In one sentence
One equation in two unknowns has infinitely many real solutions, joined up into a curve. But if you only want integer solutions, everything changes: they are the lattice points the curve happens to pass through. They are few and far between, often only finitely many, sometimes none at all.
The whole art of finding integer solutions is to use the properties only integers have — divisibility, factorization, bounds, parity — to sift infinitely many candidates down to a few you can count.
Example: positive integer solutions of 3x + 5y = 47
Step 1 / 5- Find one first
Try a few values of : at , , so . So is an integer solution, and the line passes through this lattice point.
Stop 1 · Divisibility: does the line ever hit a lattice point?
need not have integer solutions. Whatever integers and are, the left side is a multiple of the greatest common divisor of and . If the right side is not, there is no hope.
Break it
Counterexample lab: remove one condition
Conclusion: has integer solutions ← still true?
: the gcd of and is , and is a multiple of . The line has lattice points on it, one at regular intervals.
Change only the to : . The line has hardly moved, but… look carefully: which lattice point does it pass through?
Not a single one. The left side is always a multiple of , and is not. The line slips through the gaps between the lattice points.
Stop 2 · Factorization: turn a sum into a product
The equation looks hard to get a grip on. But add to both sides and the left side factors:
There are only finitely many ways for two integers to multiply to . Each way of splitting it gives one lattice point on the curve.
Predict first
How many integer solutions does have? (Count both positive and negative ones.)
has four positive divisors, , giving four ways to split it; making both factors negative gives four more. in all.
By the way, divide both sides of by and you get . So this fractional equation has the same positive integer solutions: , , , .
Stop 3 · Bounds: fence in the range, then check one by one
. A square is never negative, so and . The candidates shrink at once from “every integer” to about a dozen, and you can just check them one at a time.
The key to this move is first proving that there are only finitely many candidates. Once the range is fenced in, even the slow way is sure to finish.
Stop 4 · Parity: sometimes there are no solutions at all
One problem, several solutions
Find the positive integer solutions of .
Factor into a product
. Since and are positive integers, , so there are only two ways to split it:
- , : ,
- , : ,
Two solutions: and .
CostFastest, but the equation itself must factor.
Bound the range, then list
, so and . So , and .
Substitute one by one and see whether is a perfect square: only () and ().
CostDoes not need factoring, but takes more tries.
What happens to the same equation if we change to ?
x² − y² = 30 has no integer solutions
Step 1 / 4- Factor first, as before
.
Practice
How many positive integer solutions does have?
Among all integer solutions of , what is the largest value of ?
How many integer solutions does have?
How many positive integer solutions does have?