[Beginner] — putting numbers on the plane, so that every geometric question becomes an algebraic one.
Before you start
- You can solve and rearrange linear equations — linear equations.
- You know Pythagoras’ theorem, and the congruence results — triangles and polygons.
- Helpful: the circle theorems, which section 6 re-proves by algebra — circle theorems.
What you will be able to do after this lesson:
- Compute distance, midpoint and gradient between two points, and say which of Pythagoras’ assumptions each one inherits.
- Move fluently between the forms , point-gradient, two-point and , and say when each is the right one.
- Prove the perpendicular-gradient rule instead of memorising it, and state the case it excludes.
- Prove a geometric theorem by coordinates — reaching the same result as a synthetic proof, by different means.
Study route: read 1–6, attempt the prediction in section 5, then run the lab. Block 5 of the lab is the point of the lesson.
1. Why this exists
The previous four lessons proved things by argument: draw an auxiliary line, spot a congruence, chain the rules. It works, and it is beautiful, but it has an awkward property — you have to see the trick. The angle-sum proof needs someone to think of drawing a parallel through the apex. Nothing in the problem statement suggests it.
Descartes’ idea removes the need to be clever. Put an origin and two axes on the plane, and every point becomes a pair of numbers. Then:
- A line becomes an equation.
- “These two lines are perpendicular” becomes an arithmetic condition on two numbers.
- “This point lies on that circle” becomes substituting and checking.
Geometry becomes algebra, and algebra can be done by following rules — or by a machine. Every graphics program, every mapping application, every CAD package works this way. Nothing draws auxiliary lines.
The trade is real, though, and worth naming up front: coordinate proofs are mechanical but often longer, and they can obscure why a result is true. Both methods are worth having.
2. Terminology
| Term | Plain-English definition | Note |
|---|---|---|
| Origin | The point where the axes cross | Its position is a choice, not a fact about the figure |
| Coordinates | The pair locating a point | across, up — always that order |
| Gradient (slope) | How much changes per unit of | “Rise over run” |
| Intercept | Where a line crosses an axis | in is the -intercept |
| Collinear | Lying on one straight line | Three points are collinear when two gradients agree |
| Locus | The set of points satisfying a condition | A circle is the locus of points at distance from a centre |
3. Distance, midpoint, gradient
Take two points and .
Distance. Drop a horizontal and a vertical from the two points to form a right triangle with legs and . Pythagoras gives:
This is Pythagoras’ theorem wearing coordinates, and it inherits everything Pythagoras assumes — including that the axes are perpendicular and the scales on them equal. On a plot with a stretched -axis this formula does not measure what the picture shows.
Midpoint. The average of the coordinates:
Gradient.
with the standing exception that a vertical line has , so the gradient is undefined — not infinite, undefined. Every rule below that mentions gradients has vertical lines as an exception, and forgetting it is the most common source of a wrong answer in this topic.
Parallel and perpendicular
Parallel lines have equal gradients. Same rise per run, so the same .
Perpendicular gradients multiply to . Here is the proof, which is shorter than the rule is long-winded.
A line of gradient has direction vector — go one across, up. Similarly the other has direction . Two directions are perpendicular exactly when their dot product is zero:
The proof also shows the exception cleanly. The direction vector can never be vertical, so a vertical and a horizontal line — which are perpendicular — cannot be tested by this rule at all. Their gradients are undefined and , and the product is meaningless.
4. The equation of a straight line
Four forms, each convenient for different given information:
| Form | Equation | Use it when |
|---|---|---|
| Gradient-intercept | You know the gradient and where it crosses the -axis | |
| Point-gradient | You know one point and the gradient | |
| Two-point | You know two points | |
| General | You need to handle vertical lines, or want a formula |
The general form is the one to prefer in code. It is the only one of the four that can represent a vertical line (), and the distance formula below needs it.
Distance from a point to a line. For the line and the point :
The numerator measures how far the point is from satisfying the equation; dividing by converts that into an actual distance. This is the perpendicular distance — the shortest one — which is exactly what section 3.1 of tangents needed.
5. The circle
A circle is the locus of points at fixed distance from a centre . Writing that with the distance formula and squaring both sides:
That is the whole definition, turned into an equation with no further work. Expanding gives the general form , and completing the square takes you back.
TIP
Predict before running the lab. The line and the circle : how many points do they share? Substitute and look at the discriminant of the resulting quadratic before opening the answers. What would a discriminant of exactly zero mean geometrically?
6. Proving a theorem by coordinates
Section 3.1 of constructions needed this fact: the diagonals of a rhombus bisect each other at right angles. It was proved there by congruent triangles. Here it is by algebra.
Place the rhombus conveniently — and choosing the placement is the skill in coordinate proofs. Put one vertex at the origin and one side along the -axis:
For this to be a rhombus all four sides must be equal. and , so the condition is:
Now the diagonals are and . Their midpoints:
Identical — so the diagonals bisect each other, with no conditions at all. That half is true for any parallelogram.
Their gradients:
Multiplying:
And the rhombus condition says , so :
Perpendicular.
Notice what happened: the rhombus condition entered exactly once, at the last step. That tells you precisely which part of the theorem needs all four sides equal — the perpendicularity — and which part does not — the bisection. The synthetic proof does not make that separation nearly as obvious.
Worked example — runnable
Runnable example: save as coordinate_geometry.py in any empty directory and run python3 coordinate_geometry.py. Standard library only; writes no files.
"""Coordinate geometry, with a coordinate proof at the end.
Block 5 re-proves, numerically, the rhombus fact that the compass-and-
straightedge lesson proved by congruent triangles.
"""
import math
from fractions import Fraction
EPS = 1e-9
def dist(p, q):
return math.hypot(q[0] - p[0], q[1] - p[1])
def midpoint(p, q):
return ((p[0] + q[0]) / 2, (p[1] + q[1]) / 2)
def gradient(p, q):
"""Rise over run, or None for a vertical line."""
if abs(q[0] - p[0]) < EPS:
return None
return (q[1] - p[1]) / (q[0] - p[0])
def line_through(p, q):
"""General form (a, b, c) with ax + by + c = 0. Handles vertical lines."""
a = q[1] - p[1]
b = p[0] - q[0]
c = -(a * p[0] + b * p[1])
return (a, b, c)
def point_line_distance(pt, line):
a, b, c = line
return abs(a * pt[0] + b * pt[1] + c) / math.hypot(a, b)
def line_circle_intersections(m, k, r):
"""Where y = mx + k meets x^2 + y^2 = r^2. Returns the discriminant too."""
A = 1 + m * m
B = 2 * m * k
C = k * k - r * r
disc = B * B - 4 * A * C
if disc < -EPS:
return disc, []
if abs(disc) < EPS:
x = -B / (2 * A)
return disc, [(x, m * x + k)]
s = math.sqrt(disc)
return disc, [((-B + s) / (2 * A), m * ((-B + s) / (2 * A)) + k),
((-B - s) / (2 * A), m * ((-B - s) / (2 * A)) + k)]
if __name__ == "__main__":
A, B = (1.0, 2.0), (7.0, 10.0)
print("Block 1 - distance, midpoint, gradient")
print(f" A={A} B={B}")
print(f" distance = {dist(A, B):.6f} midpoint = {midpoint(A, B)} gradient = {gradient(A, B)}")
assert abs(dist(A, B) - 10.0) < EPS # the 6-8-10 triangle
assert midpoint(A, B) == (4.0, 6.0)
print()
print("Block 2 - perpendicular gradients multiply to -1")
for m1 in (2.0, -0.5, 3.0, 0.25):
m2 = -1 / m1
print(f" m1={m1:6.3f} m2={m2:8.4f} product={m1 * m2:.6f}")
assert abs(m1 * m2 + 1) < EPS
vertical = gradient((3.0, 1.0), (3.0, 9.0))
horizontal = gradient((1.0, 4.0), (8.0, 4.0))
print(f" vertical line gradient = {vertical}, horizontal = {horizontal}")
print(" these two ARE perpendicular, but the rule cannot express it")
assert vertical is None and horizontal == 0.0
print()
print("Block 3 - general form and distance from a point to a line")
line = line_through((0.0, 0.0), (4.0, 3.0)) # 3x - 4y = 0
print(f" line through (0,0) and (4,3): {line[0]}x + {line[1]}y + {line[2]} = 0")
for pt in [(0.0, 5.0), (4.0, 3.0), (8.0, 6.0), (-3.0, 4.0)]:
print(f" distance from {str(pt):>12} = {point_line_distance(pt, line):.6f}")
assert abs(point_line_distance((4.0, 3.0), line)) < EPS # on the line
assert abs(point_line_distance((0.0, 5.0), line) - 4.0) < EPS
print()
print("Block 4 - line meets circle: the discriminant decides")
for m, k, label in [(2.0, 1.0, "y = 2x + 1"),
(2.0, 5.0, "y = 2x + 5"),
(2.0, 6.0, "y = 2x + 6")]:
disc, pts = line_circle_intersections(m, k, math.sqrt(5))
kind = {0: "misses", 1: "tangent", 2: "secant"}[len(pts)]
print(f" {label:12} vs x^2+y^2=5 -> discriminant {disc:9.4f}, {len(pts)} point(s), {kind}")
assert len(line_circle_intersections(2.0, 1.0, math.sqrt(5))[1]) == 2
assert len(line_circle_intersections(2.0, 5.0, math.sqrt(5))[1]) == 1
assert len(line_circle_intersections(2.0, 6.0, math.sqrt(5))[1]) == 0
print()
print("Block 5 - coordinate proof: diagonals of a rhombus")
# Exact arithmetic, so 'perpendicular' means exactly -1, not nearly.
for p, q in [(Fraction(3), Fraction(4)),
(Fraction(5), Fraction(12)),
(Fraction(8), Fraction(15))]:
s = Fraction(int(math.isqrt(int(p * p + q * q)))) # side length, a whole number
assert p * p + q * q == s * s, "need a Pythagorean pair for exact arithmetic"
Av, Bv, Cv, Dv = (0, 0), (p, q), (p + s, q), (s, 0)
sides = {abs(s), abs(s)} # AD and BC
m_ac = Fraction(q, p + s)
m_bd = Fraction(q, p - s)
mid_ac = (Fraction(p + s, 2), Fraction(q, 2))
mid_bd = (Fraction(p + s, 2), Fraction(q, 2))
print(f" rhombus with p={p}, q={q}, side={s}")
print(f" midpoints coincide: {mid_ac == mid_bd}")
print(f" gradient product = {m_ac * m_bd}")
assert mid_ac == mid_bd
assert m_ac * m_bd == -1
print()
print("coordinate_geometry: passed")Expected output:
Block 1 - distance, midpoint, gradient
A=(1.0, 2.0) B=(7.0, 10.0)
distance = 10.000000 midpoint = (4.0, 6.0) gradient = 1.3333333333333333
Block 2 - perpendicular gradients multiply to -1
m1= 2.000 m2= -0.5000 product=-1.000000
m1=-0.500 m2= 2.0000 product=-1.000000
m1= 3.000 m2= -0.3333 product=-1.000000
m1= 0.250 m2= -4.0000 product=-1.000000
vertical line gradient = None, horizontal = 0.0
these two ARE perpendicular, but the rule cannot express it
Block 3 - general form and distance from a point to a line
line through (0,0) and (4,3): 3.0x + -4.0y + -0.0 = 0
distance from (0.0, 5.0) = 4.000000
distance from (4.0, 3.0) = 0.000000
distance from (8.0, 6.0) = 0.000000
distance from (-3.0, 4.0) = 5.000000
Block 4 - line meets circle: the discriminant decides
y = 2x + 1 vs x^2+y^2=5 -> discriminant 96.0000, 2 point(s), secant
y = 2x + 5 vs x^2+y^2=5 -> discriminant 0.0000, 1 point(s), tangent
y = 2x + 6 vs x^2+y^2=5 -> discriminant -44.0000, 0 point(s), misses
Block 5 - coordinate proof: diagonals of a rhombus
rhombus with p=3, q=4, side=5
midpoints coincide: True
gradient product = -1
rhombus with p=5, q=12, side=13
midpoints coincide: True
gradient product = -1
rhombus with p=8, q=15, side=17
midpoints coincide: True
gradient product = -1
coordinate_geometry: passed
Block 5 uses Fraction, not floating point, deliberately. The claim being checked is that the gradient product is exactly , and with floats the best you could report is . Exact arithmetic makes it a proof for those cases rather than strong evidence.
Common pitfalls and traps
- Forgetting vertical lines. Gradient undefined, unusable, perpendicular rule inapplicable. Any function taking a gradient needs a decision about this case; the lab returns
Noneand forces the caller to handle it. - Subtracting in inconsistent order. and agree. is the negative of both, and it is an easy slip.
- Using the distance formula on unequal axis scales. The formula assumes perpendicular axes with equal units. On a chart with a compressed axis, the visual distance and the computed distance are different things.
- Choosing an awkward placement in a coordinate proof. Putting a vertex at the origin and a side along an axis eliminates variables and shortens the algebra enormously. A general placement with four arbitrary vertices is valid but usually unmanageable — and this choice is where most coordinate proofs are won or lost.
- Over-specifying the placement. Placing the rhombus as proves the theorem for that rhombus only. The proof needs symbols with one stated condition, not a worked instance.
- Comparing floats with
==. Everywhere except block 5, the lab compares against a tolerance. Block 5 can use==only becauseFractionis exact.
Check your understanding
- Find the distance and midpoint between and .
- A line passes through with gradient . Give its equation in general form.
- Is the triangle with vertices , , right-angled? Justify with gradients.
- Find the perpendicular distance from to the line .
- A circle has equation . Find its centre and radius.
Answers — open only after an attempt
- Distance . Midpoint .
- , so , so .
- Gradients: to is ; to is ; to is . Now , , . No pair multiplies to , so it is not right-angled.
- .
- Complete the square: , so . Centre , radius .
And the prediction from section 5: substituting into gives , discriminant , so two intersection points. A discriminant of exactly zero would mean the two points have merged: the line is a tangent — the same repeated-root situation seen in the power-of-a-point lab.
Practice — independent task
Prove the midpoint theorem by coordinates, then verify it: the segment joining the midpoints of two sides of a triangle is parallel to the third side and half its length.
- Choose a placement that minimises variables. Justify your choice in a comment — how many variables did a general placement need, and how many did yours?
- Prove it symbolically on paper: show the gradients are equal and the length ratio is exactly .
- Implement
midpoint_segment(a, b, c)returning the two midpoints, and assert both properties numerically for at least five triangles, including one with a vertical side. - Use
Fractionso that “parallel” means gradients are exactly equal, and handle the vertical-side case explicitly rather than letting it raise. - Then answer: does your proof use the fact that the three points form a triangle at all? What happens if they are collinear?
Done when: your symbolic proof and your numerical check agree, the vertical case passes without special-casing the mathematics (only the representation), and you can answer step 5 with a reason rather than a guess.
Tradeoffs, limits and extensions
Coordinate proofs versus synthetic proofs. Coordinate proofs are mechanical: choose axes, write the algebra, grind. They always terminate, which synthetic proofs do not guarantee. But they can hide the reason — section 6’s algebra proves the rhombus fact without ever suggesting why equal sides should force perpendicular diagonals, whereas the congruent-triangle proof makes it visible. Prefer coordinates when you need certainty or a computer; prefer synthetic when you need insight.
The origin is a choice. Nothing in a triangle knows where your origin is. A good coordinate proof exploits that by placing the figure to zero out as many coordinates as possible. A poor one carries six arbitrary variables through pages of algebra to reach the same result.
Where this goes next. Adding a third coordinate gives lines and planes in space, where gradient stops working and direction vectors take over. Adding calculus turns “gradient of a line” into “gradient of a curve at a point”, which is the derivative.
Before moving on
You are done with this lesson when you can:
- Derive the distance formula from Pythagoras rather than recall it.
- Prove from direction vectors, and state the case it excludes.
- Convert between all four line forms, and say why the general form is the one to use in code.
- Complete a coordinate proof, choosing the placement yourself.
Recap for later lookup: ; midpoint is the average of coordinates; , undefined for vertical; parallel means equal , perpendicular means ; line forms , , ; distance from point to line ; circle .
Next: Trigonometric Ratios — coordinates give you lengths and gradients; trigonometry connects those to angles.
Related
- Constructions — the rhombus theorem proved the other way
- Circle Theorems — the same circles, by argument instead of equation
- Lines and Planes in Space — the three-dimensional continuation
- Vectors — the dot product used in the perpendicularity proof
Software engineering
Infrastructure & security
Computer science
Hardware & engineering