This course leans constantly on three tools from College Algebra: a matrix as a grid of numbers, the three row operations that solve a system by hand, and the determinant that tells a 2\times2 or 3\times3 system whether it even has a unique solution before you finish solving it. This section walks through all three again, tightened up to the exact notation and habits the rest of the course will assume you have automatically.
Think of a system of equations as a stack of index cards, one card per equation. The three row operations are the only three things you're ever allowed to do to that stack. You can shuffle two cards. You can rewrite one card by scaling every number on it by the same amount. Or you can staple a multiple of one card onto another and keep the combined version. Shuffling obviously can't change what the stack says. Scaling a card is just multiplying both sides of one equation by the same number, so it's still true. Stapling a multiple of card B onto card A builds a new true statement entirely out of two statements that were already true, so it can't smuggle in a lie. RREF is just the tidy end state. You keep stapling and shuffling until each card mentions exactly one variable, and the stack has become the answer key.
An m \times n matrix A is a rectangular array of numbers with entry a_{ij} sitting in row i, column j. Row-reducing a system means applying a sequence of elementary row operations to its augmented matrix:
- Swap two rows: R_i \leftrightarrow R_j.
- Scale a row by a nonzero constant: R_i \to cR_i.
- Replace a row with itself plus a multiple of another row: R_i \to R_i + cR_j.
None of the three changes the system's solution set. A matrix is in reduced row echelon form (RREF) once every leading entry is a 1, each sits strictly right of the leading entry in the row above it, and each is the only nonzero entry in its column — at that point the solution is read straight off the matrix, no further algebra needed.
Determinants play a different role: instead of finding the solution, one number tells you in advance whether a unique solution exists at all.
A 2\times2 determinant is one subtraction. Multiply the two entries on one diagonal of the box together, do the same for the other diagonal, then subtract the second product from the first. Geometrically, the two rows are arrows, and ad-bc is the signed area of the parallelogram they sweep out. Arrows pointing in genuinely different directions sweep out a fat parallelogram: nonzero area, one clean intersection point. Arrows that collapse onto the same line sweep out no area at all. A zero-area result is the determinant's way of saying "these rows don't actually pin down a single point together." A 3\times3 determinant upgrades the same story to a volume, since three arrows now span a box instead of two spanning a flat shape.
For a 2\times2 matrix, \det\begin{pmatrix}a&b\\c&d\end{pmatrix} = ad-bc. For a 3\times3 matrix, cofactor expansion along any row or column gives the same value, and a matrix that is upper or lower triangular (all entries on one side of the main diagonal are 0) has determinant equal to the plain product of its diagonal entries — no expansion required. A determinant of 0 signals that the matrix's rows are dependent: the system it represents has either no solution or infinitely many, never exactly one. Swapping two rows negates the determinant; scaling a row by k scales the determinant by k.
Step through the row operations yourself — predict the next legal move before the widget takes it, then watch the numbers shift:
Row-reduce a 3×3 augmented matrix
Before each click, predict which row operation clears a zero into place next, then press Next row operation to check. Watch the involved rows shade and the changed numbers turn indigo as the matrix moves toward reduced row-echelon form.
Loading interactive…
Check your understanding (counts toward your progress)
A system's coefficient matrix has determinant 0. Row reduction shows the system is consistent (no contradiction row appears). How many solutions does it have — enter 0 for none, 1 for exactly one, or 2 to mean infinitely many?
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Why
Determinant 0 rules out a unique solution. Combined with consistency (no row like 0=c with c\neq0), the only remaining option is infinitely many solutions, from at least one free variable.
Correct answer
Also accepted:
Why
Recall what it means for a system to be consistent with a free variable versus consistent with none -- how many solutions does each case give?
Compute \det\begin{pmatrix} 5 & 2 \\ 3 & 4 \end{pmatrix} — the matrix with five in the top-left corner.
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Why
Main diagonal product minus the other diagonal product: 5(4) - 2(3) = 20 - 6 = 14.
Correct answer
Also accepted:
Why
Compute \det\begin{pmatrix} 2 & -3 & 5 \\ 0 & 4 & 0 \\ 0 & 0 & -1 \end{pmatrix}.
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Why
Every entry below the main diagonal is already 0, so this matrix is upper triangular — the determinant is just the product of the diagonal entries: 2 \times 4 \times (-1) = -8.
Correct answer
Also accepted:
Why
This matrix is upper triangular -- recall what a triangular matrix's determinant reduces to.
Practice: The three row operations, read and applied (4 problems · doesn’t affect your progress)
Which of the three elementary row operations is R_3 \to R_3 - 4R_1? (For reference, the choices given are: Swap; Scale; Replace; None of these — it's not legal.)
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
A swap exchanges two whole rows with each other; this operation keeps R_3 in place and only changes its contents.
Scaling multiplies one row by a constant on its own; this operation mixes in a multiple of a different row (R_1), which makes it the replace move.
This is one of the three legal moves — replacing R_3 with itself plus a multiple of R_1 never changes the solution set.
Why
R_i \to R_i + cR_j (here c=-4, j=1) is the replace operation — it's the row-based version of the elimination move you already know from solving systems by substitution and elimination.
Replace
Which row operation on an augmented matrix could accidentally change the system's solution set if done incorrectly? (For reference, the choices given are: Swapping two rows; Scaling a row by 0; Scaling a row by -3; Replacing R_2 with R_2 + 2R_1.)
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Reordering equations never changes what they mean together — a swap is always safe.
Scaling by any nonzero constant is always legal and safe; -3 is nonzero.
Adding a multiple of one row to another is always legal and safe — it builds a new true statement out of two already-true ones.
Why
Scaling a row by 0 destroys that row's information entirely (it becomes all zeros), which is why the legal scale operation requires a nonzero constant. Every genuine elementary row operation preserves the solution set — scaling by 0 simply isn't one of the three legal moves.
Scaling a row by 0
A matrix's row 2 is entirely (0\;\;0\;\;0 \mid 7) after row reduction. What does that row say about the system? (For reference, the choices given are: It has a unique solution; It has no solution; It has infinitely many solutions; Row 2's variable is free.)
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
A unique solution shows up as a leading 1 in every variable column, not a row of all zeros before the bar.
Infinitely many solutions comes from a row that reads 0=0 (an all-zero row, augmented entry included), not 0=7.
A free variable shows up when a variable column has no leading entry — this row is a flat contradiction, not a free-variable signal.
Why
The row reads 0x+0y+0z=7, i.e. 0=7 — a statement that's never true no matter what the variables are. That's the signature of an inconsistent system: no solution.
It has no solution
Row reduction turns the coefficient matrix of a 3\times3 system into RREF, and column 3 (the z-column) never gets a leading entry, while every row before that is consistent. What does that tell you? (For reference, the choices given are: The system has no solution; z is a free variable — infinitely many solutions; x and y are both free; The system's determinant must be positive.)
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
No solution shows up as a row like (0\;0\;0\mid c) with c\neq0 — a missing leading entry in a variable column is a different signal.
Only the variable whose column has no leading entry is free; x and y still get pinned down once z is named.
Determinant sign isn't determined by which column lacks a pivot — a free-variable system like this one doesn't even have a single well-defined determinant, since its coefficient matrix isn't square with a unique-solution structure to begin with.
Why
A variable column with no leading (pivot) entry marks that variable as free — name it with a parameter, solve every other variable in terms of it, and the result is a one-parameter family: infinitely many solutions.
A missing leading entry in a variable's column signals a free variable -- think about what that means for the number of solutions when the system is consistent.
z is a free variable — infinitely many solutions
Quick review
Practice from earlier coursework, pulled in for this topic.
From College Algebra
Use row reduction (or any method) to solve \;3x + y = 9,\;\; x - 2y = -4\; for y specifically.
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Why
Augment and reduce: swap rows to put (1\; {-2} \mid {-4}) on top, then R_2 \to R_2 - 3R_1 gives (0\; 7 \mid 21), so y = 3. Back-substitute: x - 6 = -4, so x = 2. Check: 3(2) + 3 = 9 ✓.
Correct answer
Also accepted:
Why
Set up the augmented matrix and use row operations to eliminate x from one equation.
Swap the rows so a row starting with 1 in the x-column is on top, then eliminate x from the other row.
From College Algebra
Starting from \left(\begin{array}{ccc|c} 1 & 1 & 1 & 6 \\ 0 & 1 & 2 & 7 \\ 0 & -2 & 3 & 4 \end{array}\right), apply R_3 \to R_3 + 2R_2, an addition rather than a subtraction. What is the resulting entry in row 3, column 4?
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Why
New row 3 = old row 3 plus 2 times row 2, entry by entry: (0+2(0),\; -2+2(1),\; 3+2(2),\; 4+2(7)) = (0,0,7,18). The column-4 entry is 18 — and notice column 2 also cleared to 0, extending the staircase.
Correct answer
Also accepted:
Why
Compute each new row 3 entry as (old row 3 entry) plus 2 times (row 2's matching entry).
Worked example
Row-reduce a 3×3 system to its solution
-
1
Augmented matrix for x+y+z=5, 2x-y+2z=13, x+3y-z=-5. Clear column 1 below the pivot with R_2 \to R_2 - 2R_1 and R_3 \to R_3 - R_1: \left(\begin{array}{ccc|c} 1 & 1 & 1 & 5 \\ 2 & -1 & 2 & 13 \\ 1 & 3 & -1 & -5 \end{array}\right) \;\longrightarrow\; \left(\begin{array}{ccc|c} 1 & 1 & 1 & 5 \\ 0 & -3 & 0 & 3 \\ 0 & 2 & -2 & -10 \end{array}\right)
-
2
Scale row 2 by -\tfrac13: R_2 \to \big(0,\ \overset{1}{\cancel{-3}},\ 0 \mid \overset{-1}{\cancel{3}}\big), so \textcolor{#4f46e5}{y = -1} already, straight off row 2 — no back-substitution needed for that one variable.
-
3
Clear column 2 below with R_3 \to R_3 - 2R_2: \left(\begin{array}{ccc|c} 1 & 1 & 1 & 5 \\ 0 & 1 & 0 & -1 \\ 0 & 0 & -2 & -8 \end{array}\right) Row 3 reads -2z=-8, so \textcolor{#b91c1c}{z = 4}.
-
4
Back-substitute both into row 1: x + \textcolor{#4f46e5}{(-1)} + \textcolor{#b91c1c}{4} = 5, so x = 2. \boxed{(x,y,z) = (2,\ -1,\ 4)} Check in the second original equation: 2(2) - (-1) + 2(4) = 4+1+8 = 13 ✓.
Loading interactive…
Worked example
Compute a 3×3 determinant by hunting for zeros
-
1
Compute \det\begin{pmatrix} 3 & 0 & -2 \\ 1 & 4 & 5 \\ 2 & 0 & 6 \end{pmatrix}. Column 2 has entries 0, 4, 0 — expanding along it means two of the three cofactor terms are automatically 0, so only the middle entry does any work.
-
2
The surviving entry is 4, in row 2, column 2, so its sign is (-1)^{2+2} = +. Delete row 2 and column 2, leaving the minor \begin{pmatrix} 3 & -2 \\ 2 & 6 \end{pmatrix} with determinant 3(6) - (-2)(2) = 18 + 4 = 22.
-
3\det = \colorbox{#fef3c7}{$0$} + 4(22) + \colorbox{#fef3c7}{$0$} = 88
Check by expanding along row 1 instead: 3\det\begin{pmatrix}4&5\\0&6\end{pmatrix} - 0 + (-2)\det\begin{pmatrix}1&4\\2&0\end{pmatrix} = 3(24) + (-2)(-8) = 72+16=88 ✓ — same answer, a different set of zeros to lean on.
Loading interactive…
Check your understanding (counts toward your progress)
\det\begin{pmatrix} a & b \\ c & d \end{pmatrix} = 9. What is \det\begin{pmatrix} 3a & 3b \\ c & d \end{pmatrix}?
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Why
Scaling one whole row by k scales the determinant by that same k. Row 1 was scaled by 3, so the new determinant is 3 \times 9 = 27.
Correct answer
Also accepted:
Why
Recall the row-operation rule for how scaling a single row by a constant affects the determinant.
Which of the three elementary row operations is R_3 \to R_3 - 4R_1?
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
A swap exchanges two whole rows with each other; this operation keeps R_3 in place and only changes its contents.
Scaling multiplies one row by a constant on its own; this operation mixes in a multiple of a different row (R_1), which makes it the replace move.
This is one of the three legal moves — replacing R_3 with itself plus a multiple of R_1 never changes the solution set.
Why
R_i \to R_i + cR_j (here c=-4, j=1) is the replace operation — it's the row-based version of the elimination move you already know from solving systems by substitution and elimination.
Which row operation on an augmented matrix could accidentally change the system's solution set if done incorrectly?
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Reordering equations never changes what they mean together — a swap is always safe.
Scaling by any nonzero constant is always legal and safe; -3 is nonzero.
Adding a multiple of one row to another is always legal and safe — it builds a new true statement out of two already-true ones.
Why
Scaling a row by 0 destroys that row's information entirely (it becomes all zeros), which is why the legal scale operation requires a nonzero constant. Every genuine elementary row operation preserves the solution set — scaling by 0 simply isn't one of the three legal moves.
Practice: Determinants, computed and interpreted (5 problems · doesn’t affect your progress)
\det A = -6. After swapping two rows of A, what is the new determinant? (For reference, the choices given are: -6; 6; 0; Not enough information.)
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
A swap always flips the sign — the value can't stay identical unless the determinant is already 0.
A swap only negates the determinant; it can't create or destroy area, so it can't force the result to 0.
This is fully determined: swapping two rows always negates whatever the determinant was before the swap.
Why
Swapping two rows negates the determinant, and negating -6 gives 6.
Recall the row-operation rule for how swapping two rows affects the sign of a determinant.
6
Evaluate \det\begin{pmatrix} 5 & 2 \\ 3 & 4 \end{pmatrix} using the diagonal-product shortcut.
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Why
Main diagonal product minus the other diagonal product: 5(4) - 2(3) = 20 - 6 = 14.
Correct answer
Also accepted:
Why
Compute the determinant of the triangular matrix \begin{pmatrix} 2 & -3 & 5 \\ 0 & 4 & 0 \\ 0 & 0 & -1 \end{pmatrix}.
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Why
Every entry below the main diagonal is already 0, so this matrix is upper triangular — the determinant is just the product of the diagonal entries: 2 \times 4 \times (-1) = -8.
Correct answer
Also accepted:
Why
This matrix is upper triangular -- recall what a triangular matrix's determinant reduces to.
Compute \det\begin{pmatrix} 1 & 0 & 3 \\ 2 & 0 & -1 \\ 4 & 5 & 2 \end{pmatrix} by expanding along the column with the most zeros. (Work through it specifically for \det\begin{pmatrix} 1 & 0 & 3 \\ 2 & 0 & -1 \\ 4 & 5 & 2 \end{pmatrix}.)
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Why
Column 2 has entries 0, 0, 5 — only the bottom entry contributes. Its sign at position (3,2) is (-1)^{3+2}=-, and its minor (delete row 3, column 2) is \det\begin{pmatrix}1&3\\2&-1\end{pmatrix} = -1-6=-7. Total: 5 \times (-1) \times (-7) = 35. Check by expanding along row 1 instead: 1\det\begin{pmatrix}0&-1\\5&2\end{pmatrix} - 0 + 3\det\begin{pmatrix}2&0\\4&5\end{pmatrix} = 1(5) + 3(10) = 5+30=35 — same answer.
Correct answer
Also accepted:
Why
Expanding along a column with two zeros leaves only one cofactor term to compute.
A system's coefficient matrix has determinant 0. Row reduction shows the system is consistent (no contradiction row appears). How many solutions does it have — enter 0 for none, 1 for exactly one, or 2 to mean infinitely many? (Work through it specifically for 0 and 0 and 1.)
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Why
Determinant 0 rules out a unique solution. Combined with consistency (no row like 0=c with c\neq0), the only remaining option is infinitely many solutions, from at least one free variable.
Correct answer
Also accepted:
Why
Recall what it means for a system to be consistent with a free variable versus consistent with none -- how many solutions does each case give?
Quick review
Practice from earlier coursework, pulled in for this topic.
From College Algebra
This matrix is lower triangular — find \det\begin{pmatrix} 2 & 0 & 0 \\ 7 & -3 & 0 \\ 1 & 8 & 6 \end{pmatrix} by inspection.
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Why
This matrix is lower triangular (every entry above the main diagonal is 0), so the determinant is the product of the diagonal entries: 2 \times (-3) \times 6 = -36.
Correct answer
Also accepted:
Why
Check which entries are zero — above or below the diagonal — to confirm the triangular shortcut applies, then multiply the diagonal.
From College Algebra
A student tries to apply Sarrus' rule (copying columns and multiplying diagonals) to a 4\times4 matrix. What happens?
Right on try — credit. Right — the answer had already been shown, so no credit this time. Next try is worth Answer shown after three tries — this one no longer earns credit. You chose to see the answer — this one earns no credit.
Sarrus' rule doesn't generalize by simply adding more diagonals — the diagonal picture itself stops corresponding to the determinant past 3\times3.
The extra work isn't the problem — the diagonal method computes the wrong quantity entirely for 4\times4, regardless of effort.
The failure has nothing to do with zero entries — it's structural: a 4\times4 determinant has 24 signed terms, but the diagonal trick only ever produces 8 products.
Why
A 4\times4 determinant expansion has 4!=24 terms, but the diagonal trick only ever produces 2n=8 products — the two match only when n=3 (both equal 6). Past 3\times3, cofactor expansion or row reduction is required.
Compare how many terms a full n\times n determinant expansion has (n!) to how many diagonal products Sarrus' rule produces (2n).
Where this goes next
What this course builds on it
Every chapter from here forward puts these two tools straight to work: solving a matrix equation for an unknown vector is exactly the row-reduction you just reviewed, dressed in new notation; testing whether a matrix has an inverse comes down to checking that its determinant is nonzero; and finding a matrix's eigenvalues later in the course means solving a determinant equation built directly on the cofactor and triangular-matrix shortcuts practiced here. If row reduction or a 3\times3 determinant felt shaky in this section, that is worth fixing now: every later chapter assumes both are already automatic and builds new material directly on top of them.
What's next
One habit hasn't been reviewed yet: writing an argument that a claim like "the rows are dependent" or "the system has infinitely many solutions" is actually true, not just plausible from an example. Linear algebra asks for that kind of short proof constantly, and the next section builds it before the rest of the course needs it.