Inequality multiplication (positive)

If $a > b$ and $m > 0$, then $am > bm$ and $\frac{a}{m} > \frac{b}{m}$

Deep Breakdown

What it tells me (Situation Awareness): Multiplying or dividing both sides of an inequality by the same positive value preserves the direction of the inequality. This is the safe half of the multiplication rule. The dangerous half — multiplying by a negative — flips the direction (covered separately). This rule is used any time you scale a comparison: normalizing revenue by number of days, converting units, comparing per-user metrics, or scaling constraints in optimization. It is the algebraic license to rescale without breaking the relationship.

Flowchart

Real-world use

Where it is used

Normalizing metrics by a common denominator (revenue per session, cost per click, installs per day), scaling inequality constraints in LP models, converting units (daily clicks to weekly by multiplying by 7 preserves the ranking), and comparing rates where you divide a count by a positive total.

Problem it solves

In practice, you often need to rescale both sides of an inequality to make it interpretable or to isolate a variable. For example, if total revenue $R_A > R_B$ and both have run for the same number of days $d$, you want to confirm that $R_A / d > R_B / d$ — the daily rate comparison inherits the total comparison. This rule guarantees that dividing by $d > 0$ is valid.

How it solves the problem

Let $a - b > 0$ (the gap). Multiply by $m > 0$: $m(a - b) > 0$ because a positive times a positive is positive. So $ma - mb > 0$, which means $ma > mb$. Same logic for division: divide by $m > 0$ gives $(a - b)/m > 0$, so $a/m > b/m$. The key is that the gap $a - b$ stays positive after scaling by any $m > 0$.

Example in a business

Campaign A had 12,000 clicks in 30 days; Campaign B had 9,000 clicks in 30 days. So $12000 > 9000$. Divide both by 30 (positive): $400 > 300$ clicks/day. The per-day comparison is valid and the direction is preserved. Now suppose you want to scale both by a $\times 3$ budget multiplier: $36000 > 27000$. The ranking holds. This is the inequality multiplication rule in action for rate normalization.

The formula, piece by piece

• $a, b$: any real numbers with $a > b$
• $m$: a strictly positive real number ($m > 0$)
• Multiplication: $am > bm$ — direction unchanged
• Division: $\dfrac{a}{m} > \dfrac{b}{m}$ — direction unchanged (dividing by $m$ = multiplying by $\tfrac{1}{m}$, which is also positive)
• Works for all inequality types: $>, <, \geq, \leq$ — as long as $m > 0$ Full rule: $a > b$ and $m > 0 \implies am > bm$ and $\dfrac{a}{m} > \dfrac{b}{m}$

The formula in words

"Multiplying or dividing both sides of an inequality by the same positive number leaves the direction of the inequality unchanged." $9 > 6$: multiply by $4$: $36 > 24$. Divide by $3$: $3 > 2$. Direction preserved in both.

Critical Concepts

• Only safe with $m > 0$. If $m < 0$, the inequality flips: $9 > 6$ but $9 \times (-1) = -9 < -6 = 6 \times (-1)$. That is the separate negative-multiplier rule.
• Dividing is just multiplying by the reciprocal: $\frac{a}{m} = a \cdot \frac{1}{m}$. Since $m > 0 \implies \frac{1}{m} > 0$, division by a positive number falls under the same rule.
• Used to isolate variables with positive coefficients: $3x > 15 \implies x > 5$ (divide both sides by $3 > 0$, direction holds).
• Works for compound inequalities: if $a < b < c$ and $m > 0$, then $am < bm < cm$.
• Never assume $m > 0$ without checking. In data pipelines, a "denominator" could theoretically be zero or negative — always validate before dividing to normalize.

Step-by-step calculation (tiny example)

Step 1: Start with a known inequality and positive multiplier Given: $3x > 15$. We want to isolate $x$. Here $m = 3 > 0$.

Step 2: Divide both sides by 3 (positive, so direction is preserved) $\frac{3x}{3} > \frac{15}{3}$, which gives $x > 5$.

Step 3: Verify the direction is preserved Pick $x = 7 > 5$: $3(7) = 21 > 15$. Satisfied. Pick $x = 4 < 5$: $3(4) = 12 \not> 15$. Correctly excluded. Confirmed.

Code examples

Beginner example (Python)

# Inequality multiplication rule: if a > b and m > 0, then a*m > b*m and a/m > b/m
a = 12.0
b = 9.0
m_values = [0.5, 1.0, 3.0, 100.0]
print(f"Starting inequality: {a} > {b} (True: {a > b})")
print()
print(f"{'m':>8} | {'a*m':>8} | {'b*m':>8} | {'a*m > b*m':>12} | {'a/m':>8} | {'b/m':>8} | {'a/m > b/m':>12}")
print("-" * 75)
for m in m_values:
    a_times_m = a * m
b_times_m = b * m
a_div_m = a / m
b_div_m = b / m
print( f"{m:>8.1f} | {a_times_m:>8.1f} | {b_times_m:>8.1f} | {str(a_times_m > b_times_m):>12}" f" | {a_div_m:>8.3f} | {b_div_m:>8.3f} | {str(a_div_m > b_div_m):>12}" )   # Solving an inequality: 3x > 15
print("\nSolving inequality: 3x > 15")
print("Divide both sides by 3 (positive): x > 5")
threshold = 15 / 3
for x in [3, 5, 6, 10]:
    original = 3 * x > 15
simplified = x > threshold
print(f" x={x}: original (3x>15): {original}, simplified (x>5): {simplified}")

Beginner example (R)

# Inequality multiplication rule: if a > b and m > 0, then a*m > b*m and a/m > b/m
a <- 12
b <- 9
m_values <- c(0.5, 1.0, 3.0, 100.0)
cat(sprintf("Starting: %d > %d (True: %s)\n\n", a, b, a > b))
cat(sprintf("%8s | %8s | %8s | %12s | %8s | %8s | %12s\n", "m", "a*m", "b*m", "a*m>b*m", "a/m", "b/m", "a/m>b/m"))
cat(strrep("-", 80), "\n")
for (m in m_values) { cat(sprintf("%8.1f | %8.1f | %8.1f | %12s | %8.3f | %8.3f | %12s\n", m, a*m, b*m, as.character(a*m > b*m), a/m, b/m, as.character(a/m > b/m)))
}  # Solving: 3x > 15 -> x > 5
cat("\nSolving: 3x > 15 -> x > 5\n")
for (x in c(3, 5, 6, 10)) { cat(sprintf(" x=%d: original=%s, simplified=%s\n", x, 3*x>15, x>5))
}

Professional example (Python)

import pandas as pd
import numpy as np   # Scenario: normalizing campaign metrics by campaign duration (days)
# Rule: if total_clicks_A > total_clicks_B and both ran for duration > 0 days,
# then daily_clicks_A > daily_clicks_B (dividing by positive duration preserves ranking)
np.random.seed(42)
campaigns = pd.DataFrame({ "campaign": ["Alpha", "Beta", "Gamma", "Delta", "Epsilon"], "total_clicks": [36000.0, 27000.0, 15000.0, 48000.0, 21000.0], "duration_days": [30.0, 30.0, 15.0, 60.0, 21.0],
})   # Compute daily rate: dividing by duration (positive) preserves order within same-duration groups
campaigns["daily_clicks"] = campaigns["total_clicks"] / campaigns["duration_days"]   # Rank by total and by daily — they should differ when durations differ
campaigns["rank_total"] = campaigns["total_clicks"].rank(ascending=False).astype(int)
campaigns["rank_daily"] = campaigns["daily_clicks"].rank(ascending=False).astype(int)   # Verify: for same-duration pairs, multiplying/dividing by duration must preserve order
same_duration_pairs = campaigns[campaigns["duration_days"] == 30.0].copy()
p1, p2 = same_duration_pairs.iloc[0], same_duration_pairs.iloc[1]
total_order_preserved = (p1["total_clicks"] > p2["total_clicks"]) == ( p1["daily_clicks"] > p2["daily_clicks"] )
print("Campaign Metric Normalization: Inequality Multiplication Rule")
print("Rule: total_A > total_B and duration > 0 => daily_A > daily_B")
print()
print(campaigns[["campaign", "total_clicks", "duration_days", "daily_clicks", "rank_total", "rank_daily"]].to_string(index=False))
print(f"\nFor same-duration campaigns (30 days): order preserved after dividing = {total_order_preserved}")
print("Note: different durations change the ranking — normalization is essential.")

Professional example (R)

# total_A > total_B and duration > 0 => daily_A > daily_B (inequality multiplication)
campaigns <- data.frame( campaign = c("A", "B", "C"), total_clicks = c(900, 600, 450), duration_days = c(30, 30, 15)
)
campaigns$daily_clicks <- campaigns$total_clicks / campaigns$duration_days
cat("Inequality multiplication: total/duration preserves order when duration equal\n")
print(campaigns)

Source: domain knowledge

#algebra#powers-roots-and-inequalities#equations-and-inequalities