Algebra and scaling: preserving meaning while rearranging
Solve equations reversibly, compare growth laws, and expose scale with dimensionless ratios.
On this page
The question for today
What you’ll understand
- Track restrictions when rearranging an equation.
- Predict multiplicative changes from power laws.
- Distinguish absolute change from relative change and check dimensions.
Algebra replaces a problem with an equivalent one that is easier to inspect. Scaling asks a complementary question: how does a result change when an input is multiplied by some factor? Together they let you reason before calculating. This matters when comparing the energy of moving objects, estimating the memory needed by a model, or checking whether a proposed formula could possibly have the right units.
Rearrange without losing or inventing solutions
Adding the same number to both sides preserves equality. Multiplying both sides by a known nonzero number does too. Dividing by an expression requires checking that it is nonzero. For example, canceling from discards the valid solution unless you handle it separately. Instead factor to obtain , which makes both possible zero factors visible.
Some operations can introduce extra candidates. Squaring both sides of gives , which also admits . The squared equation is necessary but no longer sufficient. Treat results from a nonreversible step as candidates and substitute them into the original equation. A reliable solution records restrictions when they arise, rather than attempting to reconstruct them at the end.
Read a power as a scaling instruction
For a power law, an input change by a factor produces a predictable output factor. If area scales with the square of a length, doubling every length multiplies area by four. If volume scales with the cube, it multiplies volume by eight. The coefficients and shapes must remain fixed for that comparison. Scaling one side of a rectangle while holding the other fixed is a different experiment.
A ratio removes the coefficient and often removes units. That makes it useful for comparisons even when you do not know the absolute value. Power laws do not describe every dependence: adding fixed overhead breaks a pure power relation. In engineering, doubling user traffic need not double total resource use if a substantial fixed cost is present, and thresholds can produce entirely different behavior.
Compare like with like
Absolute change is the new value minus the old value. Relative change divides that difference by the old value, which must be nonzero. Growing from 20 to 30 adds 10 but represents a 50 percent increase. Returning from 30 to 20 is a one-third decrease, because the reference has changed. State the baseline when comparing improvements in accuracy, latency, or throughput.
Dimensional reasoning is another form of disciplined comparison. In , speed times time has units of length. In , it does not, unless the coefficient has different units from speed. Dimensionless combinations such as time divided by a reference time let you compare systems measured in different units. A unit check rejects inconsistent formulas, although many dimensionally valid formulas still describe the wrong physics.
Work it through
Separate area scaling from a fixed charge
A square panel has side length 3 m and coating usage 0.2 liters per square meter. A job also needs a fixed 1 liter for setup. How much coating is needed if the side becomes 6 m?
The side is multiplied by 6/3 = 2, so area is multiplied by the square of that factor, 4. The new area is 6 × 6 = 36 square meters.
Multiply area by usage: 36 square meters × 0.2 liters per square meter = 7.2 liters. The area units cancel, leaving liters.
Add the fixed setup amount after scaling the area-dependent part. The total is 7.2 + 1 = 8.2 liters. Multiplying the original entire total by four would incorrectly scale setup too.
The new job requires 8.2 liters under the stated constant-coverage model.
The living notebook
Explore the model
Interactive lab · function
Test a quadratic scaling law
Keep coefficient a fixed in the lab's model, and compare the outputs at x and twice x. Then change a while holding x fixed. The controls expose quadratic input scaling and linear coefficient scaling; this family has no fixed overhead term.
- y at chosen x
- 0.608
- Tangent slope · 2ax
- 1.35
- Signed area · ax³/3
- 0.182
This lab uses only y = ax². Playback traces the curve within the fixed integration interval from 0 to x; it does not estimate numerical error or change the chosen point. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.
Look forIdentify which quantities remain fixed before claiming a scaling law.
Pause & reason
Check your understanding
Your turn
Try it yourself
A toy computation takes T(n) = 2 + 0.5n² milliseconds for a positive integer n. Compare T(4) and T(8). Is total time multiplied by four?
Take a moment to make a prediction or work it out on paper. Open a hint when you need one.
Give me a hint
Scale the quadratic term, but retain the fixed 2 ms.
Compare with the solution
T(4) = 10 ms and T(8) = 34 ms. The variable part increases from 8 to 32 ms, exactly a factor of four, while the fixed 2 ms stays unchanged. Total time grows by a factor of 3.4. This follows from the toy formula, not a measured complexity claim.
Keep the idea
Take it with you
- Track every nonzero or domain assumption required by an algebraic step.
- A power law predicts ratios only while its other ingredients stay fixed.
- Dimensionless comparisons expose scale without confusing units or baselines.
Read further
Original sources and deeper explanations to follow at your own pace.
Optional progress, saved in this browser when available. Revisit any lesson at any time.
Lesson not marked complete.