The paper asks how several decode-and-forward relay nodes can jointly form a near-capacity channel code for a two-hop, half-duplex relay channel. It proposes a distributed irregular convolutional code (DIRCC): the source uses an irregular convolutional code concatenated with a unity-rate code and 4PSK, and K selected relays each re-encode only part of the decoded bits with a low-complexity component code (two repetition codes and a unit-memory convolutional code), so the destination receives one virtual irregular code that it decodes iteratively. Codes are designed by EXIT-chart curve matching
The paper asks what a coding-agent 'harness' is — the runtime that couples a large language model to the world through a loop, tools, context management, safety controls, orchestration and extension surfaces — and how production harnesses are actually built. It is a descriptive source-code study (not a benchmark) of eleven systems pinned to July 2026 releases (Claude Code, Codex CLI, Gemini CLI, Mistral Vibe, OpenHands, Aider, Mini-SWE-Agent, Hermes, Pi, OpenCode, OpenClaw), plus the Omnigent meta-harness as a contrast point. Contributions: a seven-subsystem reference anatomy with minimal and
The paper asks whether amorphous silicon can supply both intrinsic near-infrared light emission and a compact optical cavity for integrated photonics. Solution-synthesized amorphous silicon microspheres (about 500-1400 nm across) show measured room-temperature whispering-gallery-mode photoluminescence with sharp resonances spanning 800-1100 nm and a maximum measured quality factor of 522, whereas fractured spheres give only broad, weak emission. Finite-element simulations assign the peaks to TE and TM modes. The measured emission red-shifts with temperature (0.05 nm/K), with the surrounding re
Neutral-atom quantum processors start from optical tweezer arrays that are only about half filled, so atoms must be rearranged into defect-free patterns. The authors compute every intermediate hologram for a high-speed spatial light modulator analytically, as a superposition of elementary holograms (one phase gradient per trap, plus a quadratic term for axial motion), and linearly interpolate each trap's position, weight and optical phase between frames to keep phase continuity (the I-SEH protocol). On one GPU each hologram takes 0.284 ms, faster than the 0.696 ms SLM refresh, leaving time for
The paper asks how to give every crease of a self-folding origami robot its own fast, reversible, precisely controlled actuator instead of heating the whole structure at once. The authors introduce layer-by-layer direct ink writing, which embeds prefabricated flexible printed circuit boards (serpentine Joule heaters, thermistors, stiffeners, copper cooling zones) between printed liquid crystal elastomer layers at each hinge, plus a CAD workspace (OriCadLCE) that turns a crease pattern into print paths and circuit files. Measured results: the Flex-PCB hinge bends about 113 degrees at 100 C, clo
Question: how are adhesion molecules arranged in three dimensions on the microvilli (finger-like membrane protrusions) of blood stem/progenitor-like and leukemic cells, and does rolling under flow on E-selectin rearrange them? Contribution: the authors built a microfluidics-based imaging platform that sequentially combines total internal reflection fluorescence microscopy (TIRFM), which converts membrane-stain brightness in a ~120 nm evanescent field into a 3D height map of the cell's underside, with astigmatism-based 3D single-molecule localization microscopy (3D-SMLM, ~30 nm axial precision)
The paper asks whether the optimum phase-only triplicator grating profile (Gori et al.), which splits a beam into three equally intense diffraction orders with 92.6% total efficiency, can be applied to an arbitrary phase-only Fourier hologram. The authors pass the hologram phase through the triplicator function arctan[a cos(phi + cx)], showing analytically that the reconstruction contains three equal-energy terms: the target pattern in the +1 order, its inverted complex-conjugate in the -1 order, and a delta spot in the zero order, using a phase range of only 0.78 pi. Adding a second phase fun
The study asks how deliberately shaped filler particles in the poorly explored 300–1000 µm size range stiffen a soft silicone (PDMS) composite. The authors DLP-3D-printed cubes and four-armed tetrapods at two sizes (0.03–0.05 mm³ and 1 mm³), packed each at its maximum random packing fraction (35–56%, verified by micro-CT) into three PDMS matrices spanning a 6.4-fold stiffness range, and measured quasi-static compression of 51 specimens. Measured results: every filler stiffened the matrix; tetrapods reached up to ~13-fold stiffening and the highest packing-normalised modulus (44.9 MPa), outperf
Quantum key distribution (QKD) using the BB84 protocol over free-space optical (FSO) laser links needs a public classical channel so that sender and receiver can compare the polarization bases they used and sift the key; earlier QKD-FSO studies assumed this public channel was error-free. The paper models realistic, outage-prone wireless public channels and proposes NC-CQKD-FSO, in which several user pairs cooperate and use network coding (broadcast phases followed by cooperative phases carrying linear combinations of the users' messages) so that basis information survives link outages. It deri
The paper asks how to simulate, accurately and affordably, subluminal velocity-controlled spatio-temporal laser drivers, whose intensity peak moves at a chosen focal velocity independent of the envelope group velocity, for laser–plasma wakefield applications. Using OSIRIS quasi-3D particle-in-cell simulations, the authors build the pulses as superpositions of exact vacuum plane waves on the intersection of the light cone with a tilted spectral plane, give k-space sampling criteria that avoid spurious replicas, derive a paraxial spot-size scaling for near-resonant wake excitation that they chec
Soft robots need both compliant and stiff parts, but discrete soft–rigid junctions made by gluing, overmolding or commercial multi-material printing tend to fail under repeated deformation. The authors introduce hybrid epoxy–acrylate printing (HEAP): one resin cured in a dual-wavelength DLP printer, where 405 nm light polymerizes only the acrylate network (soft, about 0.4 MPa) and 365 nm light also cross-links the epoxy groups (rigid, about 1 GPa), with grayscale exposure giving intermediate hardness in one continuous network. In measured tensile tests, only 12.3% of HEAP soft–rigid specimens
The paper asks whether the function of quantum key distribution (QKD) can be obtained on standard free-space optical (FSO) links, without single-photon detectors or phase-stabilized local lasers. It proposes a protocol in which Alice sends subcarrier-intensity-modulated binary phase shift keying (SIM/BPSK) signals with a small modulation depth, and Bob detects them directly with an avalanche photodiode and two thresholds: readings beyond either threshold become key bits, readings in between are discarded, and Bob publicly announces only which time slots he kept. An eavesdropper placing an unau
The paper asks how short-lived particles such as muons and pions, often produced moving slower than light, can be accelerated to relativistic speeds before they decay. It proposes a laser wakefield scheme in which a subluminal light pulse (a space-time wave packet with focal velocity below c) drives a plasma wake, while a tailored plasma density profile continuously matches the wake's velocity to the accelerating beam, extending the dephasing length. The authors derive a theoretical model (Eq. 1) for the ideal density taper and validate it with OSIRIS particle-in-cell simulations, which show m
The paper asks how to build Kresling origami from thick material so that it still folds completely flat and still snaps between two stable states, given that thick panels are rigid (they cannot provide the stretching the Kresling's energy barrier needs) and pile up into jams when folded. It proposes a design framework: hinge types for thick panels (including double-pivot hinges that let panels fold flat against each other), trimming panel material so parts in each layer snake around each other and no point stacks more than the number of vertical folds plus one (four for the Kresling), and serp
Detailed smart-meter readings can reveal when individual appliances are used, and therefore when people are home and what they do. This paper asks how a home battery can mix its own power with grid power so that appliance switch-on and switch-off events become hard to detect in the metered trace. It introduces a selective privacy protection algorithm that splits the battery into logical energy quotas for chosen appliance groups, uses a Markov-chain one-step prediction of consumption, and extends three privacy metrics (variational distance, cluster similarity and regression). Using appliance-le
Projection 3D printers that cure liquid resin with light are limited by the fixed number of mirrors (2560×1600) on their digital micromirror device: shrinking the projected pixel for finer detail shrinks the printable area. The authors introduce segmented projection image translation (SPIN), which offsets the small objective lens sideways in the collimated 'infinity space' of the optics and orbits it around the optical axis, so the image is stepped to different positions on the build plate without moving the light engine or the resin platform. In their experimental printer this expanded the bu
The paper asks how continuous-variable quantum key distribution (CV-QKD) should correct the errors between Alice's and Bob's raw keys when blocks are short and the public classical channel used for side information is itself noisy and fading. It proposes four reconciliation schemes (Systems A–D): Polar codes protect the quantum transmission, with the Polar frozen-bit indices sent as side information, and LDPC codes protect that side information over a Rayleigh-fading classical channel ("double protection"); reverse- and direct-reconciliation protocols split Polar and LDPC decoding between Alic
The paper asks whether a near-term quantum algorithm, the quantum approximate optimization algorithm (QAOA), can perform maximum likelihood (ML) detection of binary (BPSK) symbols sent over a multiple-input multiple-output (MIMO) wireless channel, a task whose exact classical solution grows exponentially (2^Mt candidates) with the number of transmit antennas. The authors encode each transmitted bit as a qubit and map the ML objective onto an Ising problem Hamiltonian whose ground state is the ML solution; they analyse the adiabatic evolution between an easy initial Hamiltonian and this problem
The paper asks how continuous-variable quantum key distribution (CV-QKD) can reconcile Alice's and Bob's noisy correlated data into an identical key when the public classical channel used for side information is itself noisy or fading, and when the error-correcting code is not a syndrome-based LDPC code. It proposes a codeword-based reconciliation (System D): Bob sends an FEC codeword through the quantum-equivalent channel, Alice sends a separately FEC-protected codeword through the classical channel, each decodes the other's, and both take the modulo-2 sum as the key. This protects both chann
The paper asks how to better reduce the computational error of variational quantum algorithms running on noisy intermediate-scale quantum computers without prior knowledge of the noise channels. It generalizes permutation-based error mitigation (virtual distillation) into 'permutation filters': weighted combinations of the outputs of virtual distillation circuits of different orders, analogous to finite impulse response filters acting on the eigenvalue spectrum of the noisy output state. The authors propose a filter design algorithm and prove the design problem is invex, so projected gradient
The paper asks how a mobile phone can cleanly receive its own data from a base station when the radio signal arrives as several delayed, randomly fading echoes (a dispersive Rayleigh fading channel) and shares the airwaves with other users' signals. The authors propose STC-DS-RR, a downlink CDMA scheme that combines a 'double-spreading' code (a short 4-chip Walsh code to separate users, with each Walsh chip further spread by the same 8-chip random code), Alamouti two-antenna space-time block coding, a simple Rake receiver and turbo convolutional coding. They argue analytically that, in slowly
An iPaper is an interactive companion website that helps readers explore a paper’s ideas through interactions such as changing a parameter, comparing outcomes, or inspecting a mechanism. It links back to the original research and does not replace the publication.
How can researchers use iPaper.ai?
Share an iPaper alongside your publication, link to it from your lab website, or use it in a lecture or conference talk. Choose a paper whose mechanism, model, or comparison benefits from hands-on exploration.
Which papers can I turn into an iPaper?
Start with an openly accessible research article, DOI, or full-text PDF link that you are entitled to have processed under its licence, permission, or applicable law. iPaper.ai needs the full paper and does not bypass paywalls or sign-in requirements.
Do I need to code, and what do I receive?
You do not need to write code. iPaper.ai generates a hosted iPaper with a shareable link, source attribution, and a downloadable copy of your website. Review the explanation and interactions against your paper before sharing them.
Is iPaper.ai free, and are my iPapers public?
iPaper.ai is currently free to use. Your draft stays private while you review it and request changes. Choose private access, an unlisted link anyone can open, or public visibility eligible for the gallery. Each release points to a specific version.
Does an iPaper reproduce my experiments?
Interactions may use simplified educational models rather than reproduce the paper’s experiments. Review their assumptions, limits, and source references alongside the original research before sharing.
Built with researchers
Help shape what’s next.
Try an experience. Share it with a colleague. Tell us what made the science clearer—and what didn’t.