bits.

FREE CONTRIBUTION CHRONOLOGY · 628–c. 825

Numbers become state

What must a representation make possible before arithmetic becomes a reusable algorithm?

The historical question

Zero and signed quantities enlarged the states arithmetic could represent. Systematic algebra then turned whole classes of verbal problems into repeatable transformations rather than isolated tricks.

A better representation creates new valid states; an algorithm creates a reliable path between those states.

Representation and an invariant

A positional numeral stores both symbols and their positions. In base ten, 205 means 2×100 + 0×10 + 5. The zero preserves the empty tens position, so deleting it changes the represented number. An arithmetic zero, an absent record, and a Boolean false are different concepts even if a machine sometimes encodes them using the same bit pattern.

To solve x² + 10x = 39 by completing the square, add 25 to both sides: (x + 5)² = 64. Over real numbers, x + 5 is either 8 or −8, giving x = 3 or x = −13. Direct substitution checks both. The manipulation preserves equality, but taking a square root requires considering both signs. Historical treatments often organized positive quantities into separate problem classes; a modern signed-symbolic presentation should not be silently attributed to them.

This illustrates three layers. The number system specifies available values and operations. The equation expresses a constraint. The algorithm transforms the constraint while preserving its solutions. Division by a variable is dangerous because it can remove a zero-valued case. Dividing by zero is undefined in ordinary field arithmetic; extending notation does not remove the obligation to specify valid operations.

What the evidence does not establish

Extending a formal system also creates dangerous edges: some early zero rules were powerful while division-by-zero claims were not valid.

Reconstruction exercise

Solve x² + 6x = 16 by completing the square. Verify both real solutions and explain the difference between a zero value and missing data.

Worked solution

(x + 3)² = 25 gives x = 2 or x = −8. Substituting gives 4 + 12 = 16 and 64 − 48 = 16. A stored zero is a known numerical value. A missing measurement is an absence of information; replacing it with zero changes the problem and may bias a calculation. Tests should include the negative solution rather than retaining only the historically positive case.

Test the limits and transfer the idea

1. Algebraic proof audit · allow 25 minutes

Division can discard a valid solution

Solve x²=3x over the real numbers. A student divides both sides by x and reports only x=3. Identify the missing precondition, recover all solutions and verify them. Then compare solving x²=−1 over real numbers and complex numbers. Explain why changing the allowed number system changes the question rather than repairing an arithmetic mistake.

Get a starting hint

Before cancelling a factor, separate the case in which it is zero.

Compare your reasoning with the worked solution

Rewrite as x(x−3)=0, so x=0 or x=3. Substitution verifies 0²=3×0 and 3²=3×3. Division by x is equivalent only on the restricted domain x≠0; it discards the valid zero case if performed without a split. Over the reals, squares are nonnegative, so x²=−1 has no solution. Over the complex numbers it has solutions i and −i, since both square to −1. The symbol system's domain and valid operations belong in the specification. MacTutor's histories are later scholarly accounts of mathematical traditions; this modern signed and complex notation must not be attributed verbatim to their historical subjects.

Check your reasoning

  • States the nonzero precondition and recovers both roots.
  • Verifies rather than merely lists solutions.
  • Distinguishes domain extension from invalid cancellation.

2. Representation and calculation · allow 25 minutes

Zero and missing data require different states

Three observations are 0, 10 and missing. Compute the mean of observed values. A program stores missing as integer zero and averages all three slots. Compute its answer and describe the information that was lost. Design a representation that also handles a genuine zero and an explicit invalid measurement. State what assumptions would be needed to estimate a mean for the full population of three.

Get a starting hint

An encoding should preserve distinctions that later operations need, not merely fit every case in one numeric field.

Compare your reasoning with the worked solution

The observed-value mean is (0+10)/2=5. Replacing missing with zero gives 10/3≈3.333, falsely treating absence as an observation. Once both genuine zero and missing use the same untagged value, the original cases cannot be reconstructed. Use a tagged value such as Observed(number), Missing(reason) and Invalid(reason), with aggregation explicitly selecting observed values and reporting the count. The full-population mean is (10+x)/3 for the unknown third value x; no unique number follows without additional information or an imputation model. The mean of observed cases estimates the broader population only under justified assumptions about the missingness process. An arithmetic zero, a placeholder and missing evidence are different states even when convenient encodings sometimes collapse them.

Check your reasoning

  • Calculates the two different means with explicit denominators.
  • Preserves zero, absence and invalidity with tags.
  • States the extra assumption needed for a population estimate.

Check your understanding

What does completing the square preserve?

  • The set of solutions when equivalent operations are applied correctly
  • Only the positive solution by definition
  • The original written form of the equation
Explanation

(x + 3)² = 25 gives x = 2 or x = −8. Substituting gives 4 + 12 = 16 and 64 − 48 = 16. A stored zero is a known numerical value. A missing measurement is an absence of information; replacing it with zero changes the problem and may bias a calculation. Tests should include the negative solution rather than retaining only the historically positive case.

Sources and further reading