Back to app index

Galois Adjoints as a Requirement ↔ Commitment “Translator” (Rail Upgrade Example)

Toy scenario

We model two posets: Requirements (capacity tiers) and Commitments (platform/signalling/power/approvals + budget + schedule). The left adjoint f maps a requirement to its minimum package in the supplied table. The right adjoint g maps your current commitments to the strongest tier supported by that table. These rail thresholds, costs and durations illustrate the mathematics; they do not establish engineering feasibility or safety.

1) Choose a requirement (what stakeholders ask for)

Left adjoint f(requirement)

“Minimum commitment package assigned to this tier by the toy table.”

Closure g(f(requirement))

“Which tier does the table support with this minimum package?”
All five tiers stay unchanged: their packages are distinct and nested, so g(f(t)) = t.
Why this helps a project director
If a requirement changes by one notch, f gives a monotone (non-decreasing) change in every commitment dimension. That makes the assumed scope relationships legible: you can show which subsystems, approvals, and budgets the table says must move together. Try moving from tier 2 to tier 3: signalling and power both jump from stage 2 to stage 4 in f. This jump is in the package table; the closure g(f(t)) is the identity here and adds no further tier uplift.

2) Choose commitments (what you’re actually funding/scheduling/procuring)

What you’re seeing (the adjunction law)

3) Director view: a single coherent “so what?”

Strongest tier supported by the table under your commitments:
Interior f(g(commitment)) strips modelled slack: “smallest commitment that still supports your current tier in the table.”
Possible governance uses, after validating project-specific packages:
  • Change control: when a tier changes, show Δf as the “must move” list across modules.
  • Stage-gates: compare commitments with f(tier) alongside the delivery and safety evidence needed to promise that tier.
  • Negotiation: if commitments are capped, g(commitment) identifies the tier supported by the agreed table.
  • Co-design alignment: monotonicity keeps the modelled package dimensions from decreasing as requirements rise.
Note: if time-vs-money trade-offs create multiple incomparable “minimum” options, you won’t get a single weakest package; you’ll get a Pareto set (antichain). That’s exactly where monotone co-design uses set-valued solutions rather than single points.