Boo-AI — Master Artificial Intelligence by Building from Scratch

Learning Objectives

By the end of this section, you will be able to:

Understand how integration arises naturally when computing thermal energy with temperature-dependent specific heat
Apply the heat capacity integral $Q = m \int_{T_0}^{T_1} c(T) \, dT$ to calculate energy requirements
Derive and solve Newton's Law of Cooling as a separable differential equation
Calculate total heat transferred over time using integration of heat flow rates
Determine equilibrium temperatures using energy conservation principles
Connect thermal concepts to optimization and simulated annealing in machine learning

The Big Picture: Why Integration?

"Heat is energy in transit—and calculus is the language of accumulation and flow."

Heat transfer is one of the most fundamental phenomena in physics and engineering. Every time you heat water for coffee, insulate a building, or cool a computer processor, you are working with thermal energy. The mathematics of heat involves:

Energy storage: How much energy is needed to change temperature? This depends on mass, specific heat, and temperature range.
Energy flow: How fast does heat move between objects? This involves rates and differential equations.
Total energy transferred: Over a process, how much energy has flowed? This requires integration.

Integration enters whenever we need to accumulate quantities that vary. When specific heat changes with temperature, when heat flow rate changes over time, or when we sum up contributions from different parts of a system—integration is our essential tool.

Q = \int (\text{heat flow rate}) \, dt = \int mc(T) \, dT

Total thermal energy = integral of instantaneous contributions

Historical Context: The Science of Heat

The understanding of heat evolved dramatically over two centuries, transforming from a mysterious "caloric fluid" to our modern understanding of energy in motion.

Joseph Black (1728–1799)

Scottish physicist Joseph Black discovered specific heat and latent heat. He showed that different substances require different amounts of heat to achieve the same temperature change, and that phase changes (like melting ice) absorb heat without changing temperature.

Isaac Newton's Cooling Law (1701)

Newton observed that the rate of heat loss from an object is proportional to the temperature difference between the object and its surroundings. This simple observation leads to an exponential decay that we can derive using calculus.

James Prescott Joule (1818–1889)

Joule's famous experiments demonstrated the mechanical equivalent of heat—that work can be converted to heat and vice versa. This led to the first law of thermodynamics and the unit of energy bearing his name.

Units of Heat

Heat is measured in Joules (J) in SI units. Historically, the calorie was defined as the heat needed to raise 1 gram of water by 1°C. We now know 1 calorie = 4.186 J.

Heat Capacity and Thermal Energy

The fundamental equation relating heat and temperature change is:

Q = mc\Delta T

Heat = mass × specific heat × temperature change

Where:

$Q$ is the heat energy (Joules)
$m$ is the mass (kg)
$c$ is the specific heat capacity (J/kg·K)—the energy needed to raise 1 kg by 1 K
$\Delta T = T_1 - T_0$ is the temperature change

Material	Specific Heat (J/kg·K)	Physical Meaning
Water	4186	Exceptional heat storage
Air	1005	Moderate for a gas
Aluminum	900	Good for cookware
Iron	450	Dense, moderate capacity
Copper	385	Excellent conductor
Gold	129	Very low capacity

Why Water Has High Specific Heat

Water's high specific heat comes from hydrogen bonding between molecules. Much of the added energy goes into breaking these bonds rather than increasing molecular motion (temperature). This makes water excellent for thermal regulation—in oceans, organisms, and cooling systems.

Interactive: Heat Capacity Visualization

Explore how heat energy depends on material properties, mass, and temperature change. Notice that water requires far more energy than metals for the same temperature increase:

Heat Capacity and Thermal Energy: Q = mcΔT

Material

Excellent heat storage

Mass: 1.00 kg

Initial Temperature T₀: 20°C

Final Temperature T₁: 80°C

Calculation Summary

Material: Water

Specific Heat (c): 4186 J/(kg·K)

Mass (m): 1.00 kg

Temperature Change (ΔT): 60°C

Heat Energy (Q): 251.16 kJ

Physical Meaning: The shaded area represents the total thermal energy required to raise 1.0 kg of water from 20°C to 80°C. Water's high specific heat makes it excellent for thermal storage.

The Integral Form: When specific heat varies with temperature, we use integration: Q = m ∫ c(T) dT. For constant specific heat, this simplifies to Q = mcΔT.

Variable Specific Heat: When c Depends on T

For many materials, specific heat is not constant—it varies with temperature. This is especially true for:

Gases over wide temperature ranges
Metals near absolute zero
Substances near phase transitions
Many real-world engineering applications

When $c = c(T)$ , we cannot simply multiply. Instead, we integrate:

Heat Energy with Variable Specific Heat

Q = m \int_{T_0}^{T_1} c(T) \, dT

Each infinitesimal temperature increment $dT$ requires energy $mc(T) \, dT$

Example: Linear Temperature Dependence

Problem: A material has specific heat $c(T) = 1000 + 2T$ J/(kg·K). Find the heat needed to raise 3 kg from 20°C to 80°C.

Solution: We integrate the variable specific heat:

Q = m \int_{20}^{80} (1000 + 2T) \, dT = 3 \left[ 1000T + T^2 \right]_{20}^{80}

= 3 \left[ (80000 + 6400) - (20000 + 400) \right] = 3 \times 66000

Q = 198,000 \text{ J} = 198 \text{ kJ}

Compare to constant c = 1000: Q = 3 × 1000 × 60 = 180 kJ. The temperature dependence adds 18 kJ because c increases as we heat the material.

Newton's Law of Cooling

Newton observed that the rate of cooling is proportional to the temperature difference between an object and its surroundings:

\frac{dT}{dt} = -k(T - T_{ambient})

Rate of temperature change ∝ temperature difference

This is a separable first-order ODE. Let's solve it:

Step 1: Separate variables

\frac{dT}{T - T_{ambient}} = -k \, dt

Step 2: Integrate both sides

\int \frac{dT}{T - T_{ambient}} = -k \int dt

\ln|T - T_{ambient}| = -kt + C

Step 3: Solve for T

T - T_{ambient} = Ae^{-kt}

Step 4: Apply initial condition $T(0) = T_0$

A = T_0 - T_{ambient}

Newton's Law of Cooling: Solution

T(t) = T_{ambient} + (T_0 - T_{ambient})e^{-kt}

Temperature approaches ambient exponentially with time constant 1/k

Total Heat Transferred

The rate of heat flow is proportional to the temperature difference. If $hA$ is the heat transfer coefficient times surface area:

\dot{Q}(t) = hA(T(t) - T_{ambient})

The total heat transferred from time 0 to t is the integral:

Q = \int_0^t hA(T - T_{ambient}) \, d\tau = hA(T_0 - T_{ambient}) \int_0^t e^{-k\tau} \, d\tau

= hA(T_0 - T_{ambient}) \cdot \frac{1 - e^{-kt}}{k}

The Role of Integration

While the rate of heat transfer follows an exponential decay, the total heat transferred requires integrating this rate over time. The integral accumulates all the instantaneous heat flow contributions.

Interactive: Newton's Law of Cooling

Watch how temperature evolves according to Newton's Law. The shaded area represents the integral—total heat transferred to the environment:

Newton's Law of Cooling: dT/dt = -k(T - T_ambient)

Initial Temperature T₀: 90°C

Ambient Temperature: 20°C

Cooling Constant k: 0.050 min⁻¹

Higher k = faster cooling (e.g., thin metal cools faster than thick ceramic)

Current State

Time Elapsed: 0.0 min

Current Temperature: 90.0°C

Temperature Drop: 0.0°C

Relative Heat Lost: 0.0°C·equiv

The Shaded Area: Represents the integral ∫(T - T_ambient) dt, proportional to total thermal energy transferred to the environment. This integral approach is used in calculating energy consumption for cooling systems.

The Differential Equation

Newton's Law of Cooling states dT/dt = -k(T - T_ambient). This is a separable first-order ODE with solution:

T(t) = T_ambient + (T₀ - T_ambient)e^(-kt)

The total heat transferred from time 0 to t is found by integrating the heat flow rate: Q = ∫₀ᵗ hA(T - T_ambient) dt

Thermal Equilibrium

When two objects at different temperatures are brought into thermal contact, they exchange heat until reaching equilibrium. Conservation of energy requires:

\text{Heat lost by hot object} = \text{Heat gained by cold object}

Mathematically:

m_1 c_1 (T_1 - T_f) = m_2 c_2 (T_f - T_2)

Solving for the final equilibrium temperature $T_f$ :

Equilibrium Temperature

T_f = \frac{m_1 c_1 T_1 + m_2 c_2 T_2}{m_1 c_1 + m_2 c_2}

A weighted average where weights are thermal capacities mc

Interactive: Thermal Equilibrium

Explore how two bodies approach equilibrium. Notice that the body with higher thermal capacity (mc) dominates the final temperature:

Thermal Equilibrium: Energy Conservation via Integration

Hot Body (Water)

Mass: 2.0 kg

Initial Temp: 80°C

Cold Body (Aluminum)

Mass: 3.0 kg

Initial Temp: 20°C

Energy Balance

T₁ current: 80.0°C

T₂ current: 20.0°C

Equilibrium Temp: 65.4°C

Heat Transferred: 0.00 kJ

Shaded Area: The purple region between the curves represents the cumulative difference in temperatures over time, proportional to total heat transferred.

Energy Conservation: Heat lost by hot body = Heat gained by cold body. The equilibrium temperature is found by setting m₁c₁(T₁-T_f) = m₂c₂(T_f-T₂).

The Equilibrium Equation

Energy conservation requires that heat lost equals heat gained:

m₁c₁(T₁ - T_f) = m₂c₂(T_f - T₂)

Solving for T_f gives the equilibrium temperature:

T_f = (m₁c₁T₁ + m₂c₂T₂) / (m₁c₁ + m₂c₂)

Heat Flow Rate and Integration

Heat flow involves both rate (how fast energy moves) andaccumulation (total energy transferred). This naturally leads to integration.

Fourier's Law of Conduction

Heat conduction through a material follows Fourier's Law:

\dot{Q} = -kA \frac{dT}{dx}

Heat flow rate = thermal conductivity × area × temperature gradient

For a rod of length L with ends at temperatures $T_1$ and $T_2$ at steady state:

\dot{Q} = kA \frac{T_1 - T_2}{L}

The total heat transferred over time t is simply:

Q = \dot{Q} \cdot t = kA \frac{T_1 - T_2}{L} \cdot t

But for non-steady conditions—where temperatures or geometry vary—we must integrate:

Q = \int_0^t \dot{Q}(\tau) \, d\tau

Real-World Applications

Building Energy Analysis

Engineers calculate heating/cooling loads by integrating heat flow over time. The energy needed to maintain building temperature over a day:

E = \int_0^{24} \dot{Q}(t) \, dt = \int_0^{24} UA(T_{inside} - T_{outside}(t)) \, dt

where $T_{outside}(t)$ varies with time of day.

Thermal Processing in Manufacturing

Heat treatment of metals requires precise energy control. For a part with temperature-dependent specific heat:

Q = \rho V \int_{T_0}^{T_1} c(T) \, dT

Biological Systems

Thermoregulation in organisms, heat dissipation during exercise, and thermal modeling of medical procedures all use these integration techniques to predict temperature changes and energy requirements.

Machine Learning Connection

Thermal physics provides surprisingly deep analogies for optimization and machine learning.

Loss Curvature as Inverse Heat Capacity

In optimization, the Hessian (matrix of second derivatives) describes the curvature of the loss landscape. High curvature means small parameter changes cause large loss changes—like low heat capacity where small heat causes large temperature changes.

Physics	Machine Learning
Heat capacity C	Inverse curvature (Hessian⁻¹)
Q = CΔT	ΔL ≈ Δθᵀ H Δθ
High C → slow heating	Low curvature → slow convergence
Equilibrium T	Optimal parameters θ*

Simulated Annealing

This optimization algorithm is directly inspired by the physical process of annealing metals. The algorithm maintains a "temperature" parameter that controls randomness:

High temperature: Accept many moves, including uphill ones. Explore widely (like molecules at high T).
Cooling schedule: Gradually reduce temperature, following a pattern like $T(t) = T_0 e^{-kt}$ —exactly Newton's Law!
Low temperature: Only accept downhill moves. Converge to local minimum (crystallization).

The Metropolis Criterion

In simulated annealing, a move that increases energy by $\Delta E$ is accepted with probability $e^{-\Delta E / T}$ . This comes from statistical mechanics—the Boltzmann distribution describes thermal fluctuations. Higher temperature means more fluctuations, allowing escape from local minima.

Python Implementation

Heat Energy Calculations

Here's how to calculate thermal energy using Python, including cases with variable specific heat:

Heat Transfer Calculations

🐍heat_transfer.py

Explanation(6)

Code(114)

9Constant Specific Heat

When c is constant, Q = mc(T1-T0) is a simple formula. No integration is needed because we can factor c out of the integral.

20Variable Specific Heat

When specific heat depends on temperature, c(T), we must integrate: Q = m∫c(T)dT. This is common for gases and materials over wide temperature ranges.

35scipy.integrate.quad

The quad function performs adaptive numerical integration. It handles variable integrands automatically, returning both the result and an error estimate.

48Newton's Law Solution

The exponential decay solution comes from separating variables and integrating: ∫dT/(T-T_ambient) = -k∫dt, giving ln|T-T_ambient| = -kt + C.

57Heat Flow Rate

The instantaneous heat flow rate is proportional to the temperature difference. Integrating this rate gives total heat transferred.

76Energy Conservation

The equilibrium temperature is a weighted average, where the weights are the thermal capacities m*c of each body. This is conservation of energy in action.

108 lines without explanation

1import numpy as np
2from scipy import integrate
3import matplotlib.pyplot as plt
4
5def heat_energy_calculations():
6    """
7    Calculate thermal energy using integration.
8    Q = m * ∫[T0 to T1] c(T) dT
9    """
10    # Case 1: Constant specific heat
11    m = 2.0  # kg
12    c_constant = 4186  # J/(kg·K) for water
13    T0, T1 = 20, 80  # °C
14
15    Q_constant = m * c_constant * (T1 - T0)
16    print("Case 1: Constant Specific Heat")
17    print(f"Q = mc(T1 - T0) = {m} × {c_constant} × {T1-T0}")
18    print(f"Q = {Q_constant/1000:.2f} kJ\n")
19
20    # Case 2: Temperature-dependent specific heat
21    # c(T) = a + bT (realistic for some materials)
22    a = 4000  # J/(kg·K)
23    b = 2     # J/(kg·K²)
24
25    def specific_heat(T):
26        return a + b * T
27
28    # Analytical solution: ∫(a + bT)dT = aT + bT²/2
29    Q_analytical = m * (a * (T1 - T0) + (b / 2) * (T1**2 - T0**2))
30
31    # Numerical integration
32    Q_numerical, _ = integrate.quad(
33        lambda T: m * specific_heat(T),
34        T0, T1
35    )
36
37    print("Case 2: Variable Specific Heat c(T) = a + bT")
38    print(f"Analytical: Q = {Q_analytical/1000:.2f} kJ")
39    print(f"Numerical:  Q = {Q_numerical/1000:.2f} kJ\n")
40
41    return Q_constant, Q_analytical
42
43def newtons_law_cooling():
44    """
45    Newton's Law of Cooling: dT/dt = -k(T - T_ambient)
46    Solution: T(t) = T_ambient + (T0 - T_ambient) * e^(-kt)
47
48    Total heat transferred: Q = ∫[0 to t] h*A*(T - T_ambient) dt
49    """
50    T0 = 90          # Initial temperature (°C)
51    T_ambient = 20   # Ambient temperature (°C)
52    k = 0.05         # Cooling constant (1/min)
53    h_A = 10         # h*A product (W/K)
54
55    def temperature(t):
56        return T_ambient + (T0 - T_ambient) * np.exp(-k * t)
57
58    def heat_flow_rate(t):
59        """Rate of heat transfer at time t"""
60        return h_A * (temperature(t) - T_ambient)
61
62    # Total heat transferred from 0 to t
63    t_final = 60  # minutes
64    Q_total, _ = integrate.quad(heat_flow_rate, 0, t_final)
65
66    print("Newton's Law of Cooling")
67    print("=" * 40)
68    print(f"Initial temp: {T0}°C")
69    print(f"Ambient temp: {T_ambient}°C")
70    print(f"Cooling constant: k = {k} min⁻¹")
71    print(f"After {t_final} min: T = {temperature(t_final):.1f}°C")
72    print(f"Total heat transferred: Q = {Q_total:.1f} J")
73
74    # The analytical solution for total heat
75    # Q = ∫ hA(T - T_ambient) dt = hA(T0 - T_ambient)/k * (1 - e^(-kt))
76    Q_analytical = h_A * (T0 - T_ambient) / k * (1 - np.exp(-k * t_final))
77    print(f"Analytical check: Q = {Q_analytical:.1f} J\n")
78
79    return temperature, Q_total
80
81def thermal_equilibrium():
82    """
83    When two bodies exchange heat, equilibrium is reached.
84    Conservation: m1*c1*(T1 - Tf) = m2*c2*(Tf - T2)
85    Solving: Tf = (m1*c1*T1 + m2*c2*T2) / (m1*c1 + m2*c2)
86    """
87    # Body 1: Hot water
88    m1 = 2.0    # kg
89    c1 = 4186   # J/(kg·K)
90    T1 = 80     # °C
91
92    # Body 2: Cold aluminum block
93    m2 = 3.0    # kg
94    c2 = 900    # J/(kg·K)
95    T2 = 20     # °C
96
97    # Equilibrium temperature
98    Tf = (m1 * c1 * T1 + m2 * c2 * T2) / (m1 * c1 + m2 * c2)
99
100    # Heat transferred
101    Q = m1 * c1 * (T1 - Tf)
102
103    print("Thermal Equilibrium")
104    print("=" * 40)
105    print(f"Hot water: {m1} kg at {T1}°C (c = {c1} J/kg·K)")
106    print(f"Cold aluminum: {m2} kg at {T2}°C (c = {c2} J/kg·K)")
107    print(f"Equilibrium temperature: {Tf:.1f}°C")
108    print(f"Heat transferred: {Q/1000:.2f} kJ")
109
110    return Tf, Q
111
112heat_energy_calculations()
113newtons_law_cooling()
114thermal_equilibrium()

Thermal Concepts in Machine Learning

This code demonstrates the deep analogies between thermal physics and optimization algorithms:

Thermal Analogies in ML

🐍thermal_ml_analogy.py

Explanation(5)

Code(111)

8Curvature as Inverse Heat Capacity

In physics, low heat capacity means small heat input causes large temperature changes. Similarly, high loss curvature means small parameter changes cause large loss changes.

28Standard Gradient Descent

Basic gradient descent applies uniform updates regardless of curvature—like heating all materials the same regardless of their heat capacity. This can be inefficient.

38Natural Gradient

Natural gradient divides by the Hessian (curvature), making equal 'effective' steps in all directions. This is like adjusting heat input based on each material's heat capacity.

55Simulated Annealing

This optimization method uses 'temperature' to control randomness. High temperature allows escaping local minima; cooling focuses the search. The exponential cooling schedule mirrors Newton's Law!

80Metropolis Criterion

The acceptance probability exp(-ΔE/T) comes from statistical mechanics. Higher temperature means accepting more uphill moves—thermal fluctuations allow exploration.

106 lines without explanation

1import numpy as np
2
3def heat_capacity_ml_analogy():
4    """
5    In ML optimization, the Hessian (second derivative of loss)
6    acts like an 'inverse heat capacity' for parameter updates.
7
8    Low curvature = high 'heat capacity' = slow temperature change
9    High curvature = low 'heat capacity' = fast temperature change
10    """
11    print("Heat Capacity ↔ Curvature Analogy")
12    print("=" * 45)
13
14    # Simple 2D loss landscape: L(θ) = a*θ₁² + b*θ₂²
15    # where a and b are curvatures (like 1/heat_capacity)
16    a, b = 10.0, 1.0  # Different curvatures in each direction
17
18    # Hessian = [[2a, 0], [0, 2b]]
19    # High curvature in θ₁ direction (like low heat capacity)
20    # Low curvature in θ₂ direction (like high heat capacity)
21
22    def loss(theta):
23        return a * theta[0]**2 + b * theta[1]**2
24
25    def gradient(theta):
26        return np.array([2*a*theta[0], 2*b*theta[1]])
27
28    # Standard gradient descent (ignores curvature)
29    theta = np.array([1.0, 1.0])
30    lr = 0.1
31
32    print("\nStandard Gradient Descent (uniform 'heating'):")
33    for step in range(5):
34        g = gradient(theta)
35        theta = theta - lr * g
36        print(f"  Step {step+1}: θ = [{theta[0]:.4f}, {theta[1]:.4f}], L = {loss(theta):.4f}")
37
38    # Natural gradient (accounts for curvature)
39    # Like adjusting heat input based on heat capacity
40    theta = np.array([1.0, 1.0])
41    print("\nNatural Gradient (curvature-aware 'heating'):")
42
43    for step in range(5):
44        g = gradient(theta)
45        # Scale update by inverse Hessian (accounts for curvature)
46        # This is like Q = C * ΔT, where we adjust ΔT based on C
47        natural_g = np.array([g[0] / (2*a), g[1] / (2*b)])
48        theta = theta - lr * natural_g
49        print(f"  Step {step+1}: θ = [{theta[0]:.4f}, {theta[1]:.4f}], L = {loss(theta):.4f}")
50
51def cooling_schedule_analogy():
52    """
53    Simulated annealing uses temperature as a metaphor.
54    Higher temperature = more random exploration
55    Cooling schedule = gradually reducing randomness
56
57    This is analogous to Newton's Law of Cooling!
58    """
59    print("\n" + "=" * 45)
60    print("Simulated Annealing ↔ Cooling Law Analogy")
61    print("=" * 45)
62
63    T0 = 100.0        # Initial 'temperature'
64    T_final = 0.01    # Final 'temperature'
65    k = 0.1           # Cooling rate
66
67    def temperature(step, schedule='exponential'):
68        if schedule == 'exponential':
69            # T(t) = T0 * e^(-kt) - like Newton's Law!
70            return T0 * np.exp(-k * step)
71        elif schedule == 'linear':
72            return max(T_final, T0 - k * step * T0)
73
74    # Simple optimization with temperature-based acceptance
75    np.random.seed(42)
76
77    def objective(x):
78        return (x - 3)**2 + np.sin(5*x)
79
80    x = 0.0  # Initial guess
81    best_x = x
82    best_val = objective(x)
83
84    print(f"\nOptimizing with exponential cooling:")
85    print(f"{'Step':>4} | {'Temp':>8} | {'x':>8} | {'f(x)':>10}")
86    print("-" * 40)
87
88    for step in range(30):
89        T = temperature(step)
90
91        # Propose random move
92        dx = np.random.normal(0, T * 0.1)
93        new_x = x + dx
94        new_val = objective(new_x)
95
96        # Metropolis acceptance (thermal fluctuations)
97        delta = new_val - objective(x)
98        if delta < 0 or np.random.random() < np.exp(-delta / T):
99            x = new_x
100            if new_val < best_val:
101                best_val = new_val
102                best_x = new_x
103
104        if step % 5 == 0:
105            print(f"{step:>4} | {T:>8.3f} | {x:>8.4f} | {objective(x):>10.4f}")
106
107    print(f"\nBest found: x = {best_x:.4f}, f(x) = {best_val:.4f}")
108    print("(True optimum near x ≈ 3)")
109
110heat_capacity_ml_analogy()
111cooling_schedule_analogy()

Common Mistakes to Avoid

Mistake 1: Ignoring Temperature Dependence

Wrong: Always using Q = mcΔT with constant c.

Correct: When c varies significantly with temperature, use $Q = m \int c(T) \, dT$ .

Mistake 2: Confusing Rate and Total

Wrong: Reporting heat flow rate when total energy is asked.

Correct: Total energy requires integrating the rate: $Q = \int \dot{Q}(t) \, dt$ .

Mistake 3: Wrong Equilibrium Formula

Wrong: Assuming equilibrium is the simple average $T_f = (T_1 + T_2)/2$ .

Correct: Use the weighted formula with mc as weights: $T_f = (m_1 c_1 T_1 + m_2 c_2 T_2) / (m_1 c_1 + m_2 c_2)$ .

Mistake 4: Missing Phase Changes

During phase changes (melting, boiling), temperature stays constant while energy is absorbed as latent heat. The integral must include these discontinuities: $Q = mL$ for phase change plus $\int c(T) \, dT$ for temperature changes.

Test Your Understanding

Question 1 of 6

The heat energy required to raise the temperature of a substance is given by Q = mcΔT. If we have temperature-dependent specific heat c(T), how do we calculate the total heat?

Score: 0/0

Summary

Heat transfer and thermal energy provide beautiful applications of integration in physics. Whether computing energy requirements, solving cooling problems, or finding equilibrium temperatures, calculus is our essential tool.

Key Formulas

Concept	Formula
Heat energy (constant c)	Q = mcΔT
Heat energy (variable c)	Q = m ∫ c(T) dT
Newton's Law of Cooling	T(t) = T_ambient + (T₀ - T_ambient)e^(-kt)
Total heat transferred	Q = ∫ Q̇(t) dt
Equilibrium temperature	T_f = (m₁c₁T₁ + m₂c₂T₂)/(m₁c₁ + m₂c₂)

Key Takeaways

Variable specific heat: When c(T) is not constant, we must integrate to find total heat: $Q = m \int c(T) \, dT$ .
Newton's Law of Cooling: A separable ODE leading to exponential decay toward ambient temperature.
Total heat transfer: Integrate the heat flow rate over time to get cumulative energy transferred.
Equilibrium: Conservation of energy gives a weighted average with thermal capacities as weights.
ML connection: Temperature-based optimization (simulated annealing) and curvature-heat capacity analogies connect physics to learning.

The Core Insight:

"Heat is energy in transit—calculus captures both the instantaneous rate and the accumulated total, connecting differential and integral views of the same physical phenomenon."

Coming Next: In the next section on Economic Surplus, we'll use integration to calculate consumer and producer surplus in economics—another beautiful application of calculus to understanding accumulation and value.