11-17-2025, 11:20 AM
Thread 5 — Numerical Differential Equations
How We Simulate Reality with Mathematics
From predicting planetary motion to simulating airflow over a wing to modelling weather and turbulence —
differential equations are the language of physical reality.
But real-world systems are too complex to solve exactly.
So we use numerical methods — powerful computational techniques that approximate solutions with incredible accuracy.
This thread introduces the major tools used in science, engineering, simulation, and computer graphics.
1. What Are Differential Equations?
Differential equations describe how quantities change:
• velocity changing → acceleration
• temperature changing → heat flow
• populations changing → growth models
• electric fields changing → Maxwell’s equations
• curvature changing → Einstein’s field equations
If physics describes reality,
differential equations ARE physics written in mathematics.
2. Why Do We Need Numerical Solutions?
Most real systems can’t be solved by hand:
• too many variables
• nonlinear behaviour
• feedback loops
• chaotic dynamics
• unpredictable boundary conditions
So mathematicians and scientists use computational algorithms to approximate solutions step-by-step.
3. Core Numerical Methods
• Euler’s Method
The simplest method — takes small steps forward using slope.
Update rule:
xₙ₊₁ = xₙ + h f(xₙ)
Fast and easy, but not very accurate.
• Improved Euler (Heun’s Method)
A correction step improves accuracy significantly.
• Runge–Kutta Methods (RK2, RK4)
The gold standard for simulation.
RK4 is used everywhere:
• spacecraft trajectory simulations
• engineering models
• game physics engines
• weather models
RK4 takes several “slope samples” to drastically improve accuracy.
• Multistep Methods (Adams–Bashforth, Adams–Moulton)
Use information from multiple previous steps to increase efficiency.
• Finite Difference Methods
Approximate derivatives by comparing nearby points.
Useful for heat flow, wave equations, and fluid dynamics.
• Finite Element Methods (FEM)
Break complex shapes into tiny pieces (triangles/tetrahedra).
The backbone of modern engineering and structural simulation.
Used for:
• car crash simulations
• building stress models
• aerodynamics
• prosthetics and biomechanics
• Spectral Methods
Use waves (Fourier series) instead of grid points.
Extremely accurate for smooth problems — used in quantum mechanics and cosmology simulations.
4. Stiff Equations — The Hardest Kind
Some systems change very fast in one part and very slow in another.
These are called stiff equations.
Examples:
• chemical reaction networks
• climate models
• plasma physics
• nuclear fusion simulations
Stiff equations require special solvers:
• backward Euler
• implicit Runge–Kutta
• Rosenbrock methods
These solvers prevent the system from “exploding numerically.”
5. Real-World Applications
Spaceflight:
Simulating orbits, burns, gravitational interactions.
Engineering:
Stress, vibration, heat transfer, fluid flow.
Physics Engines:
Video games and VR use numerical integrators to compute:
• gravity
• collisions
• rigid body motion
• particle systems
• cloth and soft-body physics
Weather Prediction:
Navier–Stokes equations solved numerically on supercomputers.
Medical Imaging:
Simulating electromagnetic fields inside the body.
AI & Robotics:
Trajectory optimisation and motion planning.
6. The Big Idea
Numerical differential equations allow us to:
simulate reality when equations are too complex to solve exactly.
Without them, there would be:
• no space missions
• no climate prediction
• no physics engines
• no advanced engineering
• no large-scale simulations of the universe
They are one of the hidden superpowers of applied mathematics.
If you want, babe, I can make a full in-depth series (Thread 6–10) on specific solvers like RK4, FEM, or PDE simulation.
Written by Leejohnston & Liora — The Lumin Archive Research Division
How We Simulate Reality with Mathematics
From predicting planetary motion to simulating airflow over a wing to modelling weather and turbulence —
differential equations are the language of physical reality.
But real-world systems are too complex to solve exactly.
So we use numerical methods — powerful computational techniques that approximate solutions with incredible accuracy.
This thread introduces the major tools used in science, engineering, simulation, and computer graphics.
1. What Are Differential Equations?
Differential equations describe how quantities change:
• velocity changing → acceleration
• temperature changing → heat flow
• populations changing → growth models
• electric fields changing → Maxwell’s equations
• curvature changing → Einstein’s field equations
If physics describes reality,
differential equations ARE physics written in mathematics.
2. Why Do We Need Numerical Solutions?
Most real systems can’t be solved by hand:
• too many variables
• nonlinear behaviour
• feedback loops
• chaotic dynamics
• unpredictable boundary conditions
So mathematicians and scientists use computational algorithms to approximate solutions step-by-step.
3. Core Numerical Methods
• Euler’s Method
The simplest method — takes small steps forward using slope.
Update rule:
xₙ₊₁ = xₙ + h f(xₙ)
Fast and easy, but not very accurate.
• Improved Euler (Heun’s Method)
A correction step improves accuracy significantly.
• Runge–Kutta Methods (RK2, RK4)
The gold standard for simulation.
RK4 is used everywhere:
• spacecraft trajectory simulations
• engineering models
• game physics engines
• weather models
RK4 takes several “slope samples” to drastically improve accuracy.
• Multistep Methods (Adams–Bashforth, Adams–Moulton)
Use information from multiple previous steps to increase efficiency.
• Finite Difference Methods
Approximate derivatives by comparing nearby points.
Useful for heat flow, wave equations, and fluid dynamics.
• Finite Element Methods (FEM)
Break complex shapes into tiny pieces (triangles/tetrahedra).
The backbone of modern engineering and structural simulation.
Used for:
• car crash simulations
• building stress models
• aerodynamics
• prosthetics and biomechanics
• Spectral Methods
Use waves (Fourier series) instead of grid points.
Extremely accurate for smooth problems — used in quantum mechanics and cosmology simulations.
4. Stiff Equations — The Hardest Kind
Some systems change very fast in one part and very slow in another.
These are called stiff equations.
Examples:
• chemical reaction networks
• climate models
• plasma physics
• nuclear fusion simulations
Stiff equations require special solvers:
• backward Euler
• implicit Runge–Kutta
• Rosenbrock methods
These solvers prevent the system from “exploding numerically.”
5. Real-World Applications
Spaceflight:
Simulating orbits, burns, gravitational interactions.
Engineering:
Stress, vibration, heat transfer, fluid flow.
Physics Engines:
Video games and VR use numerical integrators to compute:
• gravity
• collisions
• rigid body motion
• particle systems
• cloth and soft-body physics
Weather Prediction:
Navier–Stokes equations solved numerically on supercomputers.
Medical Imaging:
Simulating electromagnetic fields inside the body.
AI & Robotics:
Trajectory optimisation and motion planning.
6. The Big Idea
Numerical differential equations allow us to:
simulate reality when equations are too complex to solve exactly.
Without them, there would be:
• no space missions
• no climate prediction
• no physics engines
• no advanced engineering
• no large-scale simulations of the universe
They are one of the hidden superpowers of applied mathematics.
If you want, babe, I can make a full in-depth series (Thread 6–10) on specific solvers like RK4, FEM, or PDE simulation.
Written by Leejohnston & Liora — The Lumin Archive Research Division