Why Science? | Institute for Mathematical Sciences

Mathematics begins by specifying objects, relations and operations precisely. Good definitions capture a useful structure without hiding assumptions.

Mathematical research makes assumptions and definitions explicit before reasoning from them. This discipline lets a result travel across applications without losing its conditions. The practical capability is mathematical-definition literacy. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Definitions create the objects of study” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: Changing a definition can change the theorem.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Next chapter

A common pattern can appear in numbers, shapes, algorithms or physical systems. Abstraction removes irrelevant detail while keeping relationships needed for reasoning.

A proof explains why a statement follows in every permitted case; computation can reveal patterns and test examples but does not automatically replace the argument. The practical capability is structural-abstraction reasoning. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Abstraction preserves structure across contexts” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: Abstract similarity does not make applications identical.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

Patterns in examples can suggest a general statement. Researchers search for proofs, boundary cases and counterexamples before accepting it.

Models simplify reality deliberately. Their value comes from transparent choices, sensitivity checks and comparisons with data, not from complexity alone. The practical capability is conjecture-testing literacy. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “A conjecture is an invitation to test” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: Many confirming examples do not prove a universal claim.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

A proof links assumptions to a conclusion through valid logical steps. It should be inspectable by other mathematicians and explicit about established results used.

Mathematical collaboration often begins when different fields discover the same structure in networks, uncertainty, optimisation, geometry or change over time. The practical capability is proof reasoning. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Proof explains why a result must hold” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: A persuasive diagram is not always a complete proof.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

One permitted case that violates a universal claim disproves it. Counterexamples reveal which hypothesis or boundary needs revision.

Mathematical research makes assumptions and definitions explicit before reasoning from them. This discipline lets a result travel across applications without losing its conditions. The practical capability is counterexample design. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Counterexamples sharpen statements” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: A counterexample must satisfy the original assumptions.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

Numerical experiments and symbolic tools can test cases, discover patterns and check algebra. Rounding, finite ranges and software errors remain visible.

A proof explains why a statement follows in every permitted case; computation can reveal patterns and test examples but does not automatically replace the argument. The practical capability is computational-mathematics literacy. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Computation explores without silently becoming proof” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: Finite computation cannot prove every infinite case without an argument.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

A probability space defines possible outcomes and their weights. Independence, conditioning and distributions turn uncertain events into quantities that can be reasoned about.

Models simplify reality deliberately. Their value comes from transparent choices, sensitivity checks and comparisons with data, not from complexity alone. The practical capability is probability reasoning. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Probability models uncertainty” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: A probability model is not the same as observed frequency in every short run.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

Estimation and testing use data to learn about broader quantities. Sampling design, model assumptions, effect size and uncertainty determine interpretation.

Mathematical collaboration often begins when different fields discover the same structure in networks, uncertainty, optimisation, geometry or change over time. The practical capability is statistical-inference literacy. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Statistics connects samples with populations” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: Statistical significance does not measure practical importance by itself.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

An objective function and feasible set define what “best” means. Convexity, algorithms, trade-offs and sensitivity determine whether solutions are unique and useful.

Mathematical research makes assumptions and definitions explicit before reasoning from them. This discipline lets a result travel across applications without losing its conditions. The practical capability is optimisation reasoning. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Optimisation searches within constraints” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: An optimum for one objective may be poor for another.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

Computers approximate continuous quantities and equations using finite operations. Stability, convergence and conditioning show how errors behave.

A proof explains why a statement follows in every permitted case; computation can reveal patterns and test examples but does not automatically replace the argument. The practical capability is numerical-analysis literacy. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Numerical analysis tracks approximation error” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: More digits do not guarantee a more accurate model.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

Difference or differential equations describe evolving states. Fixed points, cycles, sensitivity and bifurcations reveal possible behaviours.

Models simplify reality deliberately. Their value comes from transparent choices, sensitivity checks and comparisons with data, not from complexity alone. The practical capability is dynamical-systems reasoning. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Dynamical systems study change over time” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: A sensitive system is not necessarily random.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

Graphs represent entities and connections. Paths, centrality, communities and flows can illuminate systems from transport to biology.

Mathematical collaboration often begins when different fields discover the same structure in networks, uncertainty, optimisation, geometry or change over time. The practical capability is network-science literacy. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Networks make relationships measurable” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: A network diagram depends on how nodes and edges were defined.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

Geometric research studies distance, curvature, symmetry and topology across dimensions. Visual intuition helps, while definitions carry the argument beyond pictures.

Mathematical research makes assumptions and definitions explicit before reasoning from them. This discipline lets a result travel across applications without losing its conditions. The practical capability is geometric reasoning. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Geometry links form, space and transformation” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: A drawing may represent only one projection or special case.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

Variables and equations isolate relationships relevant to a scientific or social question. Calibration and validation compare model behaviour with evidence.

A proof explains why a statement follows in every permitted case; computation can reveal patterns and test examples but does not automatically replace the argument. The practical capability is mathematical-modelling literacy. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Models translate questions into mathematics” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: A mathematically correct solution can describe the wrong real-world model.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

IMS research programmes bring specialists and neighbouring fields together around a current theme. Sustained interaction supports joint questions, tutorials and new work.

Models simplify reality deliberately. Their value comes from transparent choices, sensitivity checks and comparisons with data, not from complexity alone. The practical capability is research-programme literacy. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Thematic programmes concentrate collaboration” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: Programme participation does not guarantee a particular research outcome.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

Tutorials build shared foundations, while workshops expose open problems and current results. Asking precise questions and recording sources makes both more useful.

Mathematical collaboration often begins when different fields discover the same structure in networks, uncertainty, optimisation, geometry or change over time. The practical capability is mathematical-communication literacy. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Workshops and tutorials serve different learning moves” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: A research talk summary should not overstate preliminary work.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter · Next chapter

Archive definitions, assumptions, examples, conjecture, proof or algorithm, code, precision, lemmas, counterexamples, model boundary, data comparison, review, revision and unresolved questions.

Mathematical research makes assumptions and definitions explicit before reasoning from them. This discipline lets a result travel across applications without losing its conditions. The practical capability is mathematical-sciences career preparation. Keep the question, sample or model, method, raw evidence, processing, uncertainty, review and next decision in one dated receipt. IMS and NUS control current programmes, participation and released materials; universities and responsible owners control admissions, qualifications and research decisions.

Put this into practice by turning the idea into a small, reviewable action: Create a one-page evidence card for “Keep a mathematical research receipt” with the question, source or sample, method, observation, limitation and next check. Then ask what the resulting note or artefact actually demonstrates. A useful record names context, source or instruction, action, observation, interpretation and uncertainty. It also preserves this limit: Only the responsible programme or institution controls current participation and qualification requirements.

An evidence check keeps confidence proportional to what the work can show. Separate the original observation or official statement from the learner’s explanation. Record feedback and any revision. If information is missing, write “not shown” or “not yet known” rather than guessing. That honest gap can become the next question.

Contents · Previous chapter