Ai Scientific Hpc

Scientific Computing Foundations

Doctoral work examines computation applied to problems across the sciences. Large scale computing enables investigation that experiment alone cannot reach.

Parallel Architecture Research

Research examines machine designs executing many operations simultaneously. Architecture determines what performance any program can possibly achieve.

Processor Architecture Research

Doctoral study examines the design of computing units within scientific machines. Processor design shapes how numerical work is actually executed.

Multicore Research

Research examines processors containing several independent computing cores. Core counts rose once single core clock improvement effectively stopped.

Manycore Research

Doctoral work examines processors with very large numbers of simpler cores. Many simple cores deliver throughput that few complex cores cannot.

Vector Processing Research

Research examines hardware applying one operation across many data elements. Vector execution underpins efficiency in most numerical computation.

Accelerator Architecture Research

Doctoral study examines specialised hardware supplementing general purpose processors. Accelerators now supply most computing capacity in leading systems.

Graphics Processor Computing

Research examines graphics hardware repurposed for general scientific computation. These devices transformed the economics of large scale simulation.

Specialised Accelerator Research

Doctoral work examines hardware designed for particular computational patterns. Specialisation buys efficiency at the cost of general flexibility.

Reconfigurable Hardware Research

Research examines circuits configured to match a specific computational problem. Reconfiguration suits patterns that fixed processors handle poorly.

Memory Hierarchy Research

Doctoral study examines layered storage between processors and main memory. Hierarchy behaviour governs performance more than arithmetic speed does.

Cache Architecture Research

Research examines fast small memories holding recently accessed data. Cache behaviour frequently determines whether a program runs efficiently.

High Bandwidth Memory

Doctoral work examines memory technologies delivering very high transfer rates. Bandwidth is the binding constraint for most scientific workloads.

Memory Bandwidth Research

Research examines the rate at which data reaches the computing units. Most scientific codes are limited by bandwidth rather than by arithmetic.

Nonvolatile Memory Research

Doctoral study examines memory retaining contents without any continuous power. Persistent memory blurs the boundary between memory and storage.

Near Memory Computing

Research examines performing computation close to where the data actually resides. Moving computation to data costs far less than the reverse.

Interconnect Research

Doctoral work examines networks linking computing nodes within a system. Interconnect quality determines how well applications scale across nodes.

Network Topology Research

Research examines how nodes are connected within large computing systems. Topology determines communication cost between any given pair of nodes.

Communication Latency Research

Doctoral study examines delay before transferred data begins to arrive. Latency limits algorithms requiring frequent small message exchanges.

Collective Operation Research

Research examines coordinated communication involving many processes at once. Collective performance frequently dominates overall application scaling.

Node Architecture Research

Doctoral work examines the design of individual computing units within systems. Node balance determines which workloads a system actually suits.

Heterogeneous System Research

Research examines systems combining differing kinds of computing hardware. Heterogeneity raises efficiency and complicates programming considerably.

System Balance Research

Doctoral study examines proportions between computation, memory and communication. Imbalanced systems waste capacity that applications cannot exploit.

Exascale Architecture Research

Research examines systems reaching extremely large sustained computing rates. Reaching this scale required rethinking power, resilience and programming.

Postexascale Research

Doctoral work examines architectures for the generation beyond current systems. Conventional scaling paths are approaching genuine physical limits.

Cooling And Thermal Research

Research examines removing heat generated by dense computing installations. Thermal capability constrains how densely hardware can be packed.

Power Efficiency Research

Doctoral study examines computation delivered per unit of electrical energy. Power consumption is now the principal constraint on system scale.

Energy Aware Computing

Research examines software making decisions accounting for energy consumption. Energy aware scheduling reduces cost without proportional performance loss.

Power Capping Research

Doctoral work examines operating systems within fixed electrical power limits. Capping trades peak performance for predictable facility demand.

Facility Design Research

Research examines buildings and infrastructure housing large computing systems. Facility constraints determine what systems can actually be installed.

System Reliability Research

Doctoral study examines how frequently components fail within large systems. At extreme scale some component is failing almost continuously.

Fault Tolerance Research

Research examines applications continuing correctly despite hardware failures. Tolerance is essential once runs outlast component reliability.

Checkpoint Research

Doctoral work examines saving program state to permit restart after failure. Checkpoint cost grows sharply as system memory capacity increases.

Silent Error Research

Research examines corruption occurring without any detection or warning. Undetected corruption produces wrong answers that appear entirely plausible.

Resilience Modelling Research

Doctoral study examines predicting failure behaviour of very large systems. Models guide how frequently applications should save their state.

Storage Architecture Research

Research examines systems holding the enormous datasets simulations produce. Storage capability increasingly limits what science can be attempted.

Parallel File System Research

Doctoral work examines file systems serving many processes simultaneously. File system contention is a common and underdiagnosed bottleneck.

Burst Buffer Research

Research examines fast intermediate storage absorbing sudden write demands. Buffering smooths bursts that would otherwise overwhelm storage systems.

Object Storage Research

Doctoral study examines storage organised around objects rather than files. Object approaches scale better for very large scientific datasets.

Tiered Storage Research

Research examines arranging storage across layers with differing characteristics. Tiering balances access speed against total capacity and cost.

Input Output Research

Doctoral work examines moving data between applications and storage systems. Data handling frequently consumes more time than computation itself.

Data Movement Research

Research examines transferring data within and between computing systems. Movement dominates both energy consumption and elapsed execution time.

Compression Research

Doctoral study examines reducing data volume within scientific workflows. Lossy approaches raise difficult questions about acceptable accuracy loss.

Network Protocol Research

Research examines communication rules governing data exchange between nodes. Protocol design determines achievable bandwidth and message latency.

Virtualisation Research

Doctoral work examines abstraction layers separating software from physical hardware. Abstraction eases portability and introduces performance overhead.

Container Research

Research examines packaging applications with their complete software environment. Containers substantially improve reproducibility of computational work.

Cloud Computing Research

Doctoral study examines commercial infrastructure serving scientific computing needs. Cloud offers elasticity and suits tightly coupled work poorly.

Edge Integration Research

Research examines linking instruments and remote devices with computing centres. Instrument data streams increasingly require immediate computational response.

Quantum Integration Research

Doctoral work examines quantum devices operating alongside conventional systems. Integration questions arise well before quantum advantage is achieved.

Neuromorphic Integration

Research examines brain inspired hardware within scientific computing systems. These devices promise efficiency for particular computational patterns.

Parallel Programming Research

Doctoral study examines expressing computation for simultaneous execution. Programming difficulty is the persistent barrier to exploiting hardware.

Message Passing Research

Research examines processes coordinating through explicit data exchange. Message passing remains the dominant model for distributed scientific codes.

Shared Memory Programming

Doctoral work examines threads cooperating through commonly accessible memory. Shared approaches suit work performed within a single computing node.

Task Based Programming

Research examines expressing work as tasks with explicitly declared dependencies. Task models adapt to imbalance that static approaches handle poorly.

Data Parallel Programming

Doctoral study examines applying identical operations across large data collections. This model maps naturally onto accelerator hardware designs.

Directive Based Programming

Research examines annotations guiding compilers to parallelise existing code. Directives ease adoption within very large established codebases.

Accelerator Programming Research

Doctoral work examines writing software targeting specialised computing devices. Accelerator programming demands skills most scientists do not have.

Portability Research

Research examines running the same code correctly across differing systems. Portability protects long lived codes from hardware generation change.

Performance Portability

Doctoral study examines achieving good performance across differing architectures. Running everywhere is easier than running well everywhere.

Domain Specific Language

Research examines languages designed for particular scientific problem classes. Restricted languages permit optimisation that general ones cannot.

Compiler Research

Doctoral work examines translating source code into efficient machine instructions. Compiler capability determines how much manual tuning remains necessary.

Automatic Parallelisation

Research examines compilers extracting parallelism without programmer direction. Automatic extraction succeeds for regular patterns and struggles otherwise.

Code Generation Research

Doctoral study examines producing optimised code from higher level descriptions. Generation separates scientific intent from machine specific detail.

Runtime System Research

Research examines software managing execution while a program is running. Runtime decisions adapt to conditions compilers cannot anticipate.

Job Scheduling Research

Doctoral work examines allocating system resources among competing user jobs. Scheduling policy determines both utilisation and user waiting time.

Load Balancing Research

Research examines distributing work evenly across available computing resources. Imbalance wastes capacity because idle processes still consume power.

Work Stealing Research

Doctoral study examines idle processors taking work from busier neighbours. Stealing balances load without any central coordinating authority.

Synchronisation Research

Research examines coordinating processes that must proceed together. Synchronisation cost limits scaling more than computation frequently does.

Concurrency Research

Doctoral work examines correctness when many operations proceed simultaneously. Concurrency errors are intermittent and extremely difficult to diagnose.

Communication Avoidance

Research examines algorithms deliberately minimising data exchange between processes. Avoiding communication matters because movement dominates cost.

Latency Hiding Research

Doctoral study examines overlapping communication with useful computation. Overlap conceals delays that would otherwise idle the processors.

Memory Management Research

Research examines allocating and organising memory within scientific applications. Allocation patterns strongly influence achieved memory performance.

Data Layout Research

Doctoral work examines arranging data structures for efficient hardware access. Layout choice can change performance by genuinely very large factors.

Locality Optimisation Research

Research examines organising computation to reuse nearby data repeatedly. Locality determines whether the memory hierarchy helps or hinders.

Vectorisation Research

Doctoral study examines restructuring code to exploit vector hardware units. Failed vectorisation leaves most arithmetic capability entirely unused.

Mixed Precision Research

Research examines using differing numerical precisions within one calculation. Lower precision runs faster and must not compromise final accuracy.

Reduced Precision Research

Doctoral work examines computing with fewer bits than traditional practice. Hardware increasingly favours precisions designed for learned models.

Numerical Reproducibility

Research examines obtaining identical results across runs and machines. Parallel arithmetic ordering makes exact reproduction genuinely difficult.

Performance Modelling Research

Doctoral study examines predicting how fast applications will actually run. Models guide design decisions before any code has even been written.

Performance Analysis Research

Research examines measuring and interpreting where execution time is spent. Intuition about performance bottlenecks is frequently entirely wrong.

Profiling Tool Research

Doctoral work examines instruments measuring application execution behaviour. Tool usability determines whether scientists actually analyse performance.

Tracing Research

Research examines recording detailed event sequences during program execution. Traces reveal interaction patterns that summary statistics conceal.

Bottleneck Analysis Research

Doctoral study examines identifying what actually limits application performance. Removing the wrong bottleneck yields no improvement whatsoever.

Roofline Analysis Research

Research examines graphical models relating performance to hardware limits. This framing quickly shows whether memory or arithmetic constrains a code.

Autotuning Research

Doctoral work examines automatically searching for optimal code configurations. Automatic search finds settings that human tuning would never try.

Optimisation Search Research

Research examines navigating very large spaces of possible code variants. Search efficiency determines whether tuning is practically feasible.

Benchmark Research

Doctoral study examines standard programs measuring system computing capability. Benchmark choice determines which systems appear most capable.

Benchmark Design Research

Research examines constructing tests that genuinely represent real workloads. Established benchmarks reflect scientific practice increasingly poorly.

Software Engineering Research

Doctoral work examines development practice within scientific software projects. Scientific codes are long lived and frequently poorly engineered.

Software Testing Research

Research examines verifying that scientific software behaves as intended. Testing numerical code requires approaches ordinary testing does not.

Continuous Integration Research

Doctoral study examines automated checking whenever software is modified. Automation catches breakage that manual checking routinely misses.

Version And Provenance Research

Research examines tracking software versions used to produce scientific results. Without version records, computational results cannot be reproduced.

Scientific Library Research

Doctoral work examines reusable software components underpinning scientific codes. Libraries concentrate expertise that individual projects cannot sustain.

Numerical Library Research

Research examines optimised implementations of standard mathematical operations. These libraries deliver performance that hand written code rarely matches.

Solver Library Research

Doctoral study examines reusable software solving common equation systems. Solver libraries let scientists avoid reimplementing difficult algorithms.

Simulation Framework Research

Research examines structured platforms supporting whole classes of simulation. Frameworks trade flexibility for very substantially reduced effort.

Workflow System Research

Doctoral work examines software coordinating multistep computational pipelines. Workflow systems make complex analyses repeatable and describable.

Workflow Orchestration Research

Research examines managing execution of workflows across differing resources. Orchestration handles failure and retry that manual running cannot.

Legacy Code Research

Doctoral study examines long established codes still central to scientific work. Legacy codes encode understanding nobody currently fully possesses.

Code Modernisation Research

Research examines adapting long established codes for contemporary hardware. Modernisation must preserve results that decades of use validated.

Technical Debt Research

Doctoral work examines accumulated shortcuts constraining future software change. Debt accumulates quietly and eventually halts development entirely.

Software Sustainability Research

Research examines maintaining scientific software across many funding cycles. Critical software frequently depends on very few unfunded maintainers.

Research Software Engineering

Doctoral study examines the profession developing software for research work. Recognition of this role remains inconsistent across institutions.

Open Source Research

Research examines openly developed software within the scientific computing community. Open development permits scrutiny that closed software prevents.

Documentation Research

Doctoral work examines describing scientific software so that others can use it. Poor documentation is the commonest barrier to software reuse.

Numerical Method Research

Research examines algorithms approximating solutions to mathematical problems. Method choice determines both accuracy and computational cost.

Discretisation Research

Doctoral study examines representing continuous problems using finite representations. Discretisation choice governs accuracy and computational expense.

Finite Difference Research

Research examines approximating derivatives using values at neighbouring points. This approach is simple and suits regularly structured problems.

Finite Element Research

Doctoral work examines dividing domains into elements with local approximations. This approach handles complex geometry that simpler methods cannot.

Finite Volume Research

Research examines methods conserving quantities across small control volumes. Conservation properties matter greatly in fluid and transport problems.

Spectral Method Research

Doctoral study examines representing solutions using global basis functions. Spectral approaches achieve high accuracy for sufficiently smooth problems.

Mesh Generation Research

Research examines dividing complex domains into computational elements. Mesh quality strongly determines both accuracy and solver convergence.

Adaptive Mesh Research

Doctoral work examines refining resolution where the solution demands it. Adaptivity concentrates effort where accuracy is genuinely needed.

Meshfree Method Research

Research examines methods avoiding any fixed computational mesh entirely. Meshfree approaches suit problems involving very large deformation.

Particle Method Research

Doctoral study examines representing systems as collections of moving particles. Particle methods suit problems where tracking material matters.

Linear Solver Research

Research examines solving the large linear systems that simulations produce. Solving these systems consumes most time within many applications.

Iterative Method Research

Doctoral work examines approaching solutions through successive approximation. Iterative methods suit the very large systems simulations generate.

Preconditioning Research

Research examines transforming systems so iterative solvers converge faster. Preconditioner choice matters more than solver choice frequently does.

Direct Solver Research

Doctoral study examines methods solving systems through systematic elimination. Direct methods are robust and scale poorly to very large problems.

Eigenvalue Method Research

Research examines computing characteristic values of large matrix systems. These computations underpin stability and structure analysis widely.

Sparse Matrix Research

Doctoral work examines matrices where the great majority of entries are zero. Exploiting sparsity is what makes large simulations at all feasible.

Dense Linear Algebra

Research examines operations on matrices having mostly nonzero entries. Dense operations achieve efficiency that sparse ones simply cannot approach.

Tensor Computation Research

Doctoral study examines operations on multidimensional numerical arrays. Tensor operations connect scientific computing with learned model workloads.

Fast Transform Research

Research examines efficient algorithms converting between differing representations. These transforms underpin an enormous range of scientific codes.

Multigrid Method Research

Doctoral work examines solving problems using a whole hierarchy of resolutions. Multigrid achieves convergence rates independent of problem size.

Domain Decomposition Research

Research examines dividing problems into subregions that are solved separately. Decomposition is the natural route to distributed parallel solution.

Time Integration Research

Doctoral study examines advancing solutions forward through simulated time. Step size choice balances accuracy against total computational cost.

Stiff System Research

Research examines problems containing very differing characteristic timescales. Stiff problems force impractically small steps for explicit methods.

Multiscale Method Research

Doctoral work examines coupling models across differing spatial or time scales. Scale coupling permits detail that single scale models cannot.

Multiphysics Coupling Research

Research examines joining models of differing physical processes together. Coupling introduces stability problems absent from single physics models.

Discretisation Error Research

Doctoral study examines error introduced by representing continuous problems finitely. Quantifying this error is essential for trusting any result.

Code Verification Research

Research examines confirming that software solves the equations intended. Verification asks whether the equations are being solved correctly.

Model Validation Research

Doctoral work examines comparing simulation predictions against physical measurement. Validation asks whether the correct equations are being solved.

Uncertainty Quantification

Research examines expressing confidence in computational simulation predictions. Simulations without uncertainty estimates cannot support real decisions.

Sensitivity Analysis Research

Doctoral study examines how outputs respond to changes in model inputs. Sensitivity identifies which inputs genuinely deserve careful measurement.

Inverse Problem Research

Research examines inferring causes from observed effects computationally. Inverse problems are typically ill posed and demand careful regularisation.

Optimisation Method Research

Doctoral work examines finding best solutions within scientific computing problems. Optimisation at scale requires distributed algorithmic approaches.

Monte Carlo Method Research

Research examines solving problems through repeated random sampling of outcomes. These methods parallelise very naturally across enormous systems.

Stochastic Method Research

Doctoral study examines algorithms incorporating randomness into their operation. Stochastic approaches suit problems too large for exact treatment.

Reduced Order Modelling

Research examines simplified models capturing essential system behaviour. Reduced models run fast enough for repeated evaluation and design.

Surrogate Model Research

Doctoral work examines fast approximations standing in for costly simulations. Surrogates enable exploration that direct simulation cannot support.

Physics Informed Learning

Research examines learned models constrained by known physical laws. Physical constraints improve behaviour beyond the conditions used in training.

Learned Solver Research

Doctoral study examines learned components accelerating numerical solution. Learned solvers must preserve guarantees classical methods provide.

Differentiable Simulation

Research examines simulations permitting gradients through the whole computation. Differentiability enables optimisation and inference through simulation.

Simulation Learning Coupling

Doctoral work examines combining traditional simulation with learned components. Coupling is reshaping how scientific computing systems are designed.

Foundation Model Training

Research examines training very large models on scientific computing systems. These workloads now substantially shape how new systems are procured.

Distributed Training Research

Doctoral study examines spreading model training across many computing nodes. Training scaling raises the communication problems simulation faced earlier.

Model Parallelism Research

Research examines splitting single models across differing computing devices. Splitting is necessary once models exceed any single device memory.

Training Communication Research

Doctoral work examines data exchange patterns during distributed model training. Communication frequently limits achievable training throughput.

Inference At Scale Research

Research examines serving learned models efficiently within scientific workflows. Inference cost accumulates greatly across very many evaluations.

Scientific Data Analysis

Doctoral study examines extracting understanding from very large computed datasets. Analysis capability increasingly limits what simulation delivers.

In Situ Analysis Research

Research examines analysing results while a simulation is still running. Analysing during execution avoids writing data that cannot be stored.

Scientific Visualisation Research

Doctoral work examines graphical representation of computed scientific results. Visualisation is how researchers actually interpret simulation output.

Large Data Visualisation

Research examines visualising datasets exceeding any single machine capacity. Rendering must proceed in parallel alongside the simulation itself.

Computational Steering Research

Doctoral study examines researchers influencing simulations while they are running. Steering permits exploration without waiting for complete runs.

Climate Simulation Research

Research examines computational modelling of the whole climate system. Climate models are among the largest sustained computing efforts anywhere.

Weather Prediction Research

Doctoral work examines computational forecasting of atmospheric conditions. Forecasts must complete within very strict operational time limits.

Ocean Modelling Research

Research examines computational simulation of ocean circulation and processes. Ocean models couple with atmosphere models in earth system work.

Earth System Research

Doctoral study examines coupled models spanning atmosphere, ocean, ice and land. Coupled models demand enormous sustained computing capacity.

Astrophysics Simulation

Research examines computational modelling of stars and astronomical objects. Simulation substitutes for experiments that could never be performed.

Cosmology Simulation Research

Doctoral work examines simulating structure formation across the whole universe. These simulations span an extraordinary range of physical scales.

Particle Physics Computing

Research examines computation supporting experimental particle physics research. These experiments generate among the largest datasets in all science.

Nuclear Physics Computing

Doctoral study examines computational modelling of atomic nuclei and reactions. Nuclear computation demands both accuracy and enormous resources.

Plasma Simulation Research

Research examines computational modelling of ionised gases and their behaviour. Plasma simulation underpins fusion energy development work directly.

Fluid Dynamics Computing

Doctoral work examines computational simulation of fluid flow and turbulence. Turbulence remains among the hardest problems in scientific computing.

Structural Simulation Research

Research examines computational analysis of solid mechanical structures. Structural simulation underpins engineering design across many industries.

Combustion Simulation Research

Doctoral study examines modelling chemical reaction coupled with fluid motion. Combustion couples chemistry and flow across very differing scales.

Materials Simulation Research

Research examines computational prediction of material structure and properties. Simulation accelerates discovery that laboratory synthesis alone cannot.

Quantum Chemistry Computing

Doctoral work examines computing molecular properties from quantum principles. Accuracy demands scale very steeply with molecular system size.

Molecular Dynamics Research

Research examines simulating motion of atoms and molecules over time. Reaching biologically meaningful durations remains genuinely difficult.

Electronic Structure Research

Doctoral study examines computing electron arrangements within materials and molecules. Electronic structure underlies nearly all chemical prediction.

Therapeutic Discovery Computing

Research examines computation supporting discovery of new medicines. Computational screening narrows candidates before laboratory testing begins.

Structural Biology Computing

Doctoral work examines computing biological molecule structures and interactions. Structure prediction has advanced enormously through learned models.

Genomics Computing Research

Research examines computation handling very large genetic sequence datasets. Sequencing output has grown faster than computing capacity has.

Neuroscience Computing Research

Doctoral study examines computation supporting brain research and simulation. Neural simulation demands scale far beyond current routine capability.

Medical Imaging Computing

Research examines computation reconstructing and analysing medical images. Reconstruction quality depends directly on available computing capacity.

Digital Twin Research

Doctoral work examines computational replicas mirroring real operating systems. Replicas permit testing scenarios that reality cannot safely allow.

Engineering Design Computing

Research examines simulation supporting design of engineered products. Simulation now replaces physical prototyping across many design stages.

Seismic Computing Research

Doctoral study examines computation modelling wave propagation through the earth. Seismic computation supports both hazard assessment and exploration.

Energy System Simulation

Research examines computational modelling of power networks and energy systems. System models guide planning of very long lived infrastructure.

Social Simulation Research

Doctoral work examines computational modelling of populations and their behaviour. Social models raise validation questions physical models avoid.

Scientific Data Management

Research examines organising the enormous datasets computational science produces. Management practice determines whether data remains usable later.

Data Standard Research

Doctoral study examines agreed formats for scientific computational datasets. Standards determine whether data moves between differing tools.

Computational Reproducibility

Research examines whether computed results can be independently regenerated. Environment differences defeat reproduction even with shared code.

Provenance Research

Doctoral work examines recording how computational results were actually produced. Provenance records are essential for trusting published results.

Federated Infrastructure

Research examines linking computing facilities across institutions and countries. Federation shares capacity that no single centre could provide.

Resource Allocation Research

Doctoral study examines distributing computing capacity among competing projects. Allocation decisions determine which science actually gets done.

Allocation Policy Research

Research examines rules governing access to shared computing facilities. Policy design influences which research communities are actually served.

System Utilisation Research

Doctoral work examines how effectively installed computing capacity is used. Reported utilisation frequently overstates genuinely productive computation.

Sustainability Research

Research examines environmental burden of large scale scientific computing. Computing demand is rising faster than efficiency improvements are.

Carbon Footprint Research

Doctoral study examines emissions attributable to scientific computation. Accounting must cover hardware manufacture as well as operational energy.

Access Equity Research

Research examines who can obtain access to large scale computing resources. Access concentration shapes which scientific questions get pursued.

Global Capacity Research

Doctoral work examines distribution of computing capability between countries. Capability is concentrated in a very small number of nations.

Low Resource Computing

Research examines scientific computation where facilities are severely limited. Efficient methods matter most where resources are genuinely scarce.

Training And Education

Doctoral study examines preparing scientists to use computing effectively. Most computational scientists received no formal software training.

Workforce Research

Research examines the people building and operating scientific computing systems. Combined domain and computing expertise is genuinely very scarce.

Community Building Research

Doctoral work examines communities forming around scientific software and methods. Communities sustain software that individual projects cannot.

Procurement Research

Research examines how large computing systems are specified and then acquired. Procurement decisions commit facilities for very long periods.

Economic Evaluation Research

Doctoral study examines value delivered by investment in scientific computing. Value evidence is required to justify very large public spending.

Implementation And Adoption

Research examines why computing advances reach practice or fail to do so. Scientists adopt new approaches far more slowly than developers expect.