9. Asking your library
A summary covers one search. To ask questions across everything you have collected, mosaic builds a vector index: each paper becomes one or more embeddings, lists of numbers that place texts of similar meaning close together. A question is embedded the same way, the closest papers are retrieved, and the model answers from them, citing them. This is retrieval-augmented generation (RAG), and here it runs on your machine.
It needs the rag extra (sqlite-vec) and an embedding model next to the chat model of chapter 8:
bash
ollama pull qwen3-embedding:0.6b
mosaic config --embedding-model qwen3-embedding:0.6b \
--embedding-base-url http://localhost:11434/v1 --embedding-api-key ollamaBuilding the index
console
$ mosaic index
Warning: full_text_index is enabled but pymupdf is not installed.
Run: pipx inject mosaic-search pymupdf
Falling back to metadata-only indexing.
Indexing 107 papers…
Indexed 107 new paper(s). 0 already indexed.mosaic index embeds every cached paper not indexed yet; run it again after new searches and it adds only the new ones (mosaic config --rag-auto-index does that after every search). With PyMuPDF installed (Installation), papers whose PDF is in the library are indexed by their full text, in overlapping chunks; the others by title and abstract, as here. After changing the embedding model, rebuild with mosaic index --reindex: the vectors of two models do not mix.
Searching by meaning
console
$ mosaic search "electrons crossing the magnetic field" --semantic
Searching local vector index for 'electrons crossing the magnetic field'…
# Title Authors Year DOI Source OA PDF Cited Sim.
────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
1 Cross-field chaotic Debraj 2019 10.48550/arXiv arXiv yes ✓ – 0.50
transport of electrons by Mandal et .1907.08407
$\vec E \times \vec B$ al.
electron d…
2 Rôle des instabilités Alexandre 2006 – HAL yes ✓ – 0.49
électroniques de dérive Ducrocq
dans le transport
électronique du…
3 Anomalous cross field C. 2006 10.1063/1.2360 OpenAlex no – 99 0.49
electron transport in a Boniface 182
Hall effect thruster et al.
4 Cross-field chaotic D. Mandal 2019 10.1063/1.5134 arXiv yes ✓ – 0.49
transport of electrons by et al. 148
ExB electron drift
instability in…
5 Three-dimensional Yin-Jian 2025 10.1017/S00223 Semantic yes ✓ 1 0.49
particle-in-cell Zhao et 77826102189 Scholar
simulations on Hall al.
thruster electron drift …
6 Theory for the anomalous Trevor 2016 10.1063/1.4948 OpenAlex no – 151 0.49
electron transport in Hall Lafleur et 496
effect thrusters. II. al.
Kinet…
7 Electric field induced Srimayi 2025 10.48550/arXiv arXiv yes ✓ – 0.49
Berry curvature dipole and Korrapati .2510.20237
non-linear anomalous Hall et al.
eff…
8 Modélisation Vivien 2017 10.70675/49b79 HAL yes ✓ – 0.49
bidimensionnelle de la Croes 74ezcf6dz42b1z
décharge plasma dans un 9817z0df2db1d5
propulseur de Hall 902
9 State Estimation of Troyetsky, 2023 10.1109/icops4 Crossref no – – 0.48
Anomalous Electron D. et al. 5740.2023.1048
Transport in a Hall-Effect 1135
Thruster Disc…
10 Ionization instability and Potrivitu 2022 10.21203/rs.3. Europe PMC no – – 0.48
turbulence in the plume of G & Xu S. rs-1806238/v1
sub-ampere hollow cathode…
10 result(s)--semantic ranks the indexed papers by the meaning of the query, so a paper can rank high without sharing its words. Sim. is the similarity of each paper's closest chunk. --downloaded-only keeps the papers whose PDF is in your library.
Asking a question
console
$ mosaic ask "What causes anomalous electron transport in Hall thrusters?" --show-sources
───────────────────────────────────────────── mosaic ask · mode: synthesis ─────────────────────────────────────────────
Sources retrieved (10):
[1] Optical study of anomalous electron transport in a laboratory Hall thruster — Meezan, N., Hargus, Jr., W. (1999)
[2] Anomalous cross field electron transport in a Hall effect thruster — C. Boniface, Laurent Garrigues (2006)
[3] State Estimation of Anomalous Electron Transport in a Hall-Effect Thruster Discharge — Troyetsky, D., Greve, C.
(2023)
[4] Near-Wall Pathways of Anomalous Electron Transport in Hall Thrusters Revealed by 3D PIC Simulations — Zhe Liu,
Zhongping Zhao (2026)
[5] Optimal Experimental Design for Inferring Anomalous Electron Transport in a Hall Thruster — Allen, Madison G.,
Marks, Thomas A. (2024)
[6] Comparison between ad-hoc and instability-induced electron anomalous transport in a 1D fluid simulation of
Hall-effect thruster — Roberto Martorelli, Trevor Lafleur (2019)
[7] Theory for the anomalous electron transport in Hall effect thrusters. II. Kinetic model — Trevor Lafleur, Scott
Baalrud (2016)
[8] Nonlinear simulations and anomalous transport in hall thruster plasma — Koshkarov, O., Pombo, W. Frias (2016)
[9] Anomalous transport induced by sheath instability in Hall effect thrusters — F. Taccogna, Savino Longo (2009)
[10] Cross-field chaotic transport of electrons by ExB electron drift instability in Hall thruster — D. Mandal, Y.
Elskens (2019)
What Causes Anomalous Electron Transport in Hall Thrusters? – A Synthesis of the State of the Art
1. The Core Problem
In Hall effect thrusters (HETs), electrons must cross the radial magnetic field to reach the anode, yet classical
collisional diffusion predicts a cross-field mobility that is orders of magnitude too small to sustain the observed
discharge current. This "anomalous" (or Bohm-like) transport has long been a central obstacle to first-principles
modelling of Hall thrusters. The papers collected here span optical diagnostics, kinetic and fluid modelling, kinetic
theory, sheath physics, and data-driven inference, and together they paint a picture in which azimuthal electrostatic
instabilities are the dominant driver, while wall interactions and sheath physics provide important secondary
contributions that shape the spatial distribution of the cross-field current.
2. Main Approaches and Methods
Several complementary methodological strands are represented:
• Optical and interferometric diagnostics. Meezan et al. [1] performed one of the earliest dedicated optical studies of
anomalous electron transport in a laboratory Hall thruster, establishing the need for spatially resolved plasma
measurements. Boniface et al. [2] used Fabry-Pérot interferometry on a 5 kW thruster to compare measured ion-velocity
profiles against hybrid-model predictions, allowing them to infer the cross-field conductivity inside and outside the
channel simultaneously.
• Kinetic particle-in-cell (PIC) simulations. Lafleur, Baalrud and Chabert [7] used PIC simulations to demonstrate a
direct correlation between anomalously high cross-field electron transport and the presence of azimuthal
electrostatic instabilities. Liu, Zhao and Zhao [4] performed instability-resolving, fully three-dimensional PIC
simulations with self-consistent dielectric wall charging, secondary electron emission, Monte Carlo ionisation, a
continuum neutral-gas model, and an open near-plume outflow. Koshkarov et al. [8] explored nonlinear simulation
effects on anomalous transport in Hall thruster plasmas.
• Fluid and hybrid models with instability-derived transport. Martorelli et al. [6] compared an ad-hoc
anomalous-transport model against an instability-induced closure within a one-dimensional fluid simulation, testing
whether the instability picture can reproduce results typically obtained with empirical conductivity prescriptions.
Taccogna et al. [9] performed numerical experiments specifically isolating the contribution of a lateral sheath
instability to the anomalous axial current.
• Kinetic theory and wave-particle modelling. Lafleur et al. [7] derived a kinetic theory of the enhanced electron
scattering rate driven by the azimuthal electron drift instability, predicting a cross-field mobility in good
agreement with experiment. Mandal et al. [10] modelled the anomalous transport as arising from chaotic electron
dynamics in the fluctuating electrostatic fields of the electron cyclotron drift (E×B) instability endemic to HETs.
• Data-driven and experimental-design approaches. Troyetsky et al. [3] applied state-estimation techniques to infer
anomalous electron transport from discharge measurements, while Allen et al. [5] formulated an optimal experimental
design problem for inferring the transport coefficient with fewer measurements and higher confidence.
3. Key Findings
3.1 Azimuthal Electrostatic Instabilities as the Primary Driver
The strongest and most recurring finding across the literature is that azimuthal E×B / electron drift instabilities are
the principal mechanism generating anomalous cross-field electron transport. Lafleur et al. [7] showed via PIC
simulations that the large azimuthal electron drift in HETs drives a strong instability that quickly saturates through
ion-wave trapping and wave convection, yielding an enhanced mobility many orders of magnitude above classical diffusion.
Mandal et al. [10] independently attributed the anomalous transport to chaotic electron dynamics in the fluctuating
fields of the electron cyclotron drift instability, demonstrating that wave-particle interaction raises the
perpendicular electron temperature (T⊥/T∥ ~ 4) and produces a mean axial electron velocity of the correct order of
magnitude relative to experiment. Liu et al. [4] confirmed in fully 3D PIC simulations that the transport is
instability-driven, extracting the net correlation term ⟨n_e E_y⟩ from strongly oscillatory three-dimensional fields.
3.2 Instability-Induced Transport Reproduces Ad-Hoc Models
Martorelli et al. [6] found reasonable agreement between the friction force and anomalous heating produced by a
conventional ad-hoc transport model and those arising from an instability-derived closure in a 1D fluid simulation. This
suggests that the long-used empirical Bohm-like conductivity may be a macroscopic shadow of the underlying
micro-instability physics, and it paves the way for self-consistent instability-driven closures in fluid solvers.
3.3 Spatial Non-Uniformity and Near-Wall Pathways
A particularly notable result comes from Liu et al. [4], who showed that anomalous electron transport is not distributed
uniformly across the channel cross-section. Instead, it self-organises into persistent near-wall pathways connected to
the near-exit region. Comparing conducting-wall, ceramic-wall-with-secondary-emission, and open-outflow closures, they
demonstrated that this near-wall transport topology is robust, while the boundary treatment mainly redistributes the
detailed strength of the pathway and its coupling to the exit and near-plume region. This provides a previously
unresolved spatial picture of where the instability-driven transport actually occurs.
3.4 The Role of Walls and Sheath Instability
Boniface et al. [2] challenged the long-standing assumption that Bohm anomalous conductivity dominates outside the
channel while electron-wall (near-wall) conductivity prevails inside; their interferometric comparisons suggested that
one and the same mechanism is responsible for anomalous transport in both regions. Taccogna et al. [9] identified a
distinct additional mechanism: a lateral sheath instability arising from the negative differential resistance of the
floating-wall I–V characteristic due to high secondary electron emission. Their numerical experiments indicated this
sheath instability alone can account for more than 80 % of the experimentally measured anomalous axial current. Liu et
al. [4] showed that secondary electron emission and dielectric wall charging affect the detailed strength and
exit-region coupling of the near-wall transport pathways, even though the topological structure remains robust.
3.5 Scaling and Plasma-Property Dependence
Lafleur et al. [7] emphasised that the enhanced mobility predicted by their kinetic model is a strong function of local
plasma properties (density, temperature) as well as B₀, and therefore does not in general follow a simple 1/B₀ or 1/B₀²
scaling. This has implications for scaling HET performance across power levels and magnetic-field configurations.
3.6 Inference and Experimental Design
Troyetsky et al. [3] and Allen et al. [5] address the practical challenge of inferring the anomalous transport
coefficient from limited experimental data, employing state-estimation and optimal-experimental-design frameworks
respectively. These contributions are complementary to the physics modelling, providing systematic methods to extract
the effective transport from measurements with quantified uncertainty.
4. Areas of Consensus
Among the ten papers, several points of agreement emerge:
• The classical collisional (Spitzer) conductivity is insufficient by several orders of magnitude, and some additional
mechanism is required [2][4][6][7][10].
• Azimuthal electrostatic (E×B) instabilities are the leading candidate for the anomalous transport mechanism
[4][6][7][10].
• The problem is inherently three-dimensional and spatially non-uniform, and low-dimensional or purely local models
capture only part of the physics [4][6][7].
• Wall physics (secondary emission, sheath structure) modifies but does not eliminate the instability-driven transport
[2][4][9].
• Experimental validation (Fabry-Pérot, optical, and electrical measurements) remains essential for constraining
models, and the community is moving toward more targeted, design-of-experiment approaches [1][2][3][5].
5. Notable Disagreements and Open Debates
• Single mechanism vs. multiple mechanisms. Boniface et al. [2] argued that one mechanism (implicated by their
hybrid-model comparison) explains transport both inside and outside the channel, explicitly challenging the
conventional Bohm-in / wall-conductivity-out partition. Liu et al. [4] and Taccogna et al. [9], by contrast,
highlight distinct near-wall pathways and a separate sheath-instability contribution, respectively, suggesting that
the physics is more compartmentalised than a single Bohm-like coefficient implies. Whether these are genuinely
distinct mechanisms or different spatial manifestations of the same underlying E×B instability remains an open
question.
• The spatial topology of transport. The older picture (implicit in ad-hoc models [6] and in the Bohm-vs.-wall
dichotomy of [2]) treated anomalous conductivity as a fairly homogeneous bulk property. The 3D PIC results of Liu et
al. [4] show a strongly structured, near-wall, exit-connected topology, which is not captured by 1D fluid or simple
kinetic-theory closures. How well current fluid or hybrid codes can reproduce this spatial structure is an active
area of development.
• Scaling laws. The expectation of a simple 1/B₀² (Bohm) scaling is contradicted by the kinetic-theory result of
Lafleur et al. [7], who show a strong, non-trivial dependence on local density and other plasma parameters. This has
practical consequences for thruster scaling that is not yet fully resolved.
• Adequacy of 1D / fluid closures. While Martorelli et al. [6] demonstrate that instability-induced closures can match
ad-hoc models in 1D, the 3D PIC work of Liu et al. [4] and the sheath-focused study of Taccogna et al. [9] both
suggest that important physics (azimuthal mode structure, sheath instability, wall-boundary coupling) is lost in any
purely one-dimensional treatment. The community is therefore transitioning toward fully kinetic or at least
multi-dimensional fluid descriptions, but at the significant computational cost noted in [4].
• The relative weight of the lateral sheath instability. Taccogna et al. [9] claim that the sheath instability
contributes more than 80 % of the anomalous axial current, a strong claim that sits in tension with the
instability-driven, channel-wide picture of [4][6][7][10]. Reconciling the magnitude of the sheath contribution with
the global E×B instability contribution is not yet settled in the literature presented here.
6. Summary
The state of the art, as reflected in the cited works, converges on the view that anomalous electron cross-field
transport in Hall thrusters is driven primarily by azimuthal electrostatic (E×B / electron drift) instabilities, whose
interaction with the electron population enhances the effective perpendicular mobility far beyond classical expectations
[4][6][7][10]. The transport is spatially organised into near-wall pathways rather than being a uniform bulk phenomenon
[4], and wall-related physics—secondary electron emission, dielectric charging, and lateral sheath instability—modifies
the strength and distribution of the anomalous current [2][4][9]. Empirical Bohm-like closures remain useful engineering
tools and can be partially justified by the underlying instability physics [6], but a fully predictive, spatially
resolved, self-consistent description still requires three-dimensional kinetic or high-fidelity fluid modelling [4][7].
Open questions persist regarding the exact partition of responsibility between channel instabilities and wall sheath
effects, the correct scaling with magnetic field and plasma density, and the most efficient experimental protocols for
inferring the transport coefficient in situ [3][5].
References
[1] Optical study of anomalous electron transport in a laboratory Hall thruster — Meezan, N., Hargus, Jr., W.,
Schmidt, D. et al., 1999
[2] Anomalous cross field electron transport in a Hall effect thruster — C. Boniface, Laurent Garrigues, Gerjan J. M.
Hagelaar et al., 2006
[3] State Estimation of Anomalous Electron Transport in a Hall-Effect Thruster Discharge — Troyetsky, D., Greve, C.,
Tsikata, S. et al., 2023
[4] Near-Wall Pathways of Anomalous Electron Transport in Hall Thrusters Revealed by 3D PIC Simulations — Zhe Liu,
Zhongping Zhao, Yinjian Zhao, 2026
[5] Optimal Experimental Design for Inferring Anomalous Electron Transport in a Hall Thruster — Allen, Madison G.,
Marks, Thomas A., Eckels, Joshua et al., 2024
[6] Comparison between ad-hoc and instability-induced electron anomalous transport in a 1D fluid simulation of
Hall-effect thruster — Roberto Martorelli, Trevor Lafleur, Anne Bourdon et al., 2019
[7] Theory for the anomalous electron transport in Hall effect thrusters. II. Kinetic model — Trevor Lafleur, Scott
Baalrud, Pascal Chabert, 2016
[8] Nonlinear simulations and anomalous transport in hall thruster plasma — Koshkarov, O., Pombo, W. Frias, Smolyakov,
A. I. et al., 2016
[9] Anomalous transport induced by sheath instability in Hall effect thrusters — F. Taccogna, Savino Longo, M.
Capitelli et al., 2009
[10] Cross-field chaotic transport of electrons by ExB electron drift instability in Hall thruster — D. Mandal, Y.
Elskens, N. Lemoine et al., 2019mosaic ask retrieves the ten papers closest to the question (-n changes it), sends their best-matching text to the model, and prints the answer with [n] citations; --show-sources lists the papers first. Four modes shape the answer:
--mode | Answer |
|---|---|
synthesis (default) | the state of the art on the question |
gaps | open problems, contradictions, missing evidence |
compare | methods, data and results side by side |
extract | one structured entry per paper: task, method, dataset, metric, result |
--query, --from refs.bib and --year restrict the papers it may use; --output answer.md (or .json) saves the answer with its sources.
A conversation
bash
mosaic chatmosaic chat keeps asking: each question retrieves its own papers, and the last exchanges go back to the model as context. Inside, /mode gaps switches the mode, /sources lists the papers behind the last answer, /clear forgets the conversation and /quit ends it.
What you learned
Embedding settings, mosaic index, search --semantic, mosaic ask and its modes, mosaic chat. Guide: Local RAG; reference: index, ask, chat.
Next: 10. The citation network.