The Global Centre for Risk and Innovation (GCRI)
Plans for infrastructure, reserves and portfolios must work across futures that cannot be predicted. Optimization under uncertainty chooses decisions that perform well on average, hold up in bad cases or stay acceptable across many scenarios. This page sets out the main approaches, from stochastic to robust optimization and decision-making under deep uncertainty, how their results are tested, and what evidence serves the planners who rely on them.
GCRI · In development · service level R0
Optimization under uncertainty is the choice of decisions that perform well across many possible futures instead of one forecast. Stochastic methods optimise expected performance with probabilities, robust methods guard against bad cases, and adaptive plans with triggers keep options open as uncertainty resolves.
For optimization under uncertainty, in GCRI's spatial-intelligence reading, the map opens on true-colour imagery, land cover and rain with relief in 3D and true colour from NASA GIBS, framed on North America; every other NASA layer, live feed and forecast waits in the layer list.
The interactive map needs JavaScript; the layer list and sources are below.
Layers 3
Hazards and exposure, for this page
Keyboard, with the map focused
Hold Ctrl or ⌘ and scroll to zoom
Latest available
NASA GIBS / EOSDIS · ASTER GDEM (METI, NASA) · Blue Marble (NASA) · © OpenStreetMap contributors (ODbL) · Natural Earth
Loading the latest NASA imagery.
Drag to move. Ctrl or ⌘ and scroll, pinch or the buttons to zoom; hover for the values under the pointer; reset returns to the region.
For optimization under uncertainty, in GCRI's spatial-intelligence reading, the map opens on true-colour imagery, land cover and rain with relief in 3D and true colour from NASA GIBS, framed on North America; every other NASA layer, live feed and forecast waits in the layer list.Satellite and model products as NASA publishes them, each with its own date; indicators, not assessments of any place; claim ceiling INTEGRATED.
Layers and dates
The working screen behind this subject: its structure, terms and states. Names are placeholders and values illustrative.
Optimization under uncertainty · Coupled model A · Scenario set 3
nodes: models · edges: exchanged variables
| Hydrology | 1 day | runoff | Yes |
| Power system | 1 h | demand, outages | Yes |
| Economy | 1 quarter | output, prices | Partial |
| Population | 1 year | exposure | Yes |
Robust and adaptive decisions when futures are deeply uncertain. The dossier below sets out how the risk develops, what is measured, where the data come from, how it is modelled and who decides.
A plan optimised for a single forecast can perform badly if the future differs. Representing uncertainty explicitly avoids this trap.
Stochastic programming optimises expected performance across scenarios with assigned probabilities, which needs those probabilities to be credible.
Robust optimization chooses decisions that perform acceptably for all scenarios within a set, trading average performance for protection.
Plans with decision points and triggers adapt as uncertainty resolves, avoiding commitments that lock in poor outcomes.
Each term as this page uses it. Where a standard-setting body defines the term, the wording follows that body.
| Term | Definition |
|---|---|
| Stochastic programming | Optimization that represents uncertain parameters with a set of scenarios and their probabilities. |
| Regret | The difference between a chosen plan's outcome and the best possible outcome in a given scenario. |
| Satisficing | Seeking a solution that meets acceptable thresholds instead of the single best. |
Each indicator carries its unit, its cadence, the kind of source behind it and the evidence label a value would carry. No values are shown: a programme supplies them under its own authority.
| Indicator | Unit | Cadence | Source type | Label |
|---|---|---|---|---|
| Scenarios evaluated | count | per analysis | Analysis documentation | DECLARED |
| Regret of chosen plan | difference from best plan in each scenario | per analysis | Robustness analysis | MODELLED |
| Value of the stochastic solution | difference between stochastic and deterministic plans | per analysis | Optimization outputs | MODELLED |
| Satisficing share | percent of scenarios meeting objectives | per analysis | Robustness analysis | MODELLED |
| Adaptation triggers defined | count | per plan | Plan documentation | DECLARED |
| Out-of-sample performance | objective value on held-out scenarios | per analysis | Testing | MODELLED |
The topic as a computational graph: named sources feed the methods, an evidence review gates the outputs, and the decisions stay with the authorities that hold them.
Fig. 02Nodes carry their type, two metric rows and a state; connectors show what feeds what; dashed frames name the owners; the timeline is the cadence. Design of the pipeline, not a record of a run.
Named public datasets and institutional records, each with its licence class. Sources by agreement are used only under their terms and never republished.
| Source | What it gives | Licence class |
|---|---|---|
| Society for Decision Making under Deep Uncertainty resources | Methods and case studies for robust and adaptive planning | Public, terms to check |
| Open-source optimization tools such as Pyomo and JuMP | Modelling languages for stochastic and robust optimization | Open source, BSD and MPL licences |
| EMA Workbench | Exploratory modelling and analysis toolkit | Open source, BSD licence |
| Operations research literature on stochastic programming | Methods and applications | Licence per publication |
| Published adaptive pathway plans, such as delta programmes | Applied adaptive planning | Public records, licence per government |
Systems Science and Modelling prepares evidence. The decisions below are taken by the authorities named, under their own rules, and never by the platform.
| Who decides | The question | Evidence this page serves |
|---|---|---|
| Infrastructure and resource planners | Which plans hold up across plausible futures | Robustness analyses; pathway designs |
| Analysts across Nexus platforms | Which optimization approach suits their question | Method selection guides |
| Finance and audit bodies | Whether planning analyses considered uncertainty adequately | Analysis documentation |
Classical optimization assumes a known model and known inputs. A strategy optimised for a single forecast can be finely tuned to that forecast and fragile if the future differs. In infrastructure planning, for example, a design optimised for one climate projection may fail under another.
Stochastic optimization handles uncertainty described by probability distributions, seeking the best expected performance or managing risk measures. Robust optimization seeks solutions that perform well across a set of possible conditions without probabilities.
Under deep uncertainty, when experts cannot agree on models or probabilities, methods such as robust decision making and adaptive pathways explore many futures, identify where strategies fail, and design plans that can change course when signposts are observed.
Optimising expected outcomes over scenarios, using optimization and operations research. Requires credible probabilities.
Optimising worst-case or regret across uncertainty sets. Can be conservative.
Running models across many futures to find vulnerabilities, using decision analysis. Computationally demanding.
Sequencing actions with triggers for change. Requires monitoring and institutional capacity.
These methods support judgement, they do not replace it. The choice of objectives and acceptable risks is a value choice for decision-makers.
The current service level is R0: information and enquiries. An enquiry receives a reference and names the next step; applications, payments, compute allocations and registrations open only when their team, terms and review capacity exist.