Intuition What this page is
The parent note gave you the rule: compute B 2 − 4 A C , read the shape. But a rule only becomes yours when you have watched it survive every kind of input — every sign, the tricky zeros, the collapses (degenerate cases), the empty/imaginary curves, a rotated mess, a word problem, and an exam trap. Below we first lay out a matrix of every scenario the topic can throw at you, then work an example for each cell so you never meet a case you have not already seen.
Before we start, a quick reminder of the two tools, stated in plain words:
Definition The two numbers we compute
For A x 2 + B x y + C y 2 + D x + E y + F = 0 :
Discriminant Δ 2 = B 2 − 4 A C → tells the family (ellipse / parabola / hyperbola). Here A , B , C are the coefficients of x 2 , x y , y 2 — the three degree-2 terms. B is the coefficient of the cross term x y , never of x .
Big determinant Δ = A B /2 D /2 B /2 C E /2 D /2 E /2 F → tells whether the curve is a genuine conic (Δ = 0 ) or has collapsed into a point / lines (Δ = 0 ).
The off-diagonal entries are halved because B x y splits as 2 B x y + 2 B y x to make the matrix symmetric.
Definition The rotated coordinates
x ′ , y ′ and the angle θ
When the cross term B x y is present, the conic's own axes are tilted relative to the x , y axes by some angle θ (measured anticlockwise from the x -axis). Define x ′ , y ′ to be the coordinates of the same point but measured along those tilted axes — literally, "turn your head by θ and re-read the coordinates." The parent note's Rotation of Axes substitution is
x = x ′ cos θ − y ′ sin θ , y = x ′ sin θ + y ′ cos θ .
Choosing θ so that the cross term vanishes (tan 2 θ = A − C B ) rewrites the equation as A ′ x ′2 + C ′ y ′2 + ⋯ = 0 — a standard un-tilted conic. We use x ′ , y ′ explicitly only in Ex 1; you never need them just to classify.
Prerequisite tours if any of these feel shaky: Discriminant of a Quadratic , Degenerate Conics , Rotation of Axes , Quadratic Forms and Eigenvalues .
Every classification problem falls into exactly one of these cells. The final column names the example that hits it.
#
Cell (the scenario)
What decides it
Example
1
Δ 2 < 0 , A = C or B = 0 → tilted ellipse
B 2 − 4 A C < 0 , not a circle
Ex 1
2
Δ 2 < 0 , A = C , B = 0 → circle (limiting ellipse)
equal squares, no cross term
Ex 2
3
Δ 2 = 0 → parabola (perfect square quadratic part)
B 2 − 4 A C = 0
Ex 3
4
Δ 2 > 0 → hyperbola (indefinite, factors into 2 lines)
B 2 − 4 A C > 0 , Δ = 0
Ex 4
5
Degenerate, Δ 2 > 0 → two intersecting lines
Δ 2 > 0 but Δ = 0
Ex 5
6a
Degenerate, Δ 2 = 0 → parallel distinct lines
Δ 2 = 0 , Δ = 0 , distinct roots
Ex 6
6b
Degenerate, Δ 2 = 0 → coincident line
equal roots
Ex 6b
6c
Degenerate, Δ 2 = 0 → empty (no real points)
negative constant
Ex 6c
7
Degenerate, Δ 2 < 0 → single point (collapsed ellipse)
Δ 2 < 0 and Δ = 0
Ex 7
7b
Imaginary ellipse, Δ 2 < 0 → empty set
Δ = 0 but no real points
Ex 7b
8
Zero-input edge: A = C = 0 (only x y term) → rectangular hyperbola
B = 0 , A = C = 0
Ex 8
9
Word problem — searchlight cone → identify shape
build the equation first
Ex 9
10
Exam twist — a parameter k ; find k for a parabola
solve B 2 − 4 A C = 0
Ex 10
11
Not-a-conic edge — A = B = C = 0 → single line or no locus
linear / constant only
Ex 11
The same logic reads more naturally as a decision tree — follow the arrows from the top. The two diamonds are the two numbers (Δ 2 then Δ ); each leaf is one row of the table above.
5 x 2 − 4 x y + 5 y 2 − 12 = 0 .
Forecast: Same coefficient on x 2 and y 2 but there IS an x y term. Guess: ellipse or circle? Commit before reading.
Step 1. Read A = 5 , B = − 4 , C = 5 , D = 0 , E = 0 , F = − 12 .
Why this step? Classification always starts by matching the equation to the template letter-for-letter.
Step 2. Compute the discriminant:
B 2 − 4 A C = ( − 4 ) 2 − 4 ( 5 ) ( 5 ) = 16 − 100 = − 84 < 0.
Why this step? Negative discriminant ⇒ ellipse family .
Step 3. Circle or ellipse? A circle needs A = C and B = 0 . Here A = C = 5 but B = − 4 = 0 . So it is a tilted (rotated) ellipse , not a circle.
Why this step? The parent note warns: negative alone ≠ circle. The cross term is a tilt, so the two axes of the oval have different lengths in the x , y frame.
Step 4. Confirm it's a genuine conic:
Δ = 5 − 2 0 − 2 5 0 0 0 − 12 = − 12 ( 5 ⋅ 5 − ( − 2 ) ( − 2 ) ) = − 12 ( 25 − 4 ) = − 252 = 0.
Why this step? Δ = 0 guarantees the ellipse hasn't collapsed to a point.
Verify (deriving the standard form). Since A = C , the tilt formula tan 2 θ = B / ( A − C ) has a zero denominator, so 2 θ = 9 0 ∘ and θ = 4 5 ∘ . At 4 5 ∘ , cos θ = sin θ = 2 1 , so from the Rotation of Axes rule the new coefficients are
A ′ = A cos 2 θ + B cos θ sin θ + C sin 2 θ = 5 ⋅ 2 1 + ( − 4 ) ⋅ 2 1 + 5 ⋅ 2 1 = 2 5 − 4 + 5 = 3 ,
C ′ = A sin 2 θ − B cos θ sin θ + C cos 2 θ = 5 ⋅ 2 1 − ( − 4 ) ⋅ 2 1 + 5 ⋅ 2 1 = 2 5 + 4 + 5 = 7.
With B ′ = 0 the equation becomes 3 x ′2 + 7 y ′2 − 12 = 0 , i.e. 3 x ′2 + 7 y ′2 = 12 — a genuine ellipse with semi-axes 12/3 = 2 (along x ′ ) and 12/7 (along y ′ ). Notice A ′ C ′ = 21 > 0 (same sign, closed loop) and B ′2 − 4 A ′ C ′ = − 84 = B 2 − 4 A C , matching the invariance. Ellipse. ✓
Related: Ellipse Standard Equation , Circle as Special Ellipse .
2 x 2 + 2 y 2 − 8 x + 4 y − 1 = 0 .
Forecast: No x y term, equal squares. What special ellipse is this?
Step 1. A = 2 , B = 0 , C = 2 , D = − 8 , E = 4 , F = − 1 .
Why this step? Read every coefficient off the equation first — the whole classification rests on getting A , B , C right.
Step 2. B 2 − 4 A C = 0 − 4 ( 2 ) ( 2 ) = − 16 < 0 ⇒ ellipse family.
Why this step? Establish the family before checking the special case.
Step 3. Special-case test: A = C = 2 and B = 0 . Both hold ⇒ circle .
Why this step? Equal squared coefficients with no cross term means the level curve is equally "stretched" in every direction — the definition of a round circle.
Step 4. Sanity: divide by 2 and complete the square:
x 2 + y 2 − 4 x + 2 y − 2 1 = 0 ⇒ ( x − 2 ) 2 + ( y + 1 ) 2 = 2 1 + 4 + 1 = 2 11 .
Why this step? A positive right-hand side (11/2 > 0 ) confirms a real circle of radius 11/2 .
Verify: Radius2 = 11/2 = 5.5 > 0 , centre ( 2 , − 1 ) . Circle. ✓
9 x 2 − 12 x y + 4 y 2 + x − 2 = 0 .
Forecast: Is the degree-2 part a perfect square? If yes — which family?
Step 1. A = 9 , B = − 12 , C = 4 .
Why this step? Pull the three degree-2 coefficients out; B is the coefficient of x y , so B = − 12 , not the coefficient of x .
Step 2. B 2 − 4 A C = 144 − 4 ( 9 ) ( 4 ) = 144 − 144 = 0 ⇒ parabola .
Why this step? A zero discriminant is the borderline between ellipse and hyperbola — exactly the parabola case.
Step 3. The tell-tale sign: 9 x 2 − 12 x y + 4 y 2 = ( 3 x − 2 y ) 2 , a perfect square.
Why this step? When the quadratic form squashes to a single squared expression, one direction has zero curvature — that's the borderline shape between loop and split, i.e. a parabola. See Parabola Standard Equation .
Step 4. Confirm non-degenerate. With D = 1 , E = 0 , F = − 2 :
Δ = 9 − 6 1/2 − 6 4 0 1/2 0 − 2 .
Expand along the last row: 2 1 ( ( − 6 ) ( 0 ) − 4 ( 1/2 ) ) + ( − 2 ) ( 9 ⋅ 4 − ( − 6 ) ( − 6 ) ) = 2 1 ( − 2 ) − 2 ( 36 − 36 ) = − 1 = 0.
Why this step? Δ = 0 ⇒ genuine parabola, not a pair of parallel lines.
Verify: Δ = − 1 = 0 and Δ 2 = 0 together ⇒ parabola. ✓
x 2 + 3 x y − 4 y 2 + 2 x − y + 6 = 0 .
Forecast: Does the quadratic part factor into two different lines?
Step 1. A = 1 , B = 3 , C = − 4 .
Step 2. B 2 − 4 A C = 9 − 4 ( 1 ) ( − 4 ) = 9 + 16 = 25 > 0 ⇒ hyperbola .
Step 3. Indefinite check: x 2 + 3 x y − 4 y 2 = ( x + 4 y ) ( x − y ) — two distinct linear factors, so the form takes both signs. Its level curves are open two-branch curves. See Hyperbola Standard Equation .
Step 4. Non-degenerate? D = 2 , E = − 1 , F = 6 :
Δ = 1 3/2 1 3/2 − 4 − 1/2 1 − 1/2 6 = − 4 141 = 0.
Why this step? Positive Δ 2 with Δ = 0 ⇒ genuine hyperbola (not a crossing line-pair).
Verify: Δ = − 141/4 = − 35.25 = 0 . Hyperbola. ✓
x 2 − y 2 − 2 x + 1 = 0 .
Forecast: Δ 2 > 0 screams "hyperbola" — but is it real, or has it collapsed?
Step 1. A = 1 , B = 0 , C = − 1 , D = − 2 , E = 0 , F = 1 .
Step 2. B 2 − 4 A C = 0 − 4 ( 1 ) ( − 1 ) = 4 > 0 ⇒ hyperbola family .
Step 3. Check Δ :
Δ = 1 0 − 1 0 − 1 0 − 1 0 1 = − 1 ( 1 ⋅ 1 − ( − 1 ) ( − 1 ) ) = − 1 ( 1 − 1 ) = 0.
Why this step? Δ = 0 ⇒ degenerate . The "hyperbola" has collapsed.
Step 4. Factor to see what remains:
x 2 − 2 x + 1 − y 2 = ( x − 1 ) 2 − y 2 = ( x − 1 − y ) ( x − 1 + y ) = 0.
So y = x − 1 or y = − ( x − 1 ) — two intersecting lines crossing at ( 1 , 0 ) .
Why this step? This is a hyperbola shrunk to its asymptotes — the ultimate limiting case.
Verify: Point ( 1 , 0 ) : 1 − 0 − 2 + 1 = 0 ✓, and ( 2 , 1 ) : 4 − 1 − 4 + 1 = 0 ✓. Two lines. ✓
The figure below draws exactly this collapse and is worth reading line by line. The blue line is y = x − 1 ; feed it a point like ( 2 , 1 ) and the original equation gives 0 . The yellow line is the mirror y = − ( x − 1 ) ; the point ( 0 , 1 ) sits on it and also satisfies the equation. The two lines are what the hyperbola of Cell 4 becomes when it degenerates — its two branches have flattened onto their own asymptotes. The red dot marks ( 1 , 0 ) , the single crossing point: it is the "waist" the shrinking hyperbola pinched down to, and it lies on both lines simultaneously (set x = 1 in either equation to get y = 0 ).
x 2 − 6 x + 8 = 0 (viewed in the x y -plane).
Forecast: No y at all. What does a y -free equation look like as a curve?
Step 1. A = 1 , B = 0 , C = 0 , D = − 6 , E = 0 , F = 8 .
Step 2. B 2 − 4 A C = 0 − 0 = 0 ⇒ parabola family .
Step 3. Check Δ : with C = E = 0 ,
Δ = 1 0 − 3 0 0 0 − 3 0 8 = 0
(a whole zero row) ⇒ degenerate .
Step 4. Factor: x 2 − 6 x + 8 = ( x − 2 ) ( x − 4 ) = 0 , so x = 2 or x = 4 — two parallel vertical lines (any y allowed).
Why this step? A parabola degenerates into parallel lines; here both are real and distinct because the roots 2 , 4 differ.
Verify: Roots of x 2 − 6 x + 8 are 2 and 4 ; distinct ⇒ two parallel lines. ✓
x 2 − 6 x + 9 = 0 .
Forecast: Same shape as Ex 6 but the constant nudged. When two parallel lines merge, what's left?
Step 1. A = 1 , B = 0 , C = 0 , D = − 6 , E = 0 , F = 9 .
Step 2. B 2 − 4 A C = 0 ⇒ parabola family ; and Δ = 0 (zero y -row as in Ex 6) ⇒ degenerate.
Why this step? Same family/degeneracy test; only the constant changed, so we look at the roots to see the sub-case.
Step 3. Factor: x 2 − 6 x + 9 = ( x − 3 ) 2 = 0 , so x = 3 only — a single (coincident/repeated) line .
Why this step? Equal roots mean the two parallel lines of Ex 6 have slid together into one doubled line — the limiting case between "two lines" and "empty".
Verify: Discriminant of x 2 − 6 x + 9 is 36 − 36 = 0 ⇒ one repeated root x = 3 . Coincident line. ✓
x 2 − 6 x + 10 = 0 in the real x y -plane.
Forecast: Push the constant one notch higher than Ex 6b. Can a real curve just... vanish?
Step 1. A = 1 , B = 0 , C = 0 , D = − 6 , E = 0 , F = 10 .
Step 2. B 2 − 4 A C = 0 and Δ = 0 ⇒ degenerate parabola family, again.
Step 3. Complete the square: x 2 − 6 x + 10 = ( x − 3 ) 2 + 1 . Since ( x − 3 ) 2 ≥ 0 , the left side is ≥ 1 > 0 for every real x — the equation has no real solutions .
Why this step? When the leftover constant is positive, even the doubled line disappears: the locus is the empty set — no points at all.
Verify: Discriminant of x 2 − 6 x + 10 is 36 − 40 = − 4 < 0 ⇒ no real roots. Empty set. ✓
4 x 2 + 9 y 2 − 8 x + 4 = 0 .
Forecast: Δ 2 < 0 says ellipse — but can an ellipse shrink to nothing?
Step 1. A = 4 , B = 0 , C = 9 , D = − 8 , E = 0 , F = 4 .
Step 2. B 2 − 4 A C = 0 − 4 ( 4 ) ( 9 ) = − 144 < 0 ⇒ ellipse family .
Step 3. Complete the square:
4 ( x 2 − 2 x ) + 9 y 2 + 4 = 4 ( x − 1 ) 2 − 4 + 9 y 2 + 4 = 4 ( x − 1 ) 2 + 9 y 2 = 0.
Why this step? A sum of two non-negative squares equals 0 only when both are 0 .
Step 4. So x = 1 and y = 0 : the entire "ellipse" is the single point ( 1 , 0 ) .
Check Δ = 4 0 − 4 0 9 0 − 4 0 4 = 9 ( 16 − 16 ) = 0 ⇒ degenerate, as expected.
Verify: Only ( 1 , 0 ) satisfies it: 4 − 8 + 4 = 0 ✓; any other point gives a positive sum. Single point. ✓
4 x 2 + 9 y 2 − 8 x + 8 = 0 .
Forecast: Same ellipse family as Ex 7, but nudge the constant up. Point, real oval, or nothing?
Step 1. A = 4 , B = 0 , C = 9 , D = − 8 , E = 0 , F = 8 .
Step 2. B 2 − 4 A C = − 144 < 0 ⇒ ellipse family .
Step 3. Complete the square:
4 ( x − 1 ) 2 − 4 + 9 y 2 + 8 = 4 ( x − 1 ) 2 + 9 y 2 + 4 = 0 ⇒ 4 ( x − 1 ) 2 + 9 y 2 = − 4.
Why this step? The left side is a sum of squares, so it is ≥ 0 ; it can never equal − 4 . There are no real points — an imaginary (empty) ellipse .
Step 4. Note Δ = 4 0 − 4 0 9 0 − 4 0 8 = 9 ( 32 − 16 ) = 144 = 0 . So Δ 2 < 0 with Δ = 0 passes the "genuine conic" test, yet the real graph is empty.
Why this step? Lesson: Δ = 0 guarantees non-degenerate , but for the ellipse family a further sign check decides real-vs-imaginary. The rule: an ellipse-family curve is real only when A and Δ have opposite signs. Here A = 4 > 0 and Δ = 144 > 0 share the same sign, so the oval lives only in complex coordinates — hence the empty real graph. (In Ex 1 the real ellipse had A = 5 > 0 and Δ = − 252 < 0 : opposite signs, so it was genuinely visible.)
Verify: 4 ( x − 1 ) 2 + 9 y 2 ≥ 0 > − 4 for all real x , y ⇒ no solutions. Empty (imaginary ellipse). ✓
x y = 6 .
Forecast: Just a product equals a constant. This is the classic "reciprocal" curve — which family?
Step 1. Rewrite as x y − 6 = 0 . Then A = 0 , B = 1 , C = 0 , D = 0 , E = 0 , F = − 6 .
Why this step? The x y term has coefficient 1 , so B = 1 ; there are NO x 2 or y 2 terms, so A = C = 0 .
Step 2. B 2 − 4 A C = 1 − 4 ( 0 ) ( 0 ) = 1 > 0 ⇒ hyperbola .
Why this step? Even with A = C = 0 the discriminant formula still applies verbatim; a positive value forces the hyperbola family.
Step 3. Because A = C (= 0 here) the axes lie at 4 5 ∘ : rotating gives 12 x ′2 − 12 y ′2 = 1 , a rectangular hyperbola (equal semi-axes, perpendicular asymptotes = the x , y axes themselves).
Why this step? This is the limiting "pure cross-term" input the discriminant handles just as smoothly as any other.
Step 4. Δ = 0 1/2 0 1/2 0 0 0 0 − 6 = − 6 ( 0 − 4 1 ) = 2 3 = 0 ⇒ genuine (non-degenerate) hyperbola.
Verify: Δ 2 = 1 > 0 , Δ = 1.5 = 0 . Curve passes through ( 2 , 3 ) , ( 3 , 2 ) , ( − 2 , − 3 ) — two branches. Rectangular hyperbola. ✓
The figure shows the two branches of x y = 6 . Notice the green dots at ( 2 , 3 ) , ( 3 , 2 ) , ( − 2 , − 3 ) that we checked — one branch sits in the top-right, its mirror in the bottom-left. The dashed red lines are the x - and y -axes: they are the asymptotes the branches hug but never touch, which is exactly why A = C = 0 gives a rectangular hyperbola.
Worked example A searchlight sits on the ground and points upward at a slant. The rim of its light-cone traces the equation
16 x 2 − 24 x y + 9 y 2 − 40 x − 30 y = 0 on a wall. What conic is the bright boundary?
Forecast: A slanted cone cut by a plane — the parent's flashlight picture. Loop, park, or split?
Step 1. A = 16 , B = − 24 , C = 9 , D = − 40 , E = − 30 , F = 0 .
Why this step? Translate the physical curve into coefficients before touching geometry.
Step 2. B 2 − 4 A C = 576 − 4 ( 16 ) ( 9 ) = 576 − 576 = 0 ⇒ parabola family .
Step 3. Check Δ to be sure it's a real parabola and not a line-pair:
Δ = 16 − 12 − 20 − 12 9 − 15 − 20 − 15 0 .
Expand along the last row:
− 20 ( ( − 12 ) ( − 15 ) − 9 ( − 20 ) ) − ( − 15 ) ( 16 ( − 15 ) − ( − 12 ) ( − 20 ) ) + 0
= − 20 ( 180 + 180 ) + 15 ( − 240 − 240 ) = − 20 ( 360 ) + 15 ( − 480 ) = − 7200 − 7200 = − 14400 = 0.
Why this step? Δ = 0 confirms a genuine parabola, matching the physical intuition of a beam edge that skims the wall.
Step 4. Quadratic part = ( 4 x − 3 y ) 2 , the perfect-square hallmark.
Why this step? Factoring the degree-2 group into a single square is the algebraic fingerprint of a parabola — it double-confirms the Δ 2 = 0 verdict and shows the beam-edge direction 4 x = 3 y .
Verify: Δ 2 = 0 , Δ = − 14400 = 0 ⇒ the boundary is a parabola. ✓
Worked example For which value(s) of
k does k x 2 + 12 x y + 9 y 2 + 3 x − y + 2 = 0 represent a parabola ?
Forecast: A parabola needs the discriminant to hit exactly zero. One equation, one unknown k .
Step 1. A = k , B = 12 , C = 9 .
Step 2. Parabola condition: B 2 − 4 A C = 0 .
144 − 4 ( k ) ( 9 ) = 0 ⇒ 144 = 36 k ⇒ k = 4.
Why this step? Setting the discriminant to its borderline value 0 pins down the parabola case exactly.
Step 3. Guard against collapse. At k = 4 : quadratic part = 4 x 2 + 12 x y + 9 y 2 = ( 2 x + 3 y ) 2 . Check Δ with D = 3 , E = − 1 , F = 2 :
Δ = 4 6 3/2 6 9 − 1/2 3/2 − 1/2 2 = − 4 121 = 0.
Why this step? Δ = 0 ensures k = 4 gives a real parabola, not two parallel lines.
Verify: k = 4 ⇒ B 2 − 4 A C = 144 − 144 = 0 ✓ and Δ = − 121/4 = − 30.25 = 0 ✓. Answer: k = 4 . ✓
Worked example Classify (a)
2 x + 3 y − 6 = 0 and (b) 0 ⋅ x + 0 ⋅ y + 5 = 0 .
Forecast: No degree-2 terms at all. Is the discriminant even meaningful here?
Step 1. For both, A = B = C = 0 . The general-equation requirement ( A , B , C ) = ( 0 , 0 , 0 ) is violated , so these are not second-degree equations — the conic classification simply does not apply.
Why this step? The discriminant test is defined only when a quadratic part exists; recognising when you have left that regime is itself a skill.
Step 2. (a) 2 x + 3 y − 6 = 0 is linear ⇒ a single straight line . (b) 5 = 0 is false for every ( x , y ) ⇒ the empty set (no locus). A constant like 0 = 0 would instead be the whole plane .
Why this step? These are the true edge cases below "conic": one non-zero linear equation gives a line; a purely non-zero constant gives nothing.
Verify: (a) ( 3 , 0 ) and ( 0 , 2 ) both satisfy 2 x + 3 y − 6 = 0 — two points determine the line. (b) 5 = 0 always ⇒ no points. Line / empty. ✓
Recall The whole decision, in order
Compute B 2 − 4 A C for the family, then Δ for reality. ::: B 2 − 4 A C < 0 ellipse (circle if A = C , B = 0 ), = 0 parabola, > 0 hyperbola — but if Δ = 0 it has degenerated to a point / line(s), and even with Δ = 0 an ellipse-family curve can be empty (imaginary) when A and Δ share a sign.
Mnemonic Two-number habit
Family first, Reality second. B 2 − 4 A C names the shape; Δ tells you whether that shape actually showed up or collapsed.