Search Documents

By C. Dorfling, M. Tadie, M. Tucker

Electrowinning is the final step in the hydrometallurgical production of high purity copper, involving the passage of electric current through an electrolyte and consequent plating of solid copper to form cathode sheets. Key electrowinning performance indicators are the current efficiency, specific energy consumption, yield, and quality of the metal. The high energy demand and associated cost mean that determination of performance indicators is critical during operation, but ‘online’ measurement is impractical due to the nature of the measurements and corrosive environment brought about by acid mist. This research considered the development of a predictive electrowinning model as a shift towards a pre-emptive approach to attaining plant performance data. A steady-state model was created from input variables available in the industry, using a first-principles approach combining mass balances, electrochemical and thermodynamic equations, and property correlations. A circuit diagram equivalent of an electrowinning cell, containing multiple pairs of electrodes, was used as the backbone of the model. The kinetics of each electrochemical reaction were modelled including iron as a major impurity. Losses in current efficiency due to ineffective electrode contact, short circuits, and stray currents were also incorporated. The model was used to predict the current efficiencies for tests performed at varying conditions in a bench-scale electrowinning cell; the average absolute error between the measured and predicted current efficiencies was 3.2%. The model requires a unique parameter fitting approach for each industrial operation; if parameters are customised for specific plants, the model will be a useful tool to predict performance and can aid the development of effective process control systems in copper electrowinning tank houses. Keywords: Copper; current efficiency; electrowinning; modelling; plating rate

Jan 1, 2020

By Wei Sun, Ruolin Wang, Wenjuan Sun, Junhao Fu, Haisheng Han, Zhao Wei

Pb–benzohydroxamic acid (BHA) (lead complexes of benzohydroxamic acid) exhibits high selectivity but low collecting efficiency in scheelite flotation. This paper reports the manner in which BHA and sodium dodecyl sulfate (SDS) were employed as ligands to coordinate with the lead ions to obtain a colloidal lead-group multiple ligand complex, i.e., the Pb–BHA–SDS, for improving the collecting efficiency of Pb–BHA. The flotation mechanism of Pb–BHA–SDS with respect to scheelite was investigated from the viewpoint of interface science based on the pure mineral flotation, Fourier transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), and contact angle measurements. The results showed that the BHA and SDS coordinated with Pb2+ to form novel leadgroup multiple ligand complexes, which are different from the traditional understanding associated with the mixed collectors. Pb–BHA–SDS was a better collector with stronger collecting ability and excellent selectivity in regard to a scheelite flotation than Pb–BHA. The addition of SDS to Pb–BHA enhanced the latter’s adsorption ability on the scheelite surface and improved the hydrophobicity of the collector’s nonpolar groups, rendering Pb–BHA–SDS as a more efficient collector at a low dosage. Keywords: Scheelite; Pb-BHA complexes; sodium dodecyl sulfate; multiple ligands; Pb–BHA–SDS

Jan 1, 2020

By S. Chai

Waste dumps are widely used to discharge stripped mine wastes in open pit mining methods. As mining progresses, the height of the waste dumps will increase gradually, which may induce sliding failure once the height exceeds a critical value. Therefore, estimation of the maximum dumping height is crucial to the life-cycle use of a waste dump. However, published information about the maximum height of mine waste dumps are extremely rare. In this study, the maximum dumping height is estimated based on the strength-reduction technique by numerical simulations. The influences of dump geometry and properties on the maximum height are investigated. The results show that the maximum dump height decreases as the dump slope inclination angle α or the unit weight of dump materials γ increases, but decreases when the shear parameters of dump materials (cohesion c and internal friction angle ϕ) increase. The maximum height starts to increase with the ground inclination β when shear failure occurs at the interface between the dump and the base. Furthermore the effects of the dump width B, Young’s modulus E, and Poisson’s ratio μ of dump materials on the maximum height are almost insignificant. Good agreement was found between numerical results available in the literature and those obtained in this study.

Oct 1, 2022

By R. G. Diale, H. R. Chauke, T. M. Ledwaba, P. E. Ngoepe

FeCo alloys play an important role in soft magnetic materials with a wide range of technological applications due to their high saturation magnetization and Curie temperature. However, these alloys have low ductility at room temperature. The ductility can be improved by the ternary addition of elements such as V. A density functional theory supercell approach was used to generate B2 Fe50Co50-xVx (0 ≤ x ≤ 50) structures and the properties were evaluated at different atomic percentage compositions to determine the ductility at room temperature. The structures were fully optimized to obtain better equilibrium ground-state properties such as lattice parameters and thermodynamic properties for both binary and ternary systems. The stability of Fe50Co50- xVx was evaluated from the heats of formation, elastic properties, and phonon dispersion curves. Furthermore, magnetic strength was evaluated from magnetic moments. It was found that all structures are thermodynamically stable due to the negative heats of formation. The calculated Pugh’s ratio and Poisson’s ratio confirm that alloying with V effectively improved the ductility. It was also found that Fe50Co50-xVx showed a positive shear modulus for the entire concentration range investigated, in agreement with phonon dispersion curves. The ternary addition of V to the FeCo system resulted in reduced magnetic strength due to a decrease in magnetic moments. These findings reveal that B2 FeCo-V alloys can be used as components and actuators for the automotive industry.

Jun 10, 2024

The first annual student prizegiving function of the South African Institute of Mining and Metallurgy was held in the Dorothy Susskind Auditorium, University of the Witwatersrand, on Thursday, 26th May, 1977. The address (see opposite page) was given by Dr M. G. Atmore, a director of Anglo American Corporation and Vice-President of the Institute.

Jan 1, 1977

By Y. R. Murthy, S. Angadi, A. K. Bhatnagar, S. K. Tripathy, L. O. Filippov

Dolomite is an essential raw material for the refractory industry, a flux in the iron and steel industry and also for the extraction of magnesium metal. Generally, particle size reduction, screening and classification are used for the preparation of the high-grade sized material for different applications, whereas low-grade fine fractions are not utilised. These fines or low-grade overburdens are dumped at the mine site for future utilisation, which creates environmental issues along with inadequate space management. In the present investigation, a detailed study was carried out to decrease the silica content of the low-grade dolomite fines from Gomardih , India, containing 28.6% CaO and 20.1% MgO along with 8.3% silica. Different process flow sheet options were evaluated with a combination of dry classifier, wet classifier, gravity concentrators as well as flotation. Different process variables were assessed along with mineralogical characterisation to optimise the process for efficient rejection of siliceous gangue minerals. Results indicated that dolomite, calcite and quartz are the major mineral phases present in the ore along with other minor minerals. It was concluded that two-stage flotation was an effective process for the reduction of silica up to below 1% after grinding to a particle size below 212 μm. A process flow sheet with a combination of gravitation air classifier and flotation can produce a concentrate with silica content below 4.5%. Keywords: Dolomite, flux, dry classification, flotation, characterisation, beneficiation

Jan 1, 2020

By V. Ross

"Bench-scale flotation tests play a major role in the geometallurgical process value chain. Despite their apparent simplicity, these tests pose pitfalls in terms of operation, as well as the interpretation and modelling of results. In this paper we highlight key aspects to consider in generating accurate and reproducible experimental results, and critically review some assumptions relating to the estimation of entrainment and froth recovery. We subsequently focus on first-order equations to model kinetics and propose a method that accounts more realistically for the decay in kinetics over the duration of a float, and the contributions of froth stability and hence mass pull. The method was applied to three case studies on platinum group metal (PGM) ores and matched or outperformed first-order, two-component models. Also, it allows for a deeper analysis of the integrity of the bench-scale flotation results. IntroductionGeometallurgy is a structured, multidisciplinary approach that integrates key data from the geological, mining, mineralogical, and metallurgical aspects of the value chain to construct spatial models of an ore deposit such that the overall economic benefit and sustainability of an operation can be optimized. Process geometallurgy is aimed primarily at enhancing metallurgical efficiency and plant throughput by the application of appropriate mining and processing strategies, as well as supporting production planning and operational strategies to reduce the impact of highly variable ores. With flotation being the dominant concentrating process in the mineral value chain, the bench-scale flotation test is an important characterization tool – a relatively cheap, simple, and rapid way to obtain key information about the ore and its amenability to economic extraction of valuable minerals. This includes the evaluation of different ore types and the implications of mineralogical composition and texture for comminution and floatability, the effect of the grinding environment on the liberation of minerals, the electrochemical environment in the pulp (Bradshaw et al., 2006), the screening of reagents to improve floatability (Ekmekci et al., 2003), and the role of hydrodynamics in the recovery of valuables and the entrainment of gangue (Amini et al., 2016; Wang et al., 2016). It is also a critical tool in auditing the performance of a flotation plant by hot floats, paving the way for concept and feasibility studies and providing essential data on kinetics for the purposes of scale-up, optimization, and simulation of flotation circuits."

Mar 1, 2019

By B. k. Kenzhaliyev, M. A. Naimanbayev, Zh. A. Baltabekova, N. G. Lokhova

In research various desorbents were used to desorb the following rare-earth metals from Purosorb 140 cation exchanger, i.e. ammonium chloride; ammonium sulfate; a mixture of nitric acid and ammonium nitrate; a mixture of hydrochloric acid and ammonium chloride; a mixture of sulfuric acid and ammonium sulfate. It was established that a solution containing 30 g/dm3 of H2SO4 + 300 g/dm3 of (NH4) 2SO4 is the most optimal desorbing solution to desorb REMs from macroporous Purosorb 140 cation exchanger impregnated with rare-earth metals from industrial dihydrate wet-process phosphoric acid. It was established that the residual REM content in the cation exchanger decreased from 0.89 to 0.10 wt. % with an increase in the ammonium sulfate concentration from 10 to 400 g/dm3. Technological desorption regimes for rare-earth metals from cation exchanger were developed. The optimal conditions were determined to desorb rare-earth metals from the Purosorb 140 cation exchanger, particularly temperature 25°C; L:S ratio = (0.8 ÷ 1.1):1; duration = 1.0-1.5 hours. Keywords: Rare elements, sorption, cation exchangers, wet-process phosphoric acid, desorption

Jan 1, 2020

By P. L. Ngwenyama, I. Maubane

Soundless chemical demolition agents (SCDAs) are proving to be the future of sustainable and environmental-friendly mining. These expansive chemicals become critical when there is a need to break rock close to communities, environmentally sensitive areas, and critical structures and infrastructure. Unlike explosives, the SCDAs do not generate air-blast, ground vibrations, flyrock, noise, dust, and noxious fumes, which can negatively affect the environment, surrounding communities, sensitive areas, and critical structures and infrastructure. Over the years, several types of SCDA products have been used successfully in the civil and construction industries for the demolition of old and dilapidated buildings near vibrations and environmentally sensitive areas. They have also been used to fragment boulders in the mining and construction industries. However, to date, no study has investigated the applications of soundless chemical demolition on a large-scale in opencast mines. Therefore, this study evaluated whether soundless chemical demolition can break large volumes of in situ rock. This was done by conducting five trials using Nex-Pand soundless chemical demolition agents in four different surface mining sites. Trial 1 was unsuccessful as it was unable to break the rock due to the large confinement between the drilled holes and the use of a hole diameter that was 182% larger than recommended. Trial 2 was a success as it was able to fracture the rock with a crack width that reached 110 mm after 72 hours. Trial 3 generated cracks at the biggest hole diameter of 102 mm. The cracks developed and ceased to increase after 96 hours at a crack width of 100 mm. Trial 4 achieved the smallest crack width due to a very competent rock compared to the other sites. Trial 5 generated cracks at a slower rate than other trials because of the use of Nex-Pand powder at lower ambient temperatures than recommended. These trials proved that soundless chemical demolition agents can replace explosives in areas closer to sensitive infrastructure and communities. This will ultimately enable mines to unlock value and access deposits in these areas.

Jul 23, 2025

By W. A. Smith, M. C. Bekker, C. Marnewick

Projects or project-orientated approaches have become a common form of work in nearly all sectors of economies. This has led to concepts such as ‘projectified’ and ‘project orientated’ organizations. By defining projectification of a company, industry, or economy as the share of project work in total work, one can reasonably determine the impact that project management, and by default projectification, has had on that company, industry, or economy in terms of staff optimization and allocation. This paper presents the results for such a projectification study of the South African mining industry. This sector has long been a significant contributor to the country’s economy from a gross value added (GVA) and employment point of view. Understanding the impact of projectification and the project management way of work on this industry may potentially add significant value to both the mining and project management knowledge areas. We show that although the mining industry is considered by some to operate in archaic ways, the level of projectification has increased over time, and now represents approximately one third of all work conducted.

May 1, 2021

By M. Pawlik, N. Ghaffari Touran

The interactions between apatite and hematite were studied through zeta potential and stability experiments, for this paper. The zeta potential measurements were conducted in both background electrolyte and mineral supernatants. The difference in electrokinetic properties of minerals in the presence and absence of supernatant was explained in terms of the dissolution of minerals and the subsequent readsorption of dissolved species at the mineral-water interface. The stability measurements in individual and mixed mineral systems were also performed to identify the extent of interaction between apatite and hematite upon their direct contact. The results from stability measurements in individual and mixed mineral systems indicate that there is an interaction between apatite and hematite. Based on the results, the implications for the apatite-hematite separation process were discussed. Keywords: Apatite, hematite, dissolved ions, stability, zeta potential

Jan 1, 2020

By J. J. Botha

As a Demand Side Management (DSM) Project, money was granted for the procurement and installation of a 3-Chamber Pipe Feeder System (3-CPFS) at AngloGold Ashanti?s Moab Khotsong Mine in the Vaal River Area, Orkney. The first submission was done to ESKOM in December 2002 and the contract was signed in September 2006 at a cost of R 57-million with an envisaged reduction in power of 7 Megawatt per month based on a 90% utilization factor. The project was finally commissioned in December 2008. The technology is not new, but the intent of the paper is to share with others the experience gained by using the technology in a deep-level mine in order to reduce overall power consumption.

Jan 1, 2009

[Consider the situation where we wish to estimate the value of a block of ore W from the values of n samples wt (i = 1, 2.... n). The samples can be located anywhere inside or outside W. The blocks and the samples can have any size or shape (Fig. 8.1). We define ws as the set of all the samples wi: WS = Iw1; W2; W3.. . wn and we wish to estimate the block W using the samples ws. Let xi be the value of the ith sample and µw the unknown average value of the block. We can estimate µw from the average value of all the samples. Let µw be the estimator of µw. As the samples need not be of equal size, the sample average is: µw = wi xi/~ wi. (8.1) The mean squared error made when estimating µw by µw is known as the variance of estimation of W by ws. It is the error made when giving the value of the sample set ws to the block W and is written o2E (WS to W). By definition: o2E (ws to W) = E [(µw - µw)2]. (8.2) If n = 1, this error is known as the extension variance. Since the sum of the weights given to the sample values is equal to 1,0, the estimator µw is unbiased (the question of the absence of bias is studied in more detail in Chapter 9): E [( µw - µw)1 = 0. (8.3) The error (µw - µw) has zero mean and a variance o2E = E (W8 to W). If we assume that this error is normally distributed, we can define confidence limits for µw. Let tP be the value of the standardized normal variance t such that the probability that t is less than tP is p: (t,) = .\/27T f exp (- t2/2) dt = p (8.4) The probability that µw is larger than µw + t1-P oE is p. The probability that µw is smaller than µw- t1-P oE is p. The following approximation formulae are often used to calculate 2p central confidence limits for the block value: 68% central confidence limits (p = 16%): µw - o2E, and µw + o2E. 95 % central confidence limits (p = 2,5 %): µw - 2 oE, and µw + 2 oE. Tables of the function 0 can be found in any textbook on statistics (for example, Fraser, 1958, p. 388) (see Table 7.2). [ ] The last line of Table 2.2, corresponding to f = 00, also contains the value of tP for p = 80, 90, 95 and 97,5 %.]

Jan 1, 1978

By L. Roux

This review was initially motivated by the author’s own experience in attempting to reconcile run-of mine production tons with the Coal Resource and Reserve estimation of mineable, in-situ tonnages. The primary scientific measurements and observations that are collected during exploration at the beginning of the value chain will have a profound effect on the future of the mining operation. Due diligence must be exercised during the planning of a new mine and during the life of an existing mine. The comments provided by the technical and economic assessment group of Venmyn Deloitte confirmed the need for this critical review. They found the reporting of Coal Resources in South Africa to be inconsistent. This was particularly problematic in the Waterberg Coalfield in Limpopo Province. In this coalfield there are two types of coal deposit present. The first is comprised of the thick intercalated, cyclic coal and shale/mudstone sequences of the Volksrust Formation. This overlies multiple coal seams within the Vryheid Formation, each of which is thick enough to be extracted as an individual raw coal unit. On the other hand, the interbedded coal and shale seams of the Volksrust Formation require beneficiation to separate the coal from the shale. The review examines practices and methods, investigates alternatives, provides checks and balances, and tests these against actual production reconciliations. In conclusion, the best estimates of the mineable, in-situ tonnage will be obtained from the air-dry raw material density. These estimates should be adjusted afterwards to allow for free moisture content. The adjustments are derived from reconciliation data. The greatest contributing factor to the over-estimation of Reserve tonnages is the moisture content of the raw material. This fluctuates significantly under varying conditions in situ, as well as upon exposure to the natural environment. The air-dried density of the raw material includes inherent (structurally bound) moisture within the matrix. It provides credible tonnage estimations of raw material available while also providing an indication of voids within the volume of material being assessed. The calculated solids percentage can be used to adjust the specific gravity, which is determined via the Archimedes principle. This will supply a representative estimate of the material to be mined.

Nov 1, 2021

By J. Björkvall, L. Bååth, D. Malmberg, J. Malm

Microwave technology has for decades been a tool for astronomers to map and understand the complexities of the universe its composition and extent. It is also used on a laboratory scale by spectroscopists to examine properties of atomic and molecular compounds. In this novel work we will introduce microwave technology to be used for the investigation of liquid slag structures. The preliminary results indicate that the alteration of the slag composition could be correlated to the measured microwave refractive index. Investigations have been performed in the Al2O3-CaO-SiO2slag system. Keywords: Microwave technology, slag, structure, liquid, Al2O3-CaO-SiO2

Jan 1, 2004

By R. Jacobs, R. E. L. Pobe

The Moab Khotsong Metallurgical Complex comprises a gold recovery circuit (Noligwa Gold Plant) and a uranium recovery circuit (South Uranium Plant). The metallurgical complex employs the reverse leach technology for the recovery of uranium; specifically, uranium extraction prior to gold using acid leaching processing. The inherent uranium species in the ore are uraninite, coffinite and brannerite. The slurry that has been subjected to the uranium leach is further subjected to solid-liquid separation to separate the uranium in a solution through counter-current decantation and the solids (solids carrying all the gold), which are repulped and returned to the gold plant for gold processing. The uranium leaching process is critical to the uranium playing a pivotal role as a value adding by-product to the gold operation. The benefit of adopting this reverse leach process, for improved gold recovery, is that it exposes gold particles occluded within the uranium matrices and avails them for cyanidation. However, the drive for operational efficiency in uranium extraction must be weighed against the incremental operating cost against the backdrop of depressed spot uranium market prices. Over the years, efforts were directed at maintaining the benefit of extracting uranium to enhance gold recovery and simultaneously, optimising the operating cost model for the uranium operation to remain relevant to the gold operation. The aim of this paper and the corresponding presentation is to give an account of the uranium processing optimisation journey embarked on over the years; namely, the operational tradeoffs made in the process of ensuring that uranium continues to add value to the gold operation. These challenges were faced while striking a fine balance between processing cost and operational efficiencies against the prevailing depressed uranium spot market prices. Mineralogical- and laboratory tests conducted form the support basis for the results and observations articulated in the paper. Keywords: By-products, gold extraction, reverse leach benefit, uranium extraction

Jan 1, 2020

By C. Birch

"The purpose of this study was to establish the impact of discount rates on cut-off grades for narrow tabular gold deposits as characterized by the goldfields of the Witwatersrand Basin in South Africa. There are various methods available for determining the cut-off grade, from simple breakeven calculations to sophisticated software packages that consider a variety of inputs to optimize the cut-off grade. For this study a simple financial model was created in Microsoft Excel® that links the ore flow, block listing, and the cash flow. This allows the cut-off grade to be optimized considering how the cost of capital and chosen discount rate affect the cash flow.The discounted cash flow (DCF) and resultant net present value (NPV) are a widely used valuation method for production properties according to the South African Code for the Reporting of Mineral Asset Valuation (SAMVAL Code). The financial model in this study utilizes the Solver function, as well as simple Microsoft Excel spreadsheet formulae to optimize the NPV. Solver was chosen as it is a standard feature in Excel and thus no additional software costs are incurred beyond the basic Microsoft Office suite. For the purpose of this study, just narrow tabular gold deposits of the Witwatersrand Basin were considered. An example of a typical ore block listing, as well as the costing figures, was obtained from an operating gold mine. The results obtained from the study financial model were compared to the current cut-off grades obtained from the mine using their proprietary optimizer program, and were found to be comparable. The methodology utilized for this study thus appears valid.The cut-off grade was optimized considering the cash flow, which includes the variable mineral resource royalty tax, the variable gold income tax, as well as the discount rate. By comparing the resultant NPVs using discount rates of zero, 9%, and 12%, the impact of the discount rate on cut-off grades, resultant life of mine, and average mining grades (AMG) could be compared for the example ore block listing. The South African Code for Reporting of Exploration Results, Mineral Resources and Mineral Reserves (SAMREC Code) does not require operations to state if a discount rate was considered in determining mineable reserves. Mining companies approach the question of applying discount rates in determining their cut-off grades, and thus mine planning strategy, in different ways. Due to the negative impact that optimizing purely on NPV has on the life of mine, some choose to optimize only on profit. Other companies have reverted to utilizing pay limits as their primary grade planning strategy. The calculated cut-off grade is then considered to be another planning indicator rather than a hard determination of mineable reserves."

Jan 1, 2016

By K. Thambi

The court in Baleni v Minister of Mineral Resources [2019] 2 SA 453 GP and [2020] 4 All SA 374 (GP), deliberated on the protection of rights of a community holding informal land tenure under Customary Law. The contention related to the necessary level of consent needed to acquire a mining right over such land. Moreover, whether consultations with such communities (Section 23, Mineral and Petroleum Resources Development Act 28 of 2002 (MPRDA) or consent (Section 2, Interim Protection of Informal Land Rights Act, No. 31 of 1996 (IPILRA)) was required to acquire such right. The case has a significant bearing on the granting of mining rights in South Africa, and the discretion of the Minister of Mineral Resources (the Minister) in this regard. However, the objectives of the MPRDA and IPILRA do not dovetail, therefore consultation and consent are not mutually exclusive (Tlale, 2020). This note argues that, despite the resounding victory of this case, the peripheral basis surrounding the decision and the various levels of engagement require serious deliberation. Equally, the degree of reliance on the IPILRA requires clarity to avoid aborting the fundamental objectives of the MPRDA. This paper provides considerations and recommendations that may reduce or eliminate the tensions between the statutory and socio-economic rights in the application of the two statutes.

Sep 1, 2022

By X. Jia

The ability to perform three dimensional imaging at micron-scale resolution of real mineral floc sediments and structures using commercially available Micro X-ray Tomography (XMT) instrumentation can be used for the analysis of aggregates and granules, liberation and dewatering of sediments. This paper presents are view of recent work by the authors relevant to the mining and minerals industries. It covers three main topics: X-ray imaging of agglomerates and multi-component mineralic particles,, permeability of sediments, and dissolution of granules with potential application to heap leaching.

Jan 1, 2008

By N. de Koker, Y Jooste, J. Gerber

This paper presents a novel application of the boundary element method for numerical modelling of the seismic activity and hazard that is associated with the shaft pillar extraction at Bambanani mine in South Africa. In the numerical model, the tabular mining excavations are represented by displacement discontinuity elements, and to accommodate crush, shear, and fault slip seismic events, the numerical model is populated with additional displacement discontinuity elements that represent modelled crush- and shear-type failures in the rockmass. Crush-type failures are evaluated by applying the limit equilibrium method and a Hoek-Brown strength criterion, and shear-type failures are evaluated by applying a Coulomb-friction strength criteria to Ortleppshear and geological features. In this paper, the modelling results are presented as time history analyses of modelled potency from 1 January 2010 to 30 June 2022 and normalised exceedance rates for different mining periods using an upper-truncated power law. The normalised exceedance rates of seismic events that have moment magnitude Mw ≥ 1.0 are lowest for the initial mining period (26 events per year) and highest for the final mining period (218 events per year). The relative errors of the modelled potency and normalised exceedance rates are typically less than 10% or 15%, which suggests that the modelling methodology is appropriate for medium- and longer-term forecasting of seismic activity and hazard, as presented in this paper. Shorter-term forecasting is not considered in this paper.

Nov 24, 2025