Trigonometric Modeling

⚡ Key Takeaways

Trigonometric modeling is the use of sinusoidal functions — $y = A\sin(B(x-C)) + D$ or its cosine equivalent — to describe real-world phenomena that repeat in cycles. Key result: any smooth periodic behavior (tides, daylight hours, sound, temperature, AC current) can be captured by choosing four parameters: amplitude $A$ , period $\tfrac{2\pi}{B}$ , phase shift $C$ , and vertical shift $D$ . Applies to any quantity that oscillates between a maximum and minimum at regular intervals. Who/when: rooted in ancient astronomy (Hipparchus, ~150 BCE), formalized by Fourier (1807).

💡 The Core Insight

A periodic phenomenon has four independent features: how far it swings (amplitude), how often it repeats (period/frequency), where in the cycle it starts (phase), and what value it oscillates around (midline). The four parameters $A$ , $B$ , $C$ , $D$ of a sinusoid map exactly onto these four physical features. Modeling = reading those four features off a situation and plugging them in.

DefinitionThe General Sinusoidal Function

A sinusoidal model is any function of the form

1

\quad y = A\sin\!\big(B(x - C)\big) + D

Derivation hint: start from the parent function $y=\sin x$ and apply the four standard transformations — vertical stretch, horizontal stretch, horizontal shift, vertical shift — in that order. where the symbols mean:

$A$ = amplitude = half the distance between the maximum and minimum. Always taken $A>0$ ; a reflection is handled separately. $\;A = \tfrac{\text{max}-\text{min}}{2}$ .
$B$ = angular frequency, controlling the period $P$ . Here $B>0$ .
$C$ = phase shift (horizontal shift), the $x$ -value where the sine portion begins its standard cycle.
$D$ = vertical shift = the midline $y=D$ , the average of max and min. $\;D = \tfrac{\text{max}+\text{min}}{2}$ .

The period and frequency are linked to $B$ by:

2

\quad P = \frac{2\pi}{B}, \qquad f = \frac{1}{P} = \frac{B}{2\pi}

Derivation hint: $\sin$ completes one cycle as its argument increases by $2\pi$ ; set $B(x-C)$ to increase by $2\pi$ and solve for the change in $x$ .

TheoremSine–Cosine Equivalence

Every sine model can be rewritten as a cosine model and vice versa, because cosine is sine shifted left by a quarter period:

3

\quad A\sin\!\big(B(x-C)\big) + D = A\cos\!\Big(B\big(x - C + \tfrac{P}{4}\big)\Big) + D

Derivation hint: use the identity $\sin\theta = \cos(\theta - \tfrac{\pi}{2})$ and convert $\tfrac{\pi}{2}$ into an $x$ -shift by dividing by $B$ , noting $\tfrac{\pi/2}{B} = \tfrac{P}{4}$ .

TheoremSum-of-Sinusoids Form

A combination $a\sin(Bx) + b\cos(Bx)$ is itself a single sinusoid:

4

\quad a\sin(Bx) + b\cos(Bx) = R\sin(Bx + \varphi), \quad R=\sqrt{a^2+b^2}, \quad \tan\varphi = \frac{b}{a}

Derivation hint: expand $R\sin(Bx+\varphi)=R\cos\varphi\sin(Bx)+R\sin\varphi\cos(Bx)$ and match coefficients with $a$ and $b$ .

ProofBuilding the Model from Four Features

We show that knowing max, min, period, and a reference point determines $A,B,C,D$ uniquely. 1. Midline (D). The curve oscillates symmetrically about $y=D$ . Since the max occurs at $D+A$ and the min at $D-A$ , adding them gives $\text{max}+\text{min}=2D$ , so $D=\tfrac{\text{max}+\text{min}}{2}$ . 2. Amplitude (A). Subtracting instead: $\text{max}-\text{min}=2A$ , so $A=\tfrac{\text{max}-\text{min}}{2}$ . 3. Angular frequency (B). Measure the period $P$ (time/distance for one full repeat). From equation [2], $B=\tfrac{2\pi}{P}$ . 4. Phase shift (C). For a sine model, the curve crosses its midline going upward at $x=C$ . Locate one such crossing in the data and set $C$ equal to that $x$ -value. (For a cosine model, $C$ is instead where a maximum occurs.) 5. Assembly. Substituting these four values into [1] reproduces the prescribed max, min, period, and starting point, and by uniqueness of the sinusoid through these constraints the model is determined. $\blacksquare$

∎

JF

Joseph Fourier

1768–1830

Jean-Baptiste Joseph Fourier, studying how heat diffuses through solids, made the revolutionary claim (1807) that any periodic function — even a square wave with sharp corners — can be expressed as an infinite sum of sines and cosines. This Fourier series elevated trigonometric modeling from a tool for simple oscillations to the foundation of all signal analysis. Fourier also coined the term "greenhouse effect" in his climate work. His insight underlies MP3 compression, MRI imaging, and JPEG photographs — every one decomposes a signal into sinusoids.

🧠 Intuition

Picture a point riding around a Ferris wheel at constant speed. Its height traces a perfect sine curve as time passes:

The wheel's radius is the amplitude $A$ — how high above and below center you go.
The time for one full rotation is the period $P$ — set the speed and you set $B=2\pi/P$ .
The center height of the wheel above the ground is the midline $D$ .
Where you started boarding (bottom, side, top) is the phase shift $C$ .

Change one knob and the others stay put — that independence is what makes the four-parameter form so powerful. Modeling is just watching the world's "Ferris wheels" — tides, days, sound waves — and reading off these four numbers.

Figure

A sinusoid

y=A\sin(B(x-C))+D

with its four parameters labeled

```svg

```

ExampleHours of Daylight

Given: At a city, the longest day (summer solstice) has 15 hours of daylight; the shortest (winter solstice) has 9 hours. The longest day is day number $t=172$ of the year (June 21). Model daylight hours $L(t)$ where $t$ is the day of the year ( $t=1$ on Jan 1).

Step 1 — Midline: $D = \tfrac{15+9}{2} = 12$ hours.

Step 2 — Amplitude: $A = \tfrac{15-9}{2} = 3$ hours.

Step 3 — Period: One year, so $P = 365$ days, giving $B = \tfrac{2\pi}{365} \approx 0.01721$ per day.

Step 4 — Phase: Daylight is at its maximum at $t=172$ . A cosine peaks at the start of its argument, so use a cosine with $C=172$ :

L(t) = 3\cos\!\Big(\tfrac{2\pi}{365}(t-172)\Big) + 12.

Check at $t=172$ : $\cos(0)=1$ , so $L=3(1)+12=15$ ✓. Half a year later ( $t\approx 354.5$ ), $\cos(\pi)=-1$ , giving $L=9$ ✓.

Predict March 1 ( $t=60$ ):

L(60)=3\cos\!\big(\tfrac{2\pi}{365}(60-172)\big)+12 = 3\cos(-1.928)+12 \approx 3(-0.351)+12.

\boxed{L(60) \approx 10.95 \text{ hours of daylight}}

ExampleTidal Heights

Given: A harbor's water depth is 9 m at high tide and 3 m at low tide. High tide occurs at 3:00 AM ( $t=3$ h), and the next high tide is 12.4 hours later (semidiurnal tide). Find depth $h(t)$ and the depth at noon.

Step 1 — Midline: $D=\tfrac{9+3}{2}=6$ m.

Step 2 — Amplitude: $A=\tfrac{9-3}{2}=3$ m.

Step 3 — Period: $P=12.4$ h, so $B=\tfrac{2\pi}{12.4}\approx 0.5067$ per hour.

Step 4 — Phase: Maximum at $t=3$ , so use cosine with $C=3$ :

h(t)=3\cos\!\big(0.5067(t-3)\big)+6.

Predict noon ( $t=12$ ):

h(12)=3\cos\!\big(0.5067(12-3)\big)+6 = 3\cos(4.560)+6.

\cos(4.560)\approx -0.150,\quad h(12)\approx 3(-0.150)+6.

\boxed{h(12) \approx 5.55 \text{ m deep at noon}}

⚠️ Period vs. B — they are reciprocally related, not equal

The number $B$ is not the period. A large $B$ means a short period (fast oscillation), since $P=2\pi/B$ . Students routinely write "the period is $B$ " — wrong. Always convert: if a phenomenon repeats every 24 hours, then $P=24$ and $B=\tfrac{2\pi}{24}=\tfrac{\pi}{12}$ , not $B=24$ .

⚠️ The phase shift is C, not BC

In the form $y=A\sin(B(x-C))+D$ , the horizontal shift is exactly $C$ . But if the model is given un-factored as $y=A\sin(Bx-\phi)+D$ , the shift is $\tfrac{\phi}{B}$ , not $\phi$ . You must factor out $B$ first: $Bx-\phi = B\!\left(x-\tfrac{\phi}{B}\right)$ . Forgetting to divide by $B$ is the single most common modeling error.

CounterexampleNot every repeating shape is sinusoidal

A sinusoid is smooth and has a single max and min per cycle of equal, symmetric shape. The sawtooth wave (ramps up linearly, then drops instantly) and the square wave are periodic but not sinusoids — no single $A\sin(B(x-C))+D$ fits them. They require an infinite Fourier series of sinusoids. Likewise, a heartbeat (ECG) is periodic but sharply peaked, so a single sinusoid models it poorly even though its repetition rate is well-defined.

connection

Trigonometric modeling sits at a crossroads of several ideas:

It is built directly on the Unit Circle, which generates the sine and cosine values, and on Transformations of Functions, which supply the four parameters.
The sum-of-sinusoids identity [4] is the gateway to Fourier Series, where complicated periodic signals are decomposed into sinusoids.
Modeling oscillations links to Simple Harmonic Motion in physics, where the same equation governs springs and pendulums.
Fitting a sinusoid to noisy data connects to Regression (sinusoidal regression on a calculator), and the rate of change of these models is studied in Derivatives of Trigonometric Functions.

🔬 Where this appears

⏳ History

- ~150 BCE — Hipparchus compiles the first trigonometric "table of chords," motivated by predicting the periodic motions of the sun and moon — the earliest trigonometric modeling.

- ~500 CE — Indian astronomers (Aryabhata) introduce the half-chord (jya), essentially the modern sine, for periodic celestial computation.

- ~1000 CE — Islamic Golden Age (al-Battani, Abu al-Wafa) develop accurate sine and tangent tables, refining models of solar and lunar cycles.

- 1748 — Leonhard Euler publishes

e^{i\theta}=\cos\theta+i\sin\theta

, unifying sinusoids with exponentials and giving the analytic foundation for wave models.

- 1807 — Joseph Fourier submits his theory that any periodic function is a sum of sinusoids, the birth of Fourier analysis (published fully in 1822).

- 1865 — James Clerk Maxwell describes electromagnetic waves as sinusoidal oscillations of electric and magnetic fields, modeling light itself.

- 1880s — Lord Kelvin builds a mechanical tide-predicting machine that physically sums sinusoids to forecast harbor tides — trigonometric modeling made of brass gears.

- 1965 — Cooley & Tukey publish the Fast Fourier Transform (FFT), making sinusoidal decomposition computationally cheap and enabling modern digital signal processing.

open-problem

A single sinusoid assumes the period, amplitude, and midline are constant forever — but real systems often drift. Tides are quasi-periodic (the lunar and solar periods are incommensurate, so the pattern never exactly repeats); climate has trends superimposed on cycles; biological rhythms shift with jet lag. Capturing these requires modulated models — sinusoids whose amplitude or frequency themselves vary in time ( $A(t)$ , $B(t)$ ) — leading into amplitude/frequency modulation, wavelets, and time-frequency analysis. The deep open question in many sciences is separating a true periodicity from noise or coincidence: how many cycles of data are needed before a claimed sinusoidal pattern (e.g., a climate or economic cycle) is statistically real rather than an artifact? This is an active concern in spectral statistics.

Formal Statement

Why It Holds / Derivation

Building Intuition

Worked Examples

Common Misconceptions

Connections

Real-World Applications

Historical Development

Subtleties & Limits

See Also