Chapter map

1. 1. PDE
2. 2. Grid
3. 3. Scale
4. 4. Matrix
5. 5. Initial state
6. 6. Derivative
7. 7. Euler step
8. 8. Ratio
9. 9. Stability
10. 10. Mass check
11. 11. Refine space
12. 12. Implicit comparison
13. 13. Convergence
14. 14. Conclusion

Visit the How Mathematics Works hub for the wider series and its connections to education, reasoning, science, computing, design and careers.

---

1. PDE

This step deserves a careful pause. Use u_t=u_xx on [0,1].

A concrete situation

Zero boundary values.

What the mathematics does not license

Record units and domain. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Write equation. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

2. Grid

The useful question is not only how to calculate, but what the calculation assumes. Use interior nodes 1/4,1/2,3/4.

A concrete situation

dx=1/4.

What the mathematics does not license

Do not count boundaries as unknowns. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Draw grid. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

3. Scale

A small example makes the mechanism visible. Compute 1/dx^2.

A concrete situation

Value=16.

What the mathematics does not license

Keep squared units. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Record scale. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

4. Matrix

This is where modelling choices begin to affect evidence. Assemble tridiagonal second difference.

A concrete situation

Diagonal -2, neighbours 1.

What the mathematics does not license

Insert boundaries correctly. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Write operator. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

5. Initial state

The central mathematical idea is concrete. Set U=(0,1,0).

A concrete situation

Peak at the middle.

What the mathematics does not license

Keep node order. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Record vector. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

6. Derivative

Here the geometry becomes operational. Apply the matrix times 16.

A concrete situation

U prime=(16,-32,16).

What the mathematics does not license

Check signs. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Record result. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

7. Euler step

This step deserves a careful pause. Use dt=.01.

A concrete situation

U1=(.16,.68,.16).

What the mathematics does not license

This is one time method. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Record step. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

8. Ratio

The useful question is not only how to calculate, but what the calculation assumes. Compute dt/dx^2.

A concrete situation

r=.16.

What the mathematics does not license

Compare only with eligible scheme. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Record r. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

9. Stability

A small example makes the mechanism visible. Check r<=.5.

A concrete situation

Worked explicit step satisfies it.

What the mathematics does not license

Not a universal limit. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

State scope. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

10. Mass check

This is where modelling choices begin to affect evidence. Sum interior derivative.

A concrete situation

Sum=0 here.

What the mathematics does not license

Boundary flux may change other states. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Record balance. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

11. Refine space

The central mathematical idea is concrete. Set dx=1/8.

A concrete situation

Fast eigenvalues grow in magnitude.

What the mathematics does not license

Retune dt. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Record change. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

12. Implicit comparison

Here the geometry becomes operational. Use backward Euler conceptually.

A concrete situation

Solve a linear system each step.

What the mathematics does not license

Stability is not free cost. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Record trade-off. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

13. Convergence

This step deserves a careful pause. Vary dx and dt separately.

A concrete situation

Identify dominant error.

What the mathematics does not license

Avoid coupled hiding. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Record table. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

14. Conclusion

The useful question is not only how to calculate, but what the calculation assumes. State semi-discrete operator,solver,tolerances and checks.

A concrete situation

Worked state becomes (.16,.68,.16).

What the mathematics does not license

No physical guarantee. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Write conclusion. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

A seven-day practice lab

1. Day 1: define the decision, observation unit, X variables and Y or reference condition.
2. Day 2: reproduce the smallest worked example and check every dimension, unit and sign.
3. Day 3: build a transparent baseline and write what it assumes.
4. Day 4: fit the main method inside a training-only pipeline; record every parameter.
5. Day 5: challenge one assumption with a deliberately difficult dataset or operating regime.
6. Day 6: compare validation metrics with the practical workload created by errors.
7. Day 7: present one supported conclusion, one unresolved uncertainty and one safe next experiment.

Questions for a student, parent or mentor

• Which part of the model is a mathematical definition, and which part is a choice?
• What small calculation would catch a sign, scale or dimension error?
• What evidence comes from training, validation and the untouched test?
• What benign condition could look like a fault or prediction failure?
• How would the answer change for another response, asset, school project or time period?
• What should a human review before acting on the score?

A method-specific audit before deployment

The method of lines solves a time-dependent partial differential equation by discretising space first and leaving time continuous. The result is a linked system of ordinary differential equations that can be advanced by a suitable ODE solver.

For the heat equation u_t=u_xx on 0<x<1 with zero boundary values, choose three interior nodes at x=1/4, 1/2 and 3/4. The spacing is dx=1/4, so 1/dx squared equals 16.

A centred second difference converts the spatial operator into U prime=16 times the tridiagonal matrix with -2 on the diagonal and 1 beside it, applied to the vector of interior temperatures.

With U(0)=(0,1,0), the initial derivative is (16,-32,16). Heat leaves the centre and enters its neighbours; the signs and the zero sum across the interior-plus-boundary balance match diffusion.

One forward Euler step of dt=0.01 gives (0.16,0.68,0.16). The step is a time-integration choice layered on top of the spatial method, not part of the method-of-lines definition itself.

The diffusion ratio r=dt/dx squared is 0.16. For this standard explicit heat discretisation, r no greater than 1/2 is the familiar stability restriction; an implicit solver changes the restriction and the linear algebra cost.

Did You Know? A spatial discretisation can turn a modest PDE into thousands or millions of coupled ODEs. Sparsity, stiffness and boundary treatment then determine which time solver is practical.

Boundary conditions enter the semi-discrete equations. A prescribed temperature changes known terms, while a prescribed flux changes the boundary stencil or the weak form. Silently swapping them changes the physical problem.

Stiffness appears when the fastest spatial modes decay much more quickly than the slow modes. Refining dx for diffusion pushes eigenvalues left roughly like 1/dx squared, so explicit time steps may become very small.

The spatial and temporal errors must be separated. A high-order time integrator cannot repair a coarse spatial grid, while a fine spatial grid can be wasted if the time step is too large.

Students can assemble the three-by-three matrix, calculate its action on several profiles and check whether peaks flatten. This makes the PDE-to-ODE translation visible before software hides it.

Conservation, positivity and monotonicity depend on the equation, spatial operator, time integrator and step size together. No single label guarantees all three properties.

Applications include heat transfer, reaction-diffusion, wave propagation, fluid flow and population models. A mathematically stable computation still needs credible coefficients, initial data and boundary measurements.

A convergence study refines space and time independently. Holding dt proportional to dx squared can hide which error source dominates, so additional runs should vary one scale at a time.

For parents and students, the transferable idea is decomposition: preserve one variable as continuous, approximate the others carefully, then use the best established solver for the resulting system.

A responsible report states the PDE, domain, spatial nodes or elements, boundary conditions, semi-discrete operator, ODE solver, tolerances, stiffness evidence, stability limits, convergence results and validation.

---

Frequently asked questions

What is the method of lines?

A PDE strategy that discretises selected independent variables, usually space, while keeping another, usually time, continuous.

Why does it create ODEs?

Spatial derivative formulas turn values at mesh points or elements into coupled algebraic expressions.

What is stiffness?

Widely separated time scales that can make explicit solvers impractically restrictive.

Does a stable ODE solver guarantee a correct PDE solution?

No. Spatial error, boundaries, modelling and validation still matter.

How should convergence be checked?

Refine space and time independently and compare with an exact, manufactured or benchmark solution.

What must be reported?

PDE, domain, spatial operator, boundaries, solver, tolerances, stability evidence, convergence and validation.

---

Primary source and fact-check note

John C. Strikwerda, “Initial Boundary Value Problems for the Method of Lines,” Journal of Computational Physics 34, 1980. Definitions and core claims in this article were checked against this primary publication. Calculations shown here were independently recomputed. Examples are educational illustrations; they are not reported industrial performance results.

Useful next reading

• Finite difference method
• Finite element method
• Numerical analysis hub
• How Mathematics Works
• Study and learning methods

Keep the options open. Mathematics education supports clearer questions, transferable problem-solving skills and informed choices across science, engineering, computing, AI, statistics, design and many other careers. It does not guarantee intelligence, income, admission or employment; those outcomes depend on far more than one subject or model.

Method audit: equations, structure and evidence

Translation audit

Show exactly which spatial derivatives became matrix or residual terms.

Boundary audit

Identify every known boundary contribution in the semi-discrete equations.

Stiffness audit

Estimate eigenvalue scales and explain the ODE solver choice.

Error audit

Vary space, time and solver tolerance independently.

Mechanism workshop: calculate, challenge and transfer

Matrix build

Derive a three-node Laplacian from Taylor differences.

Mode test

Apply the matrix to constant, peaked and alternating profiles.

Solver comparison

Advance one step with explicit and implicit Euler.

Stiffness experiment

Halve dx and track the explicit step restriction.

Boundary swap

Replace a value boundary with a flux boundary and rebuild the affected row.

Tolerance study

Tighten ODE tolerances while holding the mesh fixed.

Space study

Refine the mesh while holding time error negligible.

Manufactured solution

Insert a known u(x,t) and derive its forcing.

Student explanation

Explain why a PDE became coupled ODEs.

Parent explanation

Describe why simulation still needs real data.

Evidence notebook: reconstruct every step

Evidence notebook 1: PDE

Step 1 isolates pde. Use u_t=u_xx on [0,1]. The checked result is: Zero boundary values. Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Record units and domain. Write equation. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 2: Grid

Step 2 isolates grid. Use interior nodes 1/4,1/2,3/4. The checked result is: dx=1/4. Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Do not count boundaries as unknowns. Draw grid. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 3: Scale

Step 3 isolates scale. Compute 1/dx^2. The checked result is: Value=16. Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Keep squared units. Record scale. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 4: Matrix

Step 4 isolates matrix. Assemble tridiagonal second difference. The checked result is: Diagonal -2, neighbours 1. Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Insert boundaries correctly. Write operator. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 5: Initial state

Step 5 isolates initial state. Set U=(0,1,0). The checked result is: Peak at the middle. Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Keep node order. Record vector. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 6: Derivative

Step 6 isolates derivative. Apply the matrix times 16. The checked result is: U prime=(16,-32,16). Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Check signs. Record result. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 7: Euler step

Step 7 isolates euler step. Use dt=.01. The checked result is: U1=(.16,.68,.16). Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: This is one time method. Record step. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 8: Ratio

Step 8 isolates ratio. Compute dt/dx^2. The checked result is: r=.16. Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Compare only with eligible scheme. Record r. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 9: Stability

Step 9 isolates stability. Check r<=.5. The checked result is: Worked explicit step satisfies it. Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Not a universal limit. State scope. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 10: Mass check

Step 10 isolates mass check. Sum interior derivative. The checked result is: Sum=0 here. Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Boundary flux may change other states. Record balance. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 11: Refine space

Step 11 isolates refine space. Set dx=1/8. The checked result is: Fast eigenvalues grow in magnitude. Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Retune dt. Record change. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 12: Implicit comparison

Step 12 isolates implicit comparison. Use backward Euler conceptually. The checked result is: Solve a linear system each step. Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Stability is not free cost. Record trade-off. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 13: Convergence

Step 13 isolates convergence. Vary dx and dt separately. The checked result is: Identify dominant error. Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: Avoid coupled hiding. Record table. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Evidence notebook 14: Conclusion

Step 14 isolates conclusion. State semi-discrete operator,solver,tolerances and checks. The checked result is: Worked state becomes (.16,.68,.16). Keep the equation, units, node or element order, face orientation, basis convention, time level and precision beside every value. The safeguard is: No physical guarantee. Write conclusion. Recalculate after changing exactly one parameter and explain whether the change arises from spatial approximation, time integration, geometry preservation, refinement logic, numerical flux, rounding or a different physical model. Ask a second reader to reproduce the arithmetic and finish with the narrowest claim the evidence supports.

Failure cases and responsible recovery

Space-only confidence

Time integration can still dominate error. Build a small counterexample, restore the missing condition and recalculate before making the claim.

ODE-only confidence

A good ODE solve cannot repair a bad spatial operator. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Boundary omission

The semi-discrete system changes with boundary data. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Stiffness ignored

Explicit cost can explode after refinement. Build a small counterexample, restore the missing condition and recalculate before making the claim.

One coupled refinement

It can hide which error dominates. Build a small counterexample, restore the missing condition and recalculate before making the claim.

CFL copied blindly

Limits depend on equation and scheme. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Conservation assumed

The full discretisation decides it. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Validation skipped

Numerical convergence is not physical validity. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Verified case study: fourteen steps from equation to decision

1. Define the continuous model

Write the equation, variables, units, domain and initial or boundary data.

2. Identify preserved structure

Name conservation, dissipation, symplectic geometry, smoothness or localisation that should guide the discretisation.

3. Choose representation

Justify nodes, elements, local polynomials, nested grids or phase variables.

4. Derive one local rule

Connect a matrix row, canonical substep, refinement decision or face flux to the governing mathematics.

5. Calculate by hand

Use the smallest nontrivial example and keep exact values where practical.

6. Check signs and units

Audit orientation, dimensions, time levels and indexing.

7. Test an invariant

Measure mass, phase area, energy behaviour, boundary values or another eligible diagnostic.

8. Trigger a known failure

Use a bad time step, escaped feature, wrong trace or broken interface transfer.

9. Refine systematically

Vary one of space, time, polynomial degree or threshold while controlling the others.

10. Compare a baseline

Use an exact, manufactured, uniform-grid or independently implemented reference.

11. Count cost

Record degrees of freedom, cell-updates, matrix solves, memory and elapsed time.

12. Validate the model

Compare the application output with independent observations or accepted benchmarks.

13. Communicate limits

State what the result supports and what remains untested.

14. Preserve evidence

Save equations, data, code, meshes, parameters, logs and corrected calculations.

Learner lab: ten checks for transferable understanding

Lab question 1

Create a reproducible mini-investigation of semi-discrete operator, boundary rows, eigenvalue scales and space-time convergence. For task 1, state one prediction, calculate or simulate a transparent case, record units and conventions, compare an independent baseline, identify one failure mode, and explain the result once for a student and once for a parent. End with one practical next step and one claim the evidence does not support.

Lab question 2

Create a reproducible mini-investigation of semi-discrete operator, boundary rows, eigenvalue scales and space-time convergence. For task 2, state one prediction, calculate or simulate a transparent case, record units and conventions, compare an independent baseline, identify one failure mode, and explain the result once for a student and once for a parent. End with one practical next step and one claim the evidence does not support.

Lab question 3

Create a reproducible mini-investigation of semi-discrete operator, boundary rows, eigenvalue scales and space-time convergence. For task 3, state one prediction, calculate or simulate a transparent case, record units and conventions, compare an independent baseline, identify one failure mode, and explain the result once for a student and once for a parent. End with one practical next step and one claim the evidence does not support.

Lab question 4

Create a reproducible mini-investigation of semi-discrete operator, boundary rows, eigenvalue scales and space-time convergence. For task 4, state one prediction, calculate or simulate a transparent case, record units and conventions, compare an independent baseline, identify one failure mode, and explain the result once for a student and once for a parent. End with one practical next step and one claim the evidence does not support.

Lab question 5

Create a reproducible mini-investigation of semi-discrete operator, boundary rows, eigenvalue scales and space-time convergence. For task 5, state one prediction, calculate or simulate a transparent case, record units and conventions, compare an independent baseline, identify one failure mode, and explain the result once for a student and once for a parent. End with one practical next step and one claim the evidence does not support.

Lab question 6

Create a reproducible mini-investigation of semi-discrete operator, boundary rows, eigenvalue scales and space-time convergence. For task 6, state one prediction, calculate or simulate a transparent case, record units and conventions, compare an independent baseline, identify one failure mode, and explain the result once for a student and once for a parent. End with one practical next step and one claim the evidence does not support.

Lab question 7

Create a reproducible mini-investigation of semi-discrete operator, boundary rows, eigenvalue scales and space-time convergence. For task 7, state one prediction, calculate or simulate a transparent case, record units and conventions, compare an independent baseline, identify one failure mode, and explain the result once for a student and once for a parent. End with one practical next step and one claim the evidence does not support.

Lab question 8

Create a reproducible mini-investigation of semi-discrete operator, boundary rows, eigenvalue scales and space-time convergence. For task 8, state one prediction, calculate or simulate a transparent case, record units and conventions, compare an independent baseline, identify one failure mode, and explain the result once for a student and once for a parent. End with one practical next step and one claim the evidence does not support.

Lab question 9

Create a reproducible mini-investigation of semi-discrete operator, boundary rows, eigenvalue scales and space-time convergence. For task 9, state one prediction, calculate or simulate a transparent case, record units and conventions, compare an independent baseline, identify one failure mode, and explain the result once for a student and once for a parent. End with one practical next step and one claim the evidence does not support.

Lab question 10

Create a reproducible mini-investigation of semi-discrete operator, boundary rows, eigenvalue scales and space-time convergence. For task 10, state one prediction, calculate or simulate a transparent case, record units and conventions, compare an independent baseline, identify one failure mode, and explain the result once for a student and once for a parent. End with one practical next step and one claim the evidence does not support.