Chapter map

1. 1. Equation
2. 2. Sources
3. 3. Weights
4. 4. One term
5. 5. Sum
6. 6. Move point
7. 7. Influence matrix
8. 8. Singular case
9. 9. Normals
10. 10. Boundary data
11. 11. Solve
12. 12. Interior evaluation
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. Equation

A small example makes the mechanism visible. Use the 3D Laplace fundamental solution.

A concrete situation

G=1/(4 pi r).

What the mathematics does not license

State units and convention. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Record kernel. 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. Sources

This is where modelling choices begin to affect evidence. Place six points at plus/minus coordinate axes.

A concrete situation

Each is distance one from origin.

What the mathematics does not license

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

Try it, check it, transfer it

Draw geometry. 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. Weights

The central mathematical idea is concrete. Assign q=1/6 to each.

A concrete situation

Total source weight=1.

What the mathematics does not license

Boundary density is illustrative. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Record weights. 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. One term

Here the geometry becomes operational. Evaluate qG at origin.

A concrete situation

Contribution=1/(24 pi).

What the mathematics does not license

Use r=1. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Record term. 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. Sum

This step deserves a careful pause. Add six equal terms.

A concrete situation

Potential=1/(4 pi).

What the mathematics does not license

Approx .0796. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Record value. 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. Move point

The useful question is not only how to calculate, but what the calculation assumes. Evaluate at x=(.5,0,0).

A concrete situation

Distances are no longer equal.

What the mathematics does not license

Recompute all terms. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Record distances. 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. Influence matrix

A small example makes the mechanism visible. Use several targets and sources.

A concrete situation

Every entry generally nonzero.

What the mathematics does not license

Dense coupling. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Record matrix. 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. Singular case

This is where modelling choices begin to affect evidence. Let target approach source.

A concrete situation

Kernel grows singular.

What the mathematics does not license

Use special quadrature. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Record treatment. 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. Normals

The central mathematical idea is concrete. Orient each panel outward.

A concrete situation

Flux signs follow normals.

What the mathematics does not license

Check geometry. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Record convention. 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. Boundary data

Here the geometry becomes operational. Prescribe value or normal derivative.

A concrete situation

Other trace becomes unknown.

What the mathematics does not license

Choose eligible formulation. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Record unknown. 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. Solve

This step deserves a careful pause. Assemble and solve dense system.

A concrete situation

Conditioning matters.

What the mathematics does not license

Report residual. 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.

---

12. Interior evaluation

The useful question is not only how to calculate, but what the calculation assumes. Use solved density in representation formula.

A concrete situation

All panels contribute.

What the mathematics does not license

Avoid source singularity. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Record field. 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

A small example makes the mechanism visible. Refine boundary elements.

A concrete situation

Track target quantity.

What the mathematics does not license

Control solver tolerance. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Record errors. 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

This is where modelling choices begin to affect evidence. State identity,kernel,mesh,quadrature and validation.

A concrete situation

Worked origin potential=1/(4 pi).

What the mathematics does not license

Example is a kernel sum. Keeping this boundary clear makes the result easier to audit and transfer.

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 boundary element method rewrites an eligible boundary-value problem as an integral equation over its boundary. A three-dimensional volume problem can therefore be discretised on a two-dimensional surface, although the resulting matrices are usually dense.

For the three-dimensional Laplace equation, the free-space fundamental solution is G(x,y)=1/[4 pi times the distance between x and y]. It describes the potential created at x by a unit point source at y.

Place six equal source weights q=1/6 at the unit-distance points plus and minus the three coordinate axes. At the origin every source is one unit away.

Each contribution at the origin is (1/6)/(4 pi). Adding six contributions gives the potential 1/(4 pi), approximately 0.0796. The example shows the all-to-all kernel sum that underlies a boundary discretisation.

If the evaluation point moves, all six distances change and every source can contribute. This global coupling explains why a boundary-only mesh can still produce a dense matrix.

Green’s identities connect interior solutions with boundary values and normal derivatives. Depending on the formulation, one of these boundary quantities is prescribed and the other becomes the unknown density.

Did You Know? Reducing the geometric dimension by one saves mesh unknowns, but singular and nearly singular integrals demand careful quadrature. Boundary-only does not mean mathematically effortless.

The method is especially natural for linear homogeneous equations with known fundamental solutions, including Laplace, Helmholtz and elasticity problems. Strong volume nonlinearity or heterogeneous coefficients can remove its main advantage.

Collocation enforces the integral equation at selected boundary points; Galerkin formulations test it against boundary basis functions. The same physical problem can therefore produce different discrete systems.

At a source point, kernels may be weakly singular, strongly singular or hypersingular. Principal values, analytic subtraction and specialised quadrature define the intended operator.

Boundary orientation fixes outward normals and signs. A reversed panel can change flux terms while leaving the geometry visually plausible.

Students can calculate fundamental-solution distances, assemble a small dense influence matrix and check symmetry when the formulation permits it. This makes the global communication visible.

Applications include electrostatics, acoustics, fracture, elasticity and exterior wave scattering. Radiation conditions and resonance behaviour require problem-specific formulations.

Fast multipole or hierarchical matrix acceleration can reduce the cost of far-field interactions. Approximation tolerances then become part of the error budget.

For parents and students, the transferable idea is to focus work on the boundary where a decision is imposed, while remembering that every boundary piece can influence the whole interior.

A responsible report states the PDE, fundamental solution, integral identity, boundary data, element basis, normal convention, singular quadrature, solver, acceleration, convergence and validation.

---

Frequently asked questions

What is the boundary element method?

A numerical method that discretises a boundary integral equation rather than the whole domain.

Why can it reduce dimension?

Known fundamental solutions express interior fields through boundary data.

Why are matrices dense?

Every boundary element can influence every collocation or test point.

What makes singular quadrature necessary?

Fundamental-solution kernels become singular when source and target coincide.

When is BEM less natural?

Strong volume nonlinearity, variable media or unavailable fundamental solutions can remove its advantage.

What must be reported?

PDE, kernel, integral identity, boundary data, mesh, normals, quadrature, solver, convergence and validation.

---

Primary source and fact-check note

M. A. Jaswon, “Integral Equation Methods in Potential Theory. I,” Proceedings of the Royal Society A 275, 1963. 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 element method
• Discontinuous Galerkin 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: representation, transfer and evidence

Identity audit

Show the Green identity and which boundary trace is unknown.

Kernel audit

State dimension, sign, singularity and radiation behaviour.

Geometry audit

Check panel orientation, normals and element Jacobians.

Solver audit

Separate quadrature, discretisation, dense solve and acceleration errors.

Mechanism workshop: calculate, challenge and transfer

Kernel table

Calculate distances and Green-function values for six sources.

Dense matrix

Assemble a four-by-four influence example.

Normal test

Reverse one panel and locate sign changes.

Singular test

Move target toward a source and compare quadrature.

Boundary choice

Swap Dirichlet and Neumann data in an eligible model.

Interior field

Evaluate solved density at several points.

Refinement

Split panels and track a target potential.

Acceleration

Compare direct and compressed matvec tolerances.

Student explanation

Explain why only the boundary is meshed.

Parent explanation

Describe global influence and checking.

Evidence notebook: reconstruct every step

Evidence notebook 1: Equation

Step 1 isolates equation. Use the 3D Laplace fundamental solution. The checked result is: G=1/(4 pi r). Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: State units and convention. Record kernel. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 2: Sources

Step 2 isolates sources. Place six points at plus/minus coordinate axes. The checked result is: Each is distance one from origin. Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Keep ordering. Draw geometry. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 3: Weights

Step 3 isolates weights. Assign q=1/6 to each. The checked result is: Total source weight=1. Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Boundary density is illustrative. Record weights. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 4: One term

Step 4 isolates one term. Evaluate qG at origin. The checked result is: Contribution=1/(24 pi). Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Use r=1. Record term. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 5: Sum

Step 5 isolates sum. Add six equal terms. The checked result is: Potential=1/(4 pi). Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Approx .0796. Record value. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 6: Move point

Step 6 isolates move point. Evaluate at x=(.5,0,0). The checked result is: Distances are no longer equal. Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Recompute all terms. Record distances. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 7: Influence matrix

Step 7 isolates influence matrix. Use several targets and sources. The checked result is: Every entry generally nonzero. Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Dense coupling. Record matrix. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 8: Singular case

Step 8 isolates singular case. Let target approach source. The checked result is: Kernel grows singular. Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Use special quadrature. Record treatment. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 9: Normals

Step 9 isolates normals. Orient each panel outward. The checked result is: Flux signs follow normals. Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Check geometry. Record convention. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 10: Boundary data

Step 10 isolates boundary data. Prescribe value or normal derivative. The checked result is: Other trace becomes unknown. Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Choose eligible formulation. Record unknown. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 11: Solve

Step 11 isolates solve. Assemble and solve dense system. The checked result is: Conditioning matters. Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Report residual. Record result. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 12: Interior evaluation

Step 12 isolates interior evaluation. Use solved density in representation formula. The checked result is: All panels contribute. Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Avoid source singularity. Record field. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 13: Convergence

Step 13 isolates convergence. Refine boundary elements. The checked result is: Track target quantity. Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Control solver tolerance. Record errors. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Evidence notebook 14: Conclusion

Step 14 isolates conclusion. State identity,kernel,mesh,quadrature and validation. The checked result is: Worked origin potential=1/(4 pi). Keep dimensions, kernel or lattice conventions, node and direction order, normals, sign, time level and precision beside every number. The safeguard is: Example is a kernel sum. Write conclusion. Change one input, recompute the local rule and global invariant, and explain whether the difference comes from quadrature, collision, streaming, interface transport, force transfer, spatial resolution, time integration, rounding or a changed model. Ask a second reader to reproduce the arithmetic and finish with the narrow claim the evidence supports.

Failure cases and responsible recovery

Boundary-only-is-sparse

Global kernels usually create dense matrices. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Singularity ignored

Ordinary quadrature can be wrong. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Normal reversed

Flux signs change. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Wrong fundamental solution

The integral identity no longer matches the PDE. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Exterior resonance

Formulation choice matters. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Volume source omitted

Domain terms may remain. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Residual-only trust

Field errors need convergence checks. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Fast approximation hidden

Acceleration tolerance belongs in the budget. Build a small counterexample, restore the missing condition and recalculate before making the claim.

Verified case study: fourteen steps from model to decision

1. Define the model

Write the PDE or kinetic rule, variables, units, domain and boundary or initial data.

2. Identify the representation

Name boundary densities, lattice populations, a level-set field or Lagrangian markers.

3. State the transfer rule

Write the kernel, collision-streaming map, Hamilton-Jacobi speed or delta coupling.

4. Calculate a tiny case

Use the smallest nontrivial example with labelled indices.

5. Check a conserved quantity

Verify source weight, mass, volume, force or work as eligible.

6. Audit signs and geometry

Check normals, directions, level-set sign, marker coordinates and Jacobians.

7. Test a singular or sharp feature

Approach a kernel source, interface, wall or concentrated force.

8. Refine

Improve panels, lattice, interface grid and marker spacing consistently.

9. Separate errors

Distinguish modelling, discretisation, quadrature, iteration and rounding.

10. Compare a baseline

Use an exact, manufactured, direct or independently implemented reference.

11. Count cost

Record unknowns, nonzeros or dense entries, cell updates, memory and elapsed time.

12. Validate

Compare application outputs with independent measurements 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 arithmetic.

Learner lab: ten checks for transferable understanding

Lab question 1

Create a reproducible mini-investigation of Green kernels, integral identities, panels, normals, singular quadrature and dense solves. For task 1, state one prediction, calculate 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 Green kernels, integral identities, panels, normals, singular quadrature and dense solves. For task 2, state one prediction, calculate 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 Green kernels, integral identities, panels, normals, singular quadrature and dense solves. For task 3, state one prediction, calculate 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 Green kernels, integral identities, panels, normals, singular quadrature and dense solves. For task 4, state one prediction, calculate 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 Green kernels, integral identities, panels, normals, singular quadrature and dense solves. For task 5, state one prediction, calculate 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 Green kernels, integral identities, panels, normals, singular quadrature and dense solves. For task 6, state one prediction, calculate 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 Green kernels, integral identities, panels, normals, singular quadrature and dense solves. For task 7, state one prediction, calculate 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 Green kernels, integral identities, panels, normals, singular quadrature and dense solves. For task 8, state one prediction, calculate 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 Green kernels, integral identities, panels, normals, singular quadrature and dense solves. For task 9, state one prediction, calculate 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 Green kernels, integral identities, panels, normals, singular quadrature and dense solves. For task 10, state one prediction, calculate 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.