5.4 Trigonometric models

Fundamentals of Trigonometric Models

Trigonometric models use sine and cosine functions to describe anything that repeats in a cycle. Tides, temperature swings, sound waves, electrical currents: if a phenomenon oscillates, there's a good chance a trig model can capture it. This section covers the building blocks you'll need before constructing or analyzing any model.

Periodic Functions

A periodic function repeats its values at regular intervals. The length of one full cycle is called the period. Think of a heartbeat on an EKG monitor: the same shape appears over and over, spaced evenly in time.

• Frequency is the number of complete cycles per unit of time (the inverse of the period).
• Wavelength is the spatial distance between repeating parts of a wave.
• Periodic behavior shows up in nature (planetary orbits, ocean waves) and in engineered systems (alternating current, clock pendulums).

Sine and Cosine Functions

Both sine and cosine come from the unit circle. If you trace a point moving around a circle of radius 1, its vertical position over time gives you sine, and its horizontal position gives you cosine.

• $y = \sin(x)$ oscillates between $-1$ and $1$, starting at $0$.
• $y = \cos(x)$ oscillates between $-1$ and $1$, starting at $1$.
• Both have a period of $2\pi$ radians (360°).
• They're the same shape, just shifted: $\sin(x) = \cos(x - \pi/2)$.

This phase-shift relationship means you can always rewrite a sine model as a cosine model and vice versa. Choose whichever fits your data's starting point more naturally.

Amplitude and Period

Amplitude is the maximum displacement from the midline. In the general form $y = A \sin(Bx)$, the amplitude is $|A|$. A larger amplitude means taller peaks and deeper troughs.

Period is how long one full cycle takes. It's calculated as:

$T = \frac{2\pi}{|B|}$

where $B$ is the angular frequency coefficient. A larger $|B|$ compresses the wave horizontally (shorter period, more cycles in the same space). A smaller $|B|$ stretches it out.

For example, a sound wave with a high frequency (large $B$) has a short period and a high pitch. Cranking up $A$ makes it louder.

Phase Shift and Vertical Shift

The full general form of a trigonometric model is:

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

• Phase shift ($C$): slides the curve left or right. A positive $C$ shifts the graph to the right. This is useful when a cycle doesn't start at $x = 0$, like a temperature model where the hottest day isn't January 1.
• Vertical shift ($D$): moves the entire curve up or down. This sets the midline of the oscillation. If average daily temperature is 15°C and it swings ±10°C, then $D = 15$ and $A = 10$.

Applications of Trigonometric Models

Trig models aren't just textbook exercises. They show up anywhere you find repeating patterns, and recognizing when a situation is periodic is half the battle.

Real-World Phenomena

• Tides: Driven by gravitational pull from the moon and sun, tidal heights follow roughly sinusoidal patterns with a period of about 12.4 hours for semidiurnal tides.
• Circadian rhythms: Body temperature, hormone levels, and sleep-wake cycles oscillate on approximately 24-hour periods.
• Alternating current (AC): Household electricity in the U.S. follows a 60 Hz sine wave (period of about 0.0167 seconds).
• Seasonal temperature: Annual temperature in many locations can be modeled with a period of 12 months.

Sound Waves and Vibrations

Sound travels as pressure variations through a medium. A pure tone is a single sine wave where:

• Frequency determines pitch (440 Hz = the note A above middle C).
• Amplitude determines loudness.

Real-world sounds are rarely pure tones. A guitar string, for instance, produces a fundamental frequency plus overtones. You can model complex sounds as sums of simple sine waves, which is the basis of audio engineering and noise-cancellation technology.

Seasonal Patterns

Annual temperature in a city like Chicago might be modeled as:

$T(t) = 14 \sin\left(\frac{2\pi}{12}(t - 4)\right) + 10$

where $T$ is temperature in °C and $t$ is the month number. The amplitude of 14 captures the swing from winter lows to summer highs, the period is 12 months, the phase shift of 4 accounts for peak warmth in July rather than January, and the vertical shift of 10 sets the annual average.

These models also apply to daylight hours, agricultural yield forecasting, energy demand planning, and tourism patterns.

Circular Motion

When an object moves in a circle at constant speed, its $x$- and $y$-coordinates are cosine and sine functions of time, respectively. A point on a Ferris wheel at angle $\theta$ has position:

• $x = r\cos(\theta)$
• $y = r\sin(\theta)$

This connects directly to angular velocity and shows up in engineering (gears, pulleys, crankshafts) and physics (planetary orbits, satellite tracking).

Constructing Trigonometric Models

Building a trig model from data is a core skill. The process goes from raw observations to a usable equation.

Identifying Key Parameters

Given a set of periodic data, extract the four parameters step by step:

1. Find the amplitude: $A = \frac{\text{max value} - \text{min value}}{2}$

2. Find the vertical shift (midline): $D = \frac{\text{max value} + \text{min value}}{2}$

3. Find the period: Measure the time (or distance) between two consecutive peaks or troughs. Then $B = \frac{2\pi}{T}$.

4. Find the phase shift: Locate where a peak or zero-crossing occurs. If using a sine model, a peak occurs at $x = C + \frac{\pi}{2B}$, so solve for $C$.

Plug everything into $y = A \sin(B(x - C)) + D$.

Fitting Data to Models

When data is messy or doesn't perfectly match a sine curve:

1. Plot the data points to confirm the pattern looks periodic.
2. Estimate parameters visually using the steps above.
3. Use regression tools (Excel's Solver, MATLAB's fit function, Python's scipy.optimize.curve_fit) to refine the fit.
4. Check the residuals (observed minus predicted). If they show a pattern, the model may need adjustment.
5. Iterate: adjust parameters or try a different form (cosine instead of sine, or add harmonics).

Graphical Representations

A good graph makes your model's quality immediately visible. Plot the fitted curve alongside the original data points, using different colors or markers to distinguish them. Label axes with units, and mark key features like peaks, troughs, and zero-crossings. This visual check often reveals problems that numerical measures alone might miss.

Algebraic Expressions

Once you've settled on parameter values, write the model in standard form:

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

Express $B$ in terms of the period ($B = 2\pi / T$) so readers can immediately see the cycle length. Make sure your units are consistent throughout, and briefly state what each parameter means in context. For instance: "$A = 3.2$ meters represents the tidal range from mean sea level."

Analysis of Trigonometric Models

Once you have a model, you need to extract useful information from it. This section covers the main analytical techniques.

Finding Extrema

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

• Maximum value = $A + D$, occurring when $\sin(B(x - C)) = 1$
• Minimum value = $-A + D$, occurring when $\sin(B(x - C)) = -1$

To find where these occur, solve:

• Maxima at $x = C + \frac{\pi}{2B} + \frac{2\pi n}{B}$ (for integer $n$)
• Minima at $x = C + \frac{3\pi}{2B} + \frac{2\pi n}{B}$

This is how you'd answer questions like "In which month is the temperature highest?" or "At what time does the tide peak?"

Determining Zeros

Setting $y = 0$ gives:

$\sin(B(x - C)) = -\frac{D}{A}$

This only has solutions when $\left|\frac{D}{A}\right| \leq 1$. Use the inverse sine function to find the principal solution, then use periodicity to find all solutions in your domain. Zeros represent equilibrium crossings: the moments when a pendulum passes through center, or when temperature crosses the freezing point.

Symmetry and Periodicity

• Sine is an odd function about its phase-shift point ($x = C$), meaning $\sin(-\theta) = -\sin(\theta)$.
• Cosine is an even function, meaning $\cos(-\theta) = \cos(\theta)$.

Because the function repeats every period $T = 2\pi / |B|$, you only need to fully analyze one cycle. Every value, slope, and feature repeats after that. This is what makes trig models so useful for prediction: once you understand one cycle, you understand them all.

Rate of Change

The derivative tells you how fast the quantity is changing at any moment.

• For $y = A \sin(Bx)$: $y' = AB \cos(Bx)$
• The second derivative: $y'' = -AB^2 \sin(Bx)$

The rate of change is greatest at the zero-crossings (where the curve is steepest) and zero at the peaks and troughs (where the curve momentarily flattens). The second derivative identifies inflection points where the curve switches from concave up to concave down. In physics, if $y$ is position, then $y'$ is velocity and $y''$ is acceleration.

Advanced Trigonometric Modeling

Real-world signals are rarely perfect sine waves. These techniques handle more complex situations.

Composite Trigonometric Functions

When a single sine or cosine isn't enough, combine them:

• Sum of sines: $y = A_1 \sin(B_1 x) + A_2 \sin(B_2 x) + \cdots + C$
• Product forms: $y = A \sin(Bx) \cdot \cos(Cx)$

Trig identities can simplify products into sums. For example, $\sin(Bx)\cos(Cx) = \frac{1}{2}[\sin((B+C)x) + \sin((B-C)x)]$. This decomposition is the foundation of signal processing and harmonic analysis.

Damped Oscillations

Many oscillating systems lose energy over time. A swinging pendulum slows down; a plucked guitar string fades. The model multiplies a trig function by an exponential decay:

$y = Ae^{-kx} \sin(Bx)$

Here $k > 0$ is the damping coefficient. Larger $k$ means faster decay. The oscillation's "envelope" shrinks exponentially while the frequency stays roughly the same (though heavy damping can alter it). Shock absorbers and RLC circuits are classic examples.

Forced Oscillations

When an external periodic force acts on an oscillating system, you get:

$y = A \sin(\omega t) + B \sin(\Omega t)$

where $\omega$ is the system's natural frequency and $\Omega$ is the forcing frequency. When $\Omega \approx \omega$, resonance occurs and the amplitude can grow dramatically. This is why soldiers break step on bridges, and why opera singers can shatter glass at the right pitch.

Fourier Series Basics

Any periodic function (even a square wave or sawtooth) can be represented as an infinite sum of sines and cosines:

$f(x) = \frac{a_0}{2} + \sum_{n=1}^{\infty} \left(a_n \cos(nx) + b_n \sin(nx)\right)$

The coefficients $a_n$ and $b_n$ are found by integrating the function over one period. Each term in the sum represents a different frequency component. This is the mathematical basis for MP3 compression, MRI imaging, and solving partial differential equations.

Limitations and Alternatives

No model fits every situation. Knowing when a trig model doesn't work is just as important as knowing when it does.

Non-Periodic Phenomena

If the data doesn't repeat, a trig model is the wrong tool. Consider instead:

• Exponential functions for growth or decay (population growth, radioactive decay)
• Polynomial functions for smooth, non-repeating trends
• Piecewise functions for data with distinct phases

For time series data that has both a trend and a seasonal component, you can separate them: fit a linear or exponential trend first, then model the residual oscillation with a trig function.

Piecewise Functions

Sometimes behavior changes abruptly. A piecewise function uses different equations on different intervals. For example, a progressive tax system charges different rates at different income levels. When building piecewise models, check whether continuity at the transition points matters for your application.

Exponential vs. Trigonometric Models

Compare models using goodness-of-fit measures like $R^2$ and residual plots. Sometimes the best model combines both types.

Combining Model Types

Complex real-world data often requires hybrid models:

• Trend + oscillation: $y = mx + b + A\sin(Bx)$ captures a linear trend with seasonal variation.
• Damped oscillation: $y = Ae^{-kx}\sin(Bx)$ combines exponential decay with periodic behavior.
• Bounded oscillatory growth: A logistic function multiplied by a periodic term.

Regression analysis helps determine which combination best fits your data without overfitting.

Technology in Trigonometric Modeling

You'll rarely build trig models by hand in practice. Knowing which tools to reach for saves time and reduces errors.

Graphing Calculators

Graphing calculators let you visualize transformations in real time. Adjust $A$, $B$, $C$, and $D$ and watch the curve respond. Built-in solvers can find zeros and extrema numerically. These are especially useful during exams when you need quick checks.

Computer Algebra Systems

Tools like Mathematica, Maple, and Wolfram Alpha handle symbolic manipulation: simplifying trig identities, computing derivatives and integrals, solving equations exactly. They can also generate 3D plots for multivariable trig functions and compute Fourier coefficients.

Data Analysis Software

For real datasets, use R, Python (with NumPy/SciPy), or MATLAB:

• Import and clean data
• Fit trig models using nonlinear regression (curve_fit in Python, nls in R)
• Generate residual plots and compute $R^2$
• Run sensitivity analysis to see how parameter changes affect predictions

Simulation Tools

Simulation software lets you model systems with multiple interacting periodic components. You can adjust parameters in real time, add random noise to study robustness, and visualize how the system evolves. Applications range from mechanical vibration analysis to electromagnetic wave propagation.

Interpreting Trigonometric Models

A model is only useful if you can explain what it means. This is where mathematical thinking meets real-world decision-making.

Physical Significance of Parameters

Always connect parameters back to their real-world meaning:

Predicting Future Values

Because trig models are periodic, you can extrapolate to predict future cycles. But be cautious:

• The model assumes conditions stay the same. Climate change, for instance, can shift the baseline over decades.
• Use confidence intervals to express how uncertain your predictions are.
• The further you extrapolate beyond your data, the less reliable the prediction.

Error Analysis

After fitting a model, always check how well it performs:

1. Calculate residuals for each data point (observed $-$ predicted).
2. Plot residuals against $x$. Random scatter is good; a visible pattern suggests the model is missing something.
3. Compute $R^2$ to quantify overall fit (closer to 1 is better).
4. Check for outliers that might be skewing the fit.

Model Refinement Techniques

If the fit isn't good enough:

• Add higher harmonics ($A_2 \sin(2Bx)$, etc.) to capture more complex wave shapes.
• Try a different base function (cosine instead of sine, or a composite model).
• Incorporate additional variables if the system depends on more than one input.
• Use cross-validation: fit the model on part of your data and test it on the rest to guard against overfitting.