Our Work

What's Changed move PointNotInDomainError by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5041 ASMStar: enable patch coloring by @pbrubeck in https://github.com/firedrakeproject…

What's Changed move PointNotInDomainError by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5041 ASMStar: enable patch coloring by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/4912 Use firedrake-rtree for spatial indexing by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/4981 New plotting methods for VertexOnlyMesh by @achanbour in https://github.com/firedrakeproject/firedrake/pull/5039 Use mat context for Vom onto vom matfree interpolation by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5055 Merge release into main by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5056 Assembly of matfree interpolation operator by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/4778 Refactor PCPatch code to not use PyOP2 by @connorjward in https://github.com/firedrakeproject/firedrake/pull/4993 GPU offloading preconditioner by @dsroberts in https://github.com/firedrakeproject/firedrake/pull/4953 Improve building compiled extensions using Makefile by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5059 DO NOT SQUASH merge release by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5069 DO NOT SQUASH merge release into main by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5080 Keep a consistent labeling of boundaries for 2D periodic meshes by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5005 Adjoint + submesh by @KarsKnook in https://github.com/firedrakeproject/firedrake/pull/5081 DO NOT SQUASH merge release by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5115 Don't tee the cat by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5129 DO NOT SQUASH Connorjward/merge release by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5134 Specify shell for GPU usage confirmation step by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5135 Remove add, sub magic methods from matrix by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5145 Drop support for python 3.10 by @francesco-ballarin in https://github.com/firedrakeproject/firedrake/pull/5173 DO NOT SQUASH Merge release into main by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5178 Remove (very) deprecated Matrix.assemble by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5143 Fix for ufl2unicode rewrite by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5058 Interpolation into EnrichedElement by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5184 Prevent Cython 3.2.6 from being installed in build by @indiamai in https://github.com/firedrakeproject/firedrake/pull/5208 cache visible_ranks on topology by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5183 Add AMD compatibility for OffloadPC by @dsroberts in https://github.com/firedrakeproject/firedrake/pull/5220 DO NOT SQUASH merge release by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5221 (DO NOT SQUASH) Merge release into main by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5234 Revert "Prevent Cython 3.2.6 from being installed in build" by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5216 Fix Jp + multigrid by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5189 Fix solver options for Real space by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5144 DumbCheckpoint renew warning by @Ig-dolci in https://github.com/firedrakeproject/firedrake/pull/5228 Cache mesh bounding boxes correctly by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5180 Fix location of community arch by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5261 Remove old PyOP2 documentation by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5262 dmcommon: de-Python-ify mesh-topology construction (~4x faster BoxMesh build) by @miguelcoolchips in https://github.com/firedrakeproject/firedrake/pull/5231 Write VOM data to VTKFile by @achanbour in https://github.com/firedrakeproject/firedrake/pull/5244 (DON't SQUASH) Leo/merge release by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5272 Don't import interpolate from future by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5269 Add support for objective function in SNES by @stefanozampini in https://github.com/firedrakeproject/firedrake/pull/5155 Adaptive NonlinearVariationalSolver support via -snes_adapt_sequence by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5213 (DON'T SQUASH) Leo/merge release by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5284 DO NOT SQUASH Merge release into main by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5293 Access DMSwarm fields via a context manager by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5296 rtree bulk query by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5247 Connorjward/merge release by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5302 Mesh Adaptivity via DMPlexTransform by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5215 Leo/merge release by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5314 Leo/deprecate clear rtree by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5315 fix rpaths for macos by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5301 Remove dead code by @achanbour in https://github.com/firedrakeproject/firedrake/pull/5320 Fix coarse-to-fine cell maps for hierarchies built on an adaptively refined mesh by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5321 interpolator init cleanup by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5319 Fix quadratic cost in adaptive mesh refinement's coarse/fine cell-map construction by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5326 Rework PMGBase.coarsen() and add NonlinearVariationalProblem.rediscretise() by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5324 Fix broken test with exact comparisons by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5353 Contract quadrature reductions in loop nest order by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5354 tsfc: cancel Jacobian products and reciprocal factors by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5230 Test interpolation of products of coefficient evaluations by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5360 Make serial submesh construction linear in the number of points by @haldaas in https://github.com/firedrakeproject/firedrake/pull/5349 Lint cython code by @stefanozampini in https://github.com/firedrakeproject/firedrake/pull/5363 DO NOT SQUASH Connorjward/merge release by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5364 Fix cython linting by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5366 Minor updates to OffloadPC by @dsroberts in https://github.com/firedrakeproject/firedrake/pull/5164 tinyasm: avoid copying the block dofmap in BlockJacobi::solve() by @manishpaulish in https://github.com/firedrakeproject/firedrake/pull/5348 DO NOT SQUASH merge release by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5382 Check that equivalent cells compile a form the same way by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5375 Fix complex CI by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5385 Compile interpolation through the GEM optimisation pipeline by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5381 Use correct dtype during PCPatch codegen by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5395 DO NOT SQUASH Merge release by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5404 Unified cache under sys.prefix by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5400 Remove backwards compat tests for variable layer extrusion by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5418 Skip failing base form coarsening test by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5420 Test multigrid coarsening of an appctx coefficient by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5436 autoregressive covariance operator: fix scaling for dim>1 by @JHopeCollins in https://github.com/firedrakeproject/firedrake/pull/5358 p-multigrid: remove custom interpolation code by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5289 Leo/merge release by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5443 Replace :: with codeblocks in demos by @nwng04 in https://github.com/firedrakeproject/firedrake/pull/5414 Let PCASM refresh its own submatrices by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5403 Drop the MatIS special case in FunctionSpace.local_to_global_map by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5437 Make the Stokes multigrid appctx test deterministic by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5453 Extruded Periodic MeshHierarchy by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/4663 Stop DG injection reading past a coarse cell's children by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5337 Fix NumPy 2.5 deprecation warning in pyop2.utils.verify_reshape by @TrueCP6 in https://github.com/firedrakeproject/firedrake/pull/5459 DO NOT SQUASH merge release into main by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5457 fix petsc4py breaking changes by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5463 MeshHierarchy: assemble prolongation into MatAIJ by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5206 more petsc4py fixes by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5464 Support physical_normals on manifolds by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/4994 DO NOT SQUASH Merge release by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5466 Offload Nullspaces with OffloadPC by @dsroberts in https://github.com/firedrakeproject/firedrake/pull/5462 Add complex Slate support by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5470 fix assertion in vom test by @leo-collins in https://github.com/firedrakeproject/firedrake/pull/5473 Make SLATE subclass BaseForm by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/4093 Multigrid: derive hierarchy maps from DMPlex transforms by @pbrubeck in https://github.com/firedrakeproject/firedrake/pull/5460 Enable SingleMemoryStorageSchedule in test_checkpointing_validity by @sghelichkhani in https://github.com/firedrakeproject/firedrake/pull/5455 make the ensemble diagram not be square by @dham in https://github.com/firedrakeproject/firedrake/pull/5495 DO NOT SQUASH Merge main into release by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5494 Set UFL and FIAT versions by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5499 Update slepc4py install instructions by @connorjward in https://github.com/firedrakeproject/firedrake/pull/5500 Regression tests for SingleMemoryStorageSchedule checkpoint clearing by @sghelichkhani in https://github.com/firedrakeproject/firedrake/pull/5454 New Contributors @haldaas made their first contribution in https://github.com/firedrakeproject/firedrake/pull/5349 @manishpaulish made their first contribution in https://github.com/firedrakeproject/firedrake/pull/5348 @nwng04 made their first contribution in https://github.com/firedrakeproject/firedrake/pull/5414 @TrueCP6 made their first contribution in https://github.com/firedrakeproject/firedrake/pull/5459 Full Changelog: https://github.com/firedrakeproject/firedrake/compare/2026.4.3...2026.10.0

Grand strategy — the hierarchical subordination of means to ends across time and domain — has become structurally impossible in the contemporary United States and European Union. This is n…

Grand strategy — the hierarchical subordination of means to ends across time and domain — has become structurally impossible in the contemporary United States and European Union. This is not a claim about leadership, ideology, or democratic dysfunction. It is a claim about slack: the degrees of freedom available to a system for purposive reconfiguration. Cross-sector coupling, whose monotonic increase since 1945 is established in The Tragedy of the Citadel, has consumed that slack. As coupling rises, every sector becomes a veto point, every adjustment reverberates across the whole, and the system's only remaining imperative is equilibrium maintenance. What replaces strategy is adjustment: the continuous, headless, path-dependent metabolisation of crises into further coupling.  The paper is a continuation of the Citadel monograph. What it contributes is fourfold. First, the concept of strategic slack, with the formal apparatus that distinguishes it from complexity and from prosperity. Second, the three mechanisms by which coupling crowds out strategy: veto proliferation, adjustment costs, and temporal compression. Third, the kinesis/praxis distinction: the headless configuration exhibits behaviour without action, motion without agency. Fourth, the category of the headless ruling structure — a class in itself but not for itself, defined by structural position in the coupling network rather than by collective consciousness, coordinated through coupling rather than through deliberation. The category is distinguished from six adjacent literatures and integrated with the Citadel Class: the Citadel is the extraction face of the structural elite; the headless ruling structure is its governance face.  The argument bears on Ukraine, Iran, and industrial policy, and it extends the phase transition analysis the Citadel established. The headless configuration does not merely fail to satisfy the four conditions for durable decoupling. It actively prevents their satisfaction, because each condition requires an agent whose existence the configuration has eliminated. The paper raises three questions it does not answer — whether headless configurations recur, whether the modern state was an exception or a baseline, and whether the class conflict that produced the couplings has ended or merely changed form — and specifies what would be required to settle them.

Artifact for the paper "GPU-Accelerated Search for Fast Matrix Multiplication over F2" (ALENEX 2027). The archive contains the CUDA kernels for our flip-graph search over F2, the scripts that run ever…

Artifact for the paper "GPU-Accelerated Search for Fast Matrix Multiplication over F2" (ALENEX 2027). The archive contains the CUDA kernels for our flip-graph search over F2, the scripts that run every experiment in the paper's evaluation (Figures 5.1, 5.2, and 6.1 to 6.4), and the rank-245 decomposition of the 7x7x7 matrix multiplication tensor over F2 that is the paper's main result, together with an independent Python verifier for it. The scheme can be verified on any machine with Python 3.9 or newer, with no GPU. Running the experiments requires Linux, one NVIDIA GPU of compute capability 7.0 or newer, and a CUDA 12.x toolkit; the paper's measurements were taken on an NVIDIA H200. A single setup script builds everything without root access, including the two baseline implementations the paper compares against, which it downloads from their authors' public repositories. README.md inside the archive gives the setup command, the verification command, and the command for each figure.

This repository contains the numerical data, source code, processed outputs, and reference figures supporting the study “Numerical Evidence for Stable-Side Riddling and Unstable-Dimension Variab…

This repository contains the numerical data, source code, processed outputs, and reference figures supporting the study “Numerical Evidence for Stable-Side Riddling and Unstable-Dimension Variability in a Parametrically Driven Duffing Dimer.” The archive includes reproducibility material for transverse Lyapunov-exponent and blowout-threshold calculations, periodic-orbit and Floquet analyses, unstable-dimension-variability diagnostics, basin classification and adaptive refinement, full-dimensional competing-attractor capture, covariant Lyapunov-vector and stable–unstable angle analyses, transverse-distance competing-capture scaling, full-system hyperchaos calculations, and stochastic-response simulations. The final reviewer-revision data include the near-balanced finite-time regime at \(K_{\mathrm{bal}}=0.32095\), independent \(80\times80\) basin calculations under two numerical protocols, finite-range uncertainty-exponent estimates, CLV records, and the scaling analysis \(P^\ast(y_0)=C y_0^{\eta_{\mathrm{esc}}}\). The repository is intended to support direct numerical reproduction and independent verification of the reported results. All conclusions concerning stable-side riddling and unstable-dimension variability are presented as finite-resolution numerical evidence rather than as a measure-theoretic proof of a continuum riddled basin.

Online material accompanying the A&A article aa61083-26.

shawk 1.63.1 Install shawk 1.63.1 Install prebuilt binaries via shell script curl --proto '=https' --tlsv1.2 -LsSf https://github.com/oicr-gsi/shesmu/releases/download/v1.63.1/shawk-installer.sh | sh …

shawk 1.63.1 Install shawk 1.63.1 Install prebuilt binaries via shell script curl --proto '=https' --tlsv1.2 -LsSf https://github.com/oicr-gsi/shesmu/releases/download/v1.63.1/shawk-installer.sh | sh Download shawk 1.63.1 | File | Platform | Checksum | |--------|----------|----------| | shawk-aarch64-apple-darwin.tar.xz | Apple Silicon macOS | checksum | | shawk-x86_64-apple-darwin.tar.xz | Intel macOS | checksum | | shawk-x86_64-pc-windows-msvc.zip | x64 Windows | checksum | | shawk-x86_64-pc-windows-msvc.msi | x64 Windows | checksum | | shawk-x86_64-unknown-linux-gnu.tar.xz | x64 Linux | checksum | json-dir-list 1.63.1 Install json-dir-list 1.63.1 Install prebuilt binaries via shell script curl --proto '=https' --tlsv1.2 -LsSf https://github.com/oicr-gsi/shesmu/releases/download/v1.63.1/json-dir-list-installer.sh | sh Download json-dir-list 1.63.1 | File | Platform | Checksum | |--------|----------|----------| | json-dir-list-aarch64-apple-darwin.tar.xz | Apple Silicon macOS | checksum | | json-dir-list-x86_64-apple-darwin.tar.xz | Intel macOS | checksum | | json-dir-list-x86_64-pc-windows-msvc.zip | x64 Windows | checksum | | json-dir-list-x86_64-pc-windows-msvc.msi | x64 Windows | checksum | | json-dir-list-x86_64-unknown-linux-gnu.tar.xz | x64 Linux | checksum |

As published. Creating release for Zenodo archive.

Scroll Score for Shift/Work ISSOTL17 Poster Presentation, Calgary, Canada Shift/Workshops are composed and disseminated by being performed. Playing the score leads to it being re-calibrated for future…

Scroll Score for Shift/Work ISSOTL17 Poster Presentation, Calgary, Canada Shift/Workshops are composed and disseminated by being performed. Playing the score leads to it being re-calibrated for future performances. Speculations was commissioned by the Kochi-Muziris Biennale, India. It was first performed at Edinburgh Sculpture Workshop (3-4 March 2017). At ESW, three ‘branch’ scores were composed to be performed in Kochi (22-23 March 2017) and at the Kristiansand Theatre, University of Agder, Norway (28-20 Aug 2017) producing further branch scores. A SoTL analysis of Speculations was presented at ISSoTL17 in Calgary, October 2017. Recalibrated through intense peer-review, a ‘Scroll Score’ of the generative composition to date was published as an open educational resource (OER) for ISSoTL17 (12 & 14 Oct 2017) and published on shift-work.org.ukShift/Work: Speculations is an innovative methodology that synthesises practice-based and practice-led research methods to transform our understanding and knowledge of artistic research (AR). It develops powerful insights into how Peer-2-Peer methods transmute speculative artistic practices by coupling the taking of risks with experiencing their consequences. Speculations combines new theories of paragogy with SoTL (Scholarship of Teaching & Learning) and Participatory Action Research (PAR) methods to develop and enhance a metacognitive understanding of AR. Through this, it invents a new theoretical framework - the ‘paratechnic’ - a means to collectively ‘score’ and disseminate artistic research as an OER. This is of particular importance to widening participation in the arts and restoring their civic purpose.‘Speculation’ is a unifying ethics (or cognitive bias) in AR commonly expressed as ‘without a script’ or ‘not knowing’. Drawing explicit attention to an implicit ethics within AR, and exposing it to the scrutiny and practices of non-artistic learners, raises awareness of its operations and limitations. This, in turn, enables us to repurpose, invent and practice metacognitive forms of speculation that have operational significance for, and beyond, AR.Speculations (Punctum) and newmaterialism.eu have been pivotal in suggesting ludic and non-anthropocentric forms of emergence that might foreground, and move beyond, AR's cognitive biases. Generative Shift/Work composition activates material probes as autopoetic catalysts for problem creation. As component interactions made radically distinct objects from the same material, speculation enabled Shift/Workers to formulate the concepts, equipment, and techniques to generate the truly collaborative ability to fabricate common research objects.Originality:Speculations is an original contribution to artistic knowledge, art andragogy and artistic research that implements emerging methods in paragogy and pioneers SoTL in art. A theoretically reflexive, participatory, and openly accessible paragogy for artistic research, Speculations liberates artistic knowledges by common-ing the proprietary ‘secret skills’ of the guild-monotechnic.Significance:Speculations has broad political, ontological and epistemological significance. Establishing a collective ontology for art practice though paragogy, Speculations exemplifies collectivist whole-systems thinking. Speculative research methods recognise that art is a subsystem of society - rather than separate from it - and ensure it has a pragmatic sense of its transformative potential.Rigour:Speculations is a scholarly, coherent, iterative system of meta-cognitive peer review. Composition is scaffolded with an academic and artistic community of ‘Shift/Workers’. Shift/Workers compose and play-test each other’s workshop scores before calibrating them through collective peer-review. Scores are then re-performed, iteratively, at international peer-reviewed artistic and scholarly events.

We report the developments obtained after PIHR–GOLDBACH XIX (Gold-19) in the continuing attempt to isolate a rigorous analytic route toward the binary Goldbach conjecture. Gold-19 returned from …

We report the developments obtained after PIHR–GOLDBACH XIX (Gold-19) in the continuing attempt to isolate a rigorous analytic route toward the binary Goldbach conjecture. Gold-19 returned from the endpoint Möbius-variance formulation of Gold-18 to the signed determinant geometry of Gold-17. Its central candidate theorem, Endpoint Coefficient-Preserving Kloosterman Transfer (ECPKT), asked that the parity-sensitive four-weight determinant block be converted, without destructive absolute-value majorization, into bilinear Kloosterman forms with two -controlled coefficient vectors. The post-Gold-19 analysis substantially sharpens that proposal. First, a false shortcut has been eliminated. The centered endpoint large-sieve/Ramaré block does not itself contain the reciprocal geometry required by the strongest Kloosterman estimates. Opening the square produces Ramanujan sums rather than Kloosterman sums, so the modular inverse must already come from the underlying determinant or congruence geometry. Second, the algebraic core of ECPKT can be carried out explicitly. If the character diagonalization of Gold-17 is stopped one Gauss transformation earlier, a two-weight arithmetic block admits an exact Partial-Gauss Kloosterman Transfer