Isogeometric Analysis and Numerical Methods for Partial Differential Equations

Isogeometric Analysis (IGA) represents a paradigm shift in the numerical solution of partial differential equations by unifying the design and analysis stages through the use of spline-based basis functions. Unlike traditional finite element methods, which approximate geometry and solution fields separately, IGA employs the same non-uniform rational B-splines (NURBS) or B-spline representations used in computer‐aided design to achieve exact geometry representation and high inter‐element continuity. This approach yields superior approximation properties, particularly for problems requiring smooth solutions or involving complex curved domains. Over the past decade, IGA has been extended to a broad class of numerical schemes—including Galerkin, collocation, discontinuous Galerkin and mixed formulations—enabling robust and accurate simulation of phenomena in structural mechanics, fluid dynamics, electromagnetics and bioengineering. Key challenges such as efficient assembly of isogeometric matrices, the design of preconditioners and multilevel solvers, and the development of structure-preserving discretisations have driven considerable methodological innovation. In parallel, applications ranging from microstructured material modelling to cardiac electromechanics and neuronal transport have demonstrated the global significance of IGA for industrial design, scientific computing and emerging biomedical simulations.

Research from Nature Portfolio

Recent studies have harnessed the strengths of IGA to tackle complex biological transport and growth processes. One investigation developed a graph neural network (GNN) surrogate trained on high‐fidelity IGA simulations of material diffusion within branched neurite networks. The GNN predicts time‐resolved concentration fields with errors below 10 per cent while accelerating computations by over two orders of magnitude, thus enabling rapid exploration of transport dynamics across diverse neuronal geometries. Another work introduced an isogeometric collocation framework coupled with a phase field model to simulate the staged growth of neurites. By incorporating tubulin transport and interface evolution into a single spline-based discretisation, the model reproduces lamellipodia formation, neurite extension and dendritic branching with quantitative agreement to experimental observations and is released as an open-source tool for the neuroscience community.

Research from all publishers

A suite of advances has addressed the algebraic and topological aspects of IGA discretisations. Researchers constructed isogeometric discrete differential forms using smooth spline spaces and Bézier extraction, preserving the de Rham complex structure and yielding pointwise conservation of physical quantities; applications demonstrate optimal approximation of incompressible Stokes flow on curved surfaces. In the realm of solver technology, comparative studies of p-multigrid and h-multigrid strategies have shown that coarsening in spline degree combined with robust ILUT smoothing achieves convergence rates effectively independent of mesh size and polynomial degree, offering a scalable route to solve large linear systems arising from high-order IGA. On the application front, an isogeometric mixed collocation scheme for nearly incompressible electromechanics has been proposed for cardiac simulations, leveraging a staggered solution strategy and mixed finite-elasticity formulation to mitigate volumetric locking while maintaining high accuracy under h-refinement.

Isogeometric Analysis and Numerical Methods for Partial Differential Equations publication trend

The graph below shows the total number of articles in isogeometric analysis and numerical methods for partial differential equations across all publications each year (not limited to Nature Index journals).

Technical terms

Isogeometric Analysis (IGA): A numerical framework that employs CAD spline functions (NURBS, B-splines) for PDE discretisation, ensuring exact geometry and high inter-element continuity.

NURBS: Non-Uniform Rational B-Splines, a generalised spline basis capable of exactly representing conic sections and complex free-form geometries.

Galerkin method: A weighted residual technique in which test and trial functions belong to the same function space, widely used for the weak formulation of PDEs.

Collocation method: A direct discretisation approach enforcing the strong form of PDEs at selected points (collocation nodes) within the domain.

Phase field method: A diffuse-interface modelling technique using smooth order parameters to capture evolving boundaries and morphological changes.

Multigrid method: A hierarchical solver that accelerates convergence by solving the problem on a sequence of mesh resolutions, transferring corrections between levels.

Discrete differential forms: A framework for discretising differential forms on meshes that preserves topological and conservation properties of the continuous de Rham complex.

Graph Neural Network: A machine learning architecture for processing data defined on graphs, here used to learn surrogate mappings for PDE solution fields.

References

1. Deep learning of material transport in complex neurite networks. Scientific Reports (2021).
2. Modeling neuron growth using isogeometric collocation based phase field method. Scientific Reports (2022).
3. Isogeometric discrete differential forms: Non-uniform degrees, Bézier extraction, polar splines and flows on surfaces. Computer Methods in Applied Mechanics and Engineering (2021).
4. p -multigrid methods and their comparison to h -multigrid methods within Isogeometric Analysis. Computer Methods in Applied Mechanics and Engineering (2020).
5. Isogeometric mixed collocation of nearly-incompressible electromechanics in finite deformations for cardiac muscle simulations. Computer Methods in Applied Mechanics and Engineering (2023).
6. Fast and multiscale formation of isogeometric matrices of microstructured geometric models. Computational Mechanics (2021).