitrary case instead of a sample: examples, proofs, and counterexamples.

Witnesses of existence, the difference from uniqueness, and empty domains.

What for every means, how a counterexample works, and why choice order matters.

When conditions can replace one another, and why both directions need proof.

Sufficient and necessary conditions, and the only way an if–then promise can fail.

Inclusive or: one successful condition is enough, and two are allowed.

Logical and: both checks must pass for the same candidate.

Reversing a condition’s truth without losing the equality boundary.

Why a condition needs an argument, and how substitution produces a truth value.

Counting selections while removing different orders of the same chosen set.

Why a product of consecutive integers counts arrangements of distinct objects.

Indexed factors, the identity value one, and the difference between products and sums.

Indices, inclusive bounds, and accumulating the result of a finite sum.

The unique decomposition a = qm + r, its remainder range, and negative inputs.

The least integer above a value: counting pages, including the partly filled last one.

The greatest integer below a value, and why floor differs from truncation for negative inputs.

Finding an exponent, keeping the base explicit, and checking the product rule.

How a root selects a value, and why a square equation can have two solutions.

Bases, zero and negative exponents, and why parentheses change the result.

Why two increasing quantities need not be proportional.

How to state an error tolerance and decide whether an approximation is useful.

Distance from zero: negative and positive points can share an absolute value.

The number line, irrational values, and the difference between exact numbers and approximations.

Numbers as ratios of integers: equivalent fractions and the nonzero denominator requirement.

Positive and negative integers, zero, and the limits of integer division.

Which values count discrete objects, and why the zero convention must be stated.