THE TWO CONSTRAINTS — AND HOW THEY INTERACT
Every cell in Calcudoku faces two constraints simultaneously: the row/column must contain every digit exactly once (Latin square), and the cage the cell belongs to must produce its target number using the given operator.
Neither constraint alone is enough to place most digits. Together they narrow the candidate set fast. The technique is: enumerate what the cage allows, then filter by what the row and column still need. The overlap is usually one or two digits — sometimes just one.
Think of each cage as publishing a short list of possible digit sets. Your job is to find which set from the list is compatible with the row/column state.
START WITH THE EASIEST CAGES
Three cage types give you information immediately:
- 1-cell cages: the target is the digit. Place it immediately. Every 1-cell cage you resolve eliminates that digit from its entire row and column.
- Subtraction and division cages: these are always exactly 2 cells. The candidate pairs are finite and often very few — on a 4×4 grid, a 3− cage can only be {1,4} or {2,5}... but wait, 5 is out of range. Only {1,4}. That's an immediate pair placement.
- Multiplication cages on small grids: on a 4×4 grid (digits 1–4), a 6× cage in 2 cells must be {2,3}. A prime target like 5× can only be {1,5}... but 5 is out of range on a 4×4, so the puzzle can't have a 5× 2-cell cage on a 4×4. When you see prime targets, count the valid pairs immediately.
Work these before touching the addition cages. They give you the most information for the least calculation.
ENUMERATE ADDITION CAGE CANDIDATES
Addition cages are flexible — a 3-cell cage summing to 9 on a 5×5 grid could be {1,3,5}, {1,4,4}, {2,3,4}, or {2,2,5}. The grid-size digit ceiling cuts many of these down. Write out the candidate sets, even just mentally.
Two things prune the list fast: the grid size (digits 1–N only, where N = grid width) and the no-repeat row/column rule. A cage can contain repeated digits — unlike Killer Sudoku — but if both cells of a 2-cell cage are in the same row, the two digits must differ because each row uses every digit exactly once.
For a cage spanning two different rows and two different columns, a repeat IS possible: both cells could be the same digit as long as they don't share a row or column. This trips up Killer Sudoku players — the no-repeat-within-cage rule doesn't exist here.
CROSS-REFERENCE: THE MAIN TECHNIQUE
Once you have cage candidates, cross-reference with row and column state. If column 2 already contains digits 1, 3, and 5, any cage spanning column 2 can only use 2 or 4 in that cell. Intersect this with the cage candidate sets — the result is often one valid assignment.
The pattern to look for: a cage where one candidate set becomes impossible because it requires a digit that the row or column has already used. Even if that leaves two possible sets instead of one, you've narrowed the space.
Every time you place a digit, update your mental model of what each row and column still needs. Those updates are what unlock the next cross-reference.
WORK IN ROUNDS, NOT SWEEPS
Most Calcudoku grids don't solve in a single left-to-right pass. The first round places the obvious digits (singletons, forced pairs). Each placement changes the row/column state, which unlocks cage cross-references that were previously ambiguous. The second round resolves those. Repeat until the grid is full.
If you complete a full round and nothing was placed, you've either missed a cage candidate (check every subtraction and division cage again) or skipped a row/column interaction. There's no guessing in a well-formed Calcudoku — the logic is always there.
THE BEGINNER MISTAKES
Importing Killer Sudoku habits. In Killer Sudoku, digits cannot repeat within a cage. In Calcudoku they can — as long as the row/column uniqueness rule allows it. A {2,2} pair in a 2-cell 4+ cage is legal if those cells are in different rows and different columns. Always check whether the cage cells share a row or column before ruling out repeats.
Forgetting the grid-size ceiling. On a 5×5 grid, digits run 1–5. A 12× cage in 2 cells on a 5×5 grid: divisor pairs of 12 are {1,12}, {2,6}, {3,4}. Only {3,4} is in range. Always set the ceiling before evaluating candidates.
Treating subtraction/division cages as flexible. They have the most constrained candidate sets on the grid. Evaluate them first, every time you're stuck. A 1÷ cage must be a pair where one digit is a multiple of the other — on most grid sizes that's a very short list.
Not propagating after each placement. Every digit you place eliminates it from the rest of that row and column. The players who solve fastest do this automatically: place a digit, immediately update all cage candidates in the same row and column. Don't batch the propagation to the end of a round.