{"id":"4d60a60c-587d-496c-ad2a-44eb57a7b7b9","arxiv_id":"2509.01442","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Quantum Brush is an open-source painting tool whose four brushes run quantum circuits on real NISQ hardware and use the measured quantum states to recolor the canvas.","lead":"Quantum Brush is a free painting app that turns each brush stroke into a small quantum circuit and uses the measured results to change the colors on the canvas. It gives artists a working way to paint with a real quantum computer and gives the rest of us a tangible feel for what quantum effects look like.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No classical-control baseline: visuals may be produced by hand-designed decoding maps rather than quantum dynamics","rationale":"The reader's weakest assumption identifies exactly this issue: no classical baseline showing the quantum source is responsible for the aesthetic. I agree with the conditional verdict because the tool itself is real, open-source, and executed on hardware; the concern is not that the tool is fake, but that the scientific framing overreaches. The load-bearing concern is the attribution of aesthetic novelty to quantum dynamics. This concern is concrete and testable via a simple control mode. I did not identify any internal inconsistency that would justify rejection. The scaling claim is secondary because the paper frames it as a soft guideline, and the implementation caps are pragmatic; however, even the core attribution lacks evidence. The proposed test is feasible given the existing open-source repository and would settle the question directly. Therefore the reader's CONDITIONAL verdict is appropriate; no change in the verdict is needed.","tokens_in":10803,"tokens_out":4841,"duration_ms":60269,"concrete_test":"Add a 'classical control' mode to the open-source Quantum Brush app: for each brush, replace the quantum circuit execution with a seeded pseudo-random number generator that produces measurement outcomes matching the device's single-qubit noise statistics (e.g., from IQM calibration data), while keeping the entire encoding/decoding pipeline (App. A, Eq. 11, Eq. 19) identical. Then run a blinded perceptual study where artists or naive users view pairs of images (quantum-hardware vs. classical-control) generated from the same input and must choose which is 'quantum'. If discrimination does not exceed chance, the visual signature is not attributable to quantum dynamics.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that the tool's aesthetic output is 'shaped through quantum mechanics' rests on the unverified premise that the visual results are attributable to quantum dynamics rather than to the authors' hand-designed decoding maps. In every brush, the quantum circuit output is reduced to a small set of observables: Heisenbrush uses only the global magnetization ⟨Mz⟩(t) (Eq. 10) mapped by Eq. 11; Collage uses Pauli expectation values (Eq. 18) mapped by the ad hoc Eq. 19; Aquarela and Smudge use single-qubit Bloch angles reconstructed via Eq. A4. These maps are deterministic, nonlinear functions chosen without a stated design principle or comparison against classical alternatives. The paper provides no control experiment in which the quantum circuit is replaced by a classical random process (same encoding/decoding, same measurement statistics) to show that the distinctive visuals disappear. Additionally, all demonstrations are limited to 10 qubits (Heisenbrush) or fewer, so exact classical simulation of the circuits is trivial; the Sec. II claim of a scaling regime requiring a QPU is asserted but never exercised. Without a baseline, the unique contribution of the quantum hardware to the aesthetic remains unsupported, even though the tool demonstrably runs and produces images.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"Quantum Brush is an open-source digital painting application that maps brush strokes and selected regions to small quantum circuits. Four brushes are presented: Aquarela (single-qubit HL encoding with an ancilla-driven color-steering interaction), Heisenbrush (Trotterized 1D Heisenberg evolution whose magnetization is mapped to color), Smudge (an amplitude-damping/pumping cascade built from a shared ancilla), and Collage (an asymmetric 1→2 quantum cloning circuit whose Pauli expectation values are mapped back to singular values used in image reconstruction). The authors show results from a noiseless simulator and from IQM's Sirius device, without error mitigation, and include an artist's qualitative account of using the tool. The paper's central claim is that the visual output is genuinely shaped by quantum dynamics, not merely by classical post-processing of measurement statistics.","tokens_in":11137,"tokens_out":6216,"duration_ms":80928,"significance":"If the central claim is supported, the paper is a useful contribution to quantum generative art: it provides a usable open-source tool, concrete circuits for four different quantum effects, hardware demonstrations on current NISQ devices, and a thoughtful account of artistic workflow. The paper is also honest about many of its constraints, such as the 10-qubit cap and the deliberate omission of error mitigation. However, the manuscript's own stated criterion--that the algorithms must admit a classically intractable scaling regime--is not met by any presented example, and, more importantly, the attribution of the aesthetic to quantum dynamics is not tested against classical alternatives. The strengths are the open-source release, the clear circuit diagrams, and the fact that genuine hardware executions are shown; the weakness is the absence of a control experiment that isolates the quantum contribution from the hand-designed decoding maps.","major_comments":[{"comment":"The paper's central claim that the art is 'shaped through quantum mechanics' (Introduction) is not isolated from the hand-designed decoding maps. Each brush reduces the quantum output to a few expectation values and then applies a deterministic, nonlinear classical map: Eq. (11) for Heisenbrush, Eq. (19) for Collage, and Eq. (A4) for Aquarela/Smudge. These maps were chosen ad hoc and the paper provides no control experiment in which the quantum circuit is replaced by a classical random process with the same encoding/decoding and the same measurement statistics. Without such a baseline, the reported differences between simulator and Sirius, and between brushes, could be produced by the decoding functions rather than by quantum dynamics. I request a control: rerun the key examples with random Bloch vectors or random expectation values (e.g., uniformly sampled on the Bloch sphere, or classi","section":"Sec. II (Aquarela, Heisenbrush, Collage) and App. A"},{"comment":"The first guideline states that 'the algorithms must admit a scaling regime where classical simulation becomes intractable, requiring a true QPU.' The presented implementations do not enter that regime: Heisenbrush is capped at 10 qubits and 10 Trotter steps, Smudge uses N+1 qubits on small demonstrated examples, and Collage uses three qubits. All of these circuits are trivially simulable on a classical computer. No analysis is given of how large N must be, what the classical simulation cost would be, or why the NISQ demonstrations are relevant to the intractable regime. Since the only hardware runs are at classically easy sizes, the stated reason for requiring a QPU is unsupported. Please either provide a concrete scalability analysis (resource estimates and, ideally, a larger demonstration) or explicitly remove this constraint from the claimed design guidelines.","section":"Sec. II, 'soft guidelines', and Sec. II.B.3"},{"comment":"The reconstruction formula tan θ = sqrt(⟨X⟩²+⟨Y⟩²)/⟨Z⟩ is not valid as written for states with ⟨Z⟩<0. Under the mapping θ = π L, any color with luminosity L>0.5 gives a negative cos θ, and the principal arctangent returns θ in the wrong quadrant. Unless the authors use atan2 or an explicit branch correction, the reconstructed luminosity will be wrong or ambiguous for dark colors. This is load-bearing for Aquarela and Smudge, whose final colors are reconstructed using Eq. (A4). The text should state the branch handling and verify that the displayed dark tones are not artifacts of this ambiguity.","section":"App. A, Eq. (A4)"},{"comment":"The Collage decoding map is specified only by two endpoint properties (perfect recovery for a unit Bloch vector pointing along log S, and uniform singular values for a maximally mixed state). The paper does not provide a design principle for the interpolation, nor a sensitivity check. Since the visual outcome of the Collage brush is largely determined by this equation, the claims about 'fidelity' and 'no-cloning' being visualized should be backed by at least a brief rationale for Eq. (19), or by a comparison with simpler classical interpolation schemes. This issue is related to the missing classical-control experiment in Major Comment 1, but Eq. (19) deserves explicit justification on its own.","section":"Sec. II.D.3, Eq. (19)"}],"minor_comments":[{"comment":"The notation 'mod 1' is used for all three HSL components. For hue, modular arithmetic is natural; for saturation and luminosity, a component-wise modulo creates discontinuities at the boundary that are not discussed. Please clarify whether the modulo is applied component-wise and how hue wrap-around vs. luminosity/saturation clipping are handled.","section":"Sec. II.B.3, Eq. (11)"},{"comment":"The artist's experience is qualitative and valuable, but it would be strengthened by a short description of the actual parameter ranges used (brush strengths, stroke counts, number of hardware shots) and how many runs were discarded before a satisfying result was obtained. This would help readers calibrate the reproducibility of the workflow.","section":"Sec. IV and Fig. 11"},{"comment":"The circuit diagram and equations are mostly consistent, but the sign conventions in Eqs. (4)-(8) should be stated more explicitly (e.g., Rxx(−Δt)=e^{iΔtXX/2} while Rx(Δt)=e^{-iΔtX/2}). A reader may otherwise wonder whether the overall Trotter step has the correct sign. This is a presentation issue, not a correctness issue.","section":"Sec. II.B.3 and Fig. 6"},{"comment":"The description of the Stroke Manager says 'supports running each brush concurrently' and snapshots are used, but it is not clear whether strokes from different brushes can be interleaved or whether the snapshot is taken per stroke or per brush. Please clarify the workflow.","section":"Sec. III"},{"comment":"The concluding statement that quantum dynamics provide 'an artistic advantage over machine learning approaches' is not supported by the evidence in this paper. I recommend softening this claim or adding a more nuanced discussion, since the paper does not compare against any machine-learning-based painting tool.","section":"Sec. V"}],"recommendation":"major_revision","confidential_remarks":"This manuscript is best judged as an applications/creative-computing contribution rather than a quantum-information result. The circuit implementations appear coherent and the open-source tool is a real contribution. The main load-bearing gap is methodological: the visual attribution to quantum dynamics is not tested against classical controls, and the paper's own intractability guideline is not met by the presented demonstrations. Both are fixable within the manuscript's scope. The self-citations (refs [7,8]) are background and not problematic. The comment about machine learning flattening artistic expression is unsupported and should be softened."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Brief take: this paper delivers a working open-source painting tool and shows it running on IQM's Sirius. That part is real. The four brush circuits—Aquarela, Heisenbrush, Smudge, Collage—are clearly described, the visuals are shown for both simulation and hardware, and the authors do not overclaim the hardware results. The open-source app and the artist's section add credibility; you can go run it.\n\nWhat is genuinely new: the specific mapping from brush strokes to quantum circuits and the integration into an interactive canvas. The underlying primitives (Trotter, amplitude damping, asymmetric cloning) are textbook, and the paper does not pretend otherwise. As an outreach and artistic tool, it is a solid contribution with real artifacts.\n\nWhere it is soft: the central language says the output is 'shaped through quantum mechanics,' but there is no classical control. The decoding maps—Eqs. 11, 19, A4—are hand-designed nonlinear functions that turn a few expectation values into colors. Without a baseline where the same maps are fed by classical random numbers or a classically simulated equivalent, you cannot tell how much of the aesthetic comes from the quantum state rather than from the map. That is a genuine gap, and it is fixable.\n\nSecond soft spot: the brushes are all small. Heisenbrush caps at 10 qubits, Collage uses 3, and time steps are limited to 10. Exact simulation is trivial. The paper's own guideline about an intractability regime (Sec. II) is asserted, not demonstrated, and the paper even says each brush may not satisfy all constraints. That should be stated more plainly, or the guideline dropped.\n\nFinally, the color maps are justified only by 'artistic outcomes.' That is fine for an art paper, but it makes the scientific claim about quantum aesthetics harder to evaluate. Tomography uncertainties or a comparison against a classical color-mixing baseline would help.\n\nOverall: the tool is real, the writing is clear, and the claims are mostly honest. The missing baseline is the main scientific weakness, but it does not sink the paper's value as a working system and outreach artifact. I would send it to peer review; a referee should ask for the classical control and a softening of the scaling claim, not for a rewrite.","headline":"A working open-source quantum painting tool with honest hardware runs, but the soft spot is the missing classical baseline for attributing the aesthetics to quantum dynamics.","tokens_in":11611,"tokens_out":1744,"would_cite":false,"duration_ms":18523,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Quantum Brush is an open-source digital painting tool that turns strokes into quantum circuits and maps the measured states back to colour, so the visuals are shaped by real quantum dynamics.","keywords":["quantum art","digital painting","NISQ hardware","quantum algorithms","HSL colour encoding","Bloch sphere","no-cloning theorem","open-source tool"],"falsifier":"A reader could take one of the brush circuits, replace the quantum hardware with a classical pseudo-random number generator that feeds the same decoding equations (Eq. A4 and the brush-specific maps, Eqs. 11 and 19), and compare the resulting images side by side; if the classical-random images are indistinguishable from the hardware images, the claim that the visuals are shaped by quantum dynamics would be falsified.","tokens_in":10704,"feed_emoji":"🎨","tokens_out":8803,"duration_ms":85604,"temperature":0.7,"pith_summary":"Quantum Brush is an open-source digital painting tool that turns each brush stroke into a quantum circuit and runs that circuit on a real quantum processor. Four brushes are presented, each built on a different quantum effect: entanglement-based colour blending, time evolution of a Heisenberg spin chain, information erasure through an amplitude-damping channel, and the no-cloning theorem realised by an asymmetric quantum cloning circuit. After the circuit runs, the qubits are measured and the resulting expectation values are decoded into new hue and luminosity values, repainting the pixels under the stroke. The authors deliberately omit error mitigation, so the noise of today's quantum hardware becomes part of the aesthetic. The paper's central claim is that this pipeline produces visual output that is genuinely shaped by quantum dynamics, giving artists a new medium in which the physics of the device participates in the image.","feed_headline":"Quantum Brush paints with quantum circuits, not classical filters","feed_subtitle":"A new open-source tool maps hue and light onto qubits, then lets quantum hardware shape the art.","key_machinery":"The load-bearing object is the HSL-to-qubit encoding |(φ,θ)⟩ = Rz(φ)Ry(θ)|0⟩ (Eq. A1), which puts hue and luminosity on the Bloch sphere so that quantum gates act directly on colour. On top of this encoding, four circuits carry the argument: the Aquarela circuit's sequential controlled rotations with a brush ancilla; the Heisenbrush's Trotterised Heisenberg evolution (using first-order Trotter-Suzuki decomposition with XX, YY, ZZ and field rotations); the Smudge circuit's shared-ancilla amplitude-damping/pumping channel; and the Collage circuit's universal asymmetric quantum cloning built from four CNOT gates plus a state-preparation unitary. Each circuit's measured Pauli expectation values","core_discovery":"The central discovery is a working quantum-native colour space and a set of four brush circuits that operate on it. In this colour space, hue and luminosity of an image region are mapped to the polar and azimuthal angles of a single qubit via |(φ,θ)⟩ = Rz(φ)Ry(θ)|0⟩, leaving saturation untouched. Each brush then applies a different quantum mechanism: Aquarela entangles the brush colour with the canvas colours through sequential controlled rotations; Heisenbrush simulates the time evolution of a one-dimensional Heisenberg model and uses the measured magnetisation to shift the HSL values; Smudge feeds all stroke colours through a shared ancilla that implements an amplitude-damping/pumping chan","pith_inferences":["Because the paper does not include a baseline run in which the same decoding maps are fed with classically generated random numbers, the claim that the aesthetics come from quantum dynamics rather than from the hand-designed maps remains an inference the reader must test for themselves.","The demonstrated implementations use at most 10 qubits (Heisenbrush) and 3 qubits (Collage), so the soft guideline of a classically intractable scaling regime is not exercised in the shown results; a testable extension would be to scale both brushes to the maximum circuit width a current device supports and compare visual output.","A direct extension would be to swap the quantum circuit for a classical circuit that applies the same unitary evolution using matrix multiplication and then samples from the same final distribution—comparing the resulting images would isolate the contribution of the hardware's noise and the measurement randomness from the ideal quantum dynamics.","The HSL-to-qubit mapping could be reused by other quantum generative art tools, not only for brushes but for any effect that needs to apply quantum logic to colour fields."],"forward_implications":["If the central claim holds, the Quantum Brush is a working open-source tool in which the visual output is shaped by the quantum circuits actually executed, not by classical image filters.","Each brush demonstrates a concrete route from a quantum principle to a visual effect: Aquarela maps entanglement to colour blending, Heisenbrush maps time-evolved magnetisation to colour shifts, Smudge maps information erasure to cascading colour changes, and Collage maps the no-cloning trade-off to copy fidelity.","Because the authors deliberately run without error mitigation, the hardware's intrinsic noise becomes a deliberate compositional parameter, available to any artist using the tool.","The snapshot-based stroke manager lets a user re-run the same stroke until a satisfying outcome, meaning the stochasticity of quantum computing is harnessed inside an interactive workflow.","The quantum-native colour space, in which saturation is deliberately left unchanged, offers a principled way to apply quantum operations to colour without RGB gamut overflow."],"supporting_citations":[{"why":"Supplies the first-order Trotter-Suzuki decomposition used to approximate the Heisenberg time evolution in the Heisenbrush circuit.","marker":"[18]"},{"why":"Supplies the original Trotter product formula that justifies the decomposition applied in the Heisenbrush implementation.","marker":"[19]"},{"why":"Provides the amplitude-damping/pumping channel circuit that the Smudge brush uses to push canvas colours toward |0⟩ or |1⟩.","marker":"[21]"},{"why":"Supplies the universal asymmetric quantum cloning protocol (Pauli cloning) that the Collage brush implements to produce approximate copies.","marker":"[23]"},{"why":"Extends asymmetric quantum cloning to the general d-level setting and provides the formalism used to fix the cloning parameters in the Collage brush.","marker":"[24]"}],"fun_headline_variants":["Quantum Brush turns strokes into qubit circuits","Paint with qubits: new tool for NISQ art","Digital painting meets quantum hardware","Quantum brush: art generated by quantum algorithms","From stroke to qubit: a quantum painting tool"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The paper does not compare the quantum-hardware output against a baseline where the same colour-decoding maps are driven by classically generated random numbers, so the claim that the novel visuals come from the quantum dynamics rather than from the chosen decoding maps rests on an untested assumption.","fun_headline_variants_meta":{"raw":{"variants":["Quantum Brush turns strokes into qubit circuits","Paint with qubits: new tool for NISQ art","Digital painting meets quantum hardware","Quantum brush: art generated by quantum algorithms","From stroke to qubit: a quantum painting tool"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":8e-05,"raw_usage":{"total_tokens":703,"prompt_tokens":582,"completion_tokens":121,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":326,"completion_tokens_details":{"reasoning_tokens":52}},"tokens_in":326,"tokens_out":121,"duration_ms":2292,"temperature":1.0,"reasoning_tokens":52,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T12:31:58.230490+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A reader could take one of the brush circuits, replace the quantum hardware with a classical pseudo-random number generator that feeds the same decoding equations (Eq. A4 and the brush-specific maps, Eqs. 11 and 19), and compare the resulting images side by side; if the classical-random images are indistinguishable from the hardware images, the claim that the visuals are shaped by quantum dynamics would be falsified.","supporting_citations":[{"cited_title":"Lioret, Quantum Art (The Eurographics Association, 2016)","cited_arxiv_id":null,"evidence_quote":"Supplies the first-order Trotter-Suzuki decomposition used to approximate the Heisenberg time evolution in the Heisenbrush circuit."},{"cited_title":"Quantum computing inspired paintings: reinterpreting classical masterpieces","cited_arxiv_id":"2411.09549","evidence_quote":"Supplies the original Trotter product formula that justifies the decomposition applied in the Heisenbrush implementation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the amplitude-damping/pumping channel circuit that the Smudge brush uses to push canvas colours toward |0⟩ or |1⟩."},{"cited_title":"Baumgarten, Quantum jungle: Playful installation vi- sualizing quantum physics (2023)","cited_arxiv_id":null,"evidence_quote":"Supplies the universal asymmetric quantum cloning protocol (Pauli cloning) that the Collage brush implements to produce approximate copies."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Extends asymmetric quantum cloning to the general d-level setting and provides the formalism used to fix the cloning parameters in the Collage brush."}],"review_version":1}