First PrinciplesStart anywhere. Prove it, then move on.

Rung 30

Systems, determinants, rank

When a system has one solution, none, or infinitely many — and why.

Best attempted after 29. Matrices as transformations. Nothing stops you trying this now — the gate will tell you if you were right.

The gate

Solve a 3×3 system by elimination, by hand, showing every row operation. Then deliberately construct a system with no unique solution, compute its determinant and its rank, and explain in terms of rung 29's picture what the transformation did to the space to make the solution non-unique.

Nobody checks this but you. Do it honestly and the rungs above hold; do it loosely and they will not, somewhere further up where the cause is much harder to find.

"Solve for x and y" was rung 5 with two unknowns. Add more and the arithmetic becomes unmanageable, so it gets systematic: write the coefficients as a matrix, eliminate downward, back-substitute. That is Gaussian elimination, and it is what every solver does underneath.

The more interesting question is when the procedure fails, because that failure is informative.

Why this is on the ladder

Because "does this system have a unique solution" is the same question as "is this transformation reversible", and reversibility governs a great deal further up. When a model has more parameters than data — which is normal — the system determining them is underdetermined, with infinitely many solutions. Knowing what that means structurally is why regularisation makes sense later instead of seeming arbitrary.

Do this

Write a 3×3 system with a unique solution and solve it by elimination. Every row operation named and written down, rung 5 style — each one is legal because doing the same thing to both sides of an equation preserves truth.

Now break it on purpose. Make the third equation the sum of the first two. It adds no information. Solve and watch a row collapse to 0 = 0.

Compute the determinant: zero. Compute the rank — the number of genuinely independent rows: two, not three.

Now connect it to rung 29. Determinant zero means the transformation squashes space flat. Once flattened, many different inputs land on the same output, so you cannot work backwards to a single answer. The algebra and the picture are the same fact.

Finally make it inconsistent: same left-hand sides, a right-hand side that contradicts. Now there is no solution at all. Say what distinguishes that from the infinite case.

Where people get stuck

Treating a zero determinant as an error. It is a finding: the transformation destroyed a dimension. That is the whole content.

Rank is also commonly reduced to a number to compute. Read it as "how many dimensions survive the transformation" and it becomes something you can reason with — including at rung 37, where deliberately keeping only the surviving dimensions is the entire technique.

Reading