pith. sign in

arxiv: 2512.09154 · v2 · submitted 2025-12-09 · 💻 cs.CE

PILL-CoDe: Inverse Design of Polypills via Automatic Differentiation for Prescribed Drug-Release Kinetics

Rahul Kumar Padhy , Aaditya Chandrasekhar , Amir M. Mirzendehdel This is my paper

Pith reviewed 2026-05-16 23:00 UTC · model grok-4.3

classification 💻 cs.CE
keywords inverse designpolypillautomatic differentiationdrug releasesupershapeneural networkdissolution modelinggradient optimization
0
0 comments X

The pith

PILL-CoDe uses automatic differentiation to co-optimize polypill shape and excipient distribution for any target drug release profile.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Current polypill design relies on ad hoc parameter sweeps that cannot efficiently explore the space of shapes, compositions, and release behaviors. The paper presents PILL-CoDe, an end-to-end differentiable framework that parametrizes geometry with supershapes and material distribution with a coordinate-based neural network, then simulates dissolution with coupled modified Allen-Cahn and Fickian equations. Automatic differentiation supplies exact gradients so that gradient-based optimization can adjust both geometry and composition at once to match prescribed kinetics while satisfying manufacturability constraints. A sympathetic reader would care because this converts manual tuning into systematic inverse design, enabling precise patient-specific multi-phase release tablets.

Core claim

PILL-CoDe simultaneously optimizes tablet geometry and excipient distribution to match prescribed drug-release kinetics. The framework couples a supershape parametrization of the pill geometry with a coordinate-based neural network representation of the excipient distribution, and governs dissolution through a coupled system of modified Allen-Cahn and Fickian diffusion equations. Implemented in JAX, the entire pipeline is end-to-end differentiable, with automatic differentiation providing exact sensitivities for gradient-based co-optimization of shape and composition under manufacturability constraints. Demonstrations show accurate matching of both monotonic and non-monotonic target release

What carries the argument

The JAX-implemented end-to-end differentiable pipeline that couples supershape geometry parametrization, neural network excipient fields, and physics-based dissolution simulation to enable gradient-based joint optimization.

Load-bearing premise

The chosen modified Allen-Cahn plus Fickian diffusion model accurately captures the dissolution kinetics of the multi-material tablets.

What would settle it

Produce the optimized polypill designs via additive manufacturing and perform in-vitro dissolution tests to measure the actual release kinetics and compare them to the simulated targets.

Figures

Figures reproduced from arXiv: 2512.09154 by Aaditya Chandrasekhar, Amir M. Mirzendehdel, Rahul Kumar Padhy.

Figure 1
Figure 1. Figure 1: Graphical abstract: Given (a) a design domain and a set of candidate materials, the framework employs multimaterial topology opti [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (a) Given a computational domain Ω, we optimize the pill shape (region where ϕ = 1) and the distribution of excipients (γ1, . . . , γS ) so that the resulting dissolution kinetics match the prescribed target shown in (b). excipient distribution to maximize design freedom (Section 2.2.2). Furthermore, the dissolution process is governed by a modified Allen-Cahn phase-field equation [33, 34], coupled with a … view at source ↗
Figure 3
Figure 3. Figure 3: Variations in the supershape’s parameters. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: (a) The supershape as defined by its parameters ( [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The neural network maps spatial coordinates to the material distribution. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: (a) Comparison of the mass release rate, where the optimized release profile (solid green) matches the monotonic target release (dashed [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: (a) Comparison of mass release rates, demonstrating the failure of the single-material design (solid green) to match the non-monotonic [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: The specific parameter bounds for this initialization are detailed in Table 3. [PITH_FULL_IMAGE:figures/full_fig_p014_8.png] view at source ↗
Figure 8
Figure 8. Figure 8: Convergence history of the multi-material optimization. The sequence (top to bottom) shows the update of the design and release curve [PITH_FULL_IMAGE:figures/full_fig_p015_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Dissolution of the optimized pill. Design Type Curvature (n) Lobes (m) Scale (s) Spike [0.5, 2.0] [5, 11] [0.1, 0.4] Circle [1.67, 2.0] [1, 3] [0.1, 0.4] Sunflower [2.5, 4.0] [10, 14] [0.1, 0.4] [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Optimization results for “circle” and “sunflower” initializations. (a, c) Release profiles demonstrate that the optimized designs accurately [PITH_FULL_IMAGE:figures/full_fig_p016_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Dissolution rate constants plotted for three available feedstock grades—fresh, moderately aged, and heavily aged. [PITH_FULL_IMAGE:figures/full_fig_p017_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Optimized for varying minimum volume fraction constraints ( [PITH_FULL_IMAGE:figures/full_fig_p017_12.png] view at source ↗
read the original abstract

Polypills are single oral dosage forms that combine multiple active pharmaceutical ingredients and excipients, enabling fixed-dose combination therapies, coordinated multi-phase release, and precise customization of patient-specific treatment protocols. Recent advances in additive manufacturing facilitate the physical realization of multi-material excipients, offering superior customization of target release profiles. However, polypill formulations remain tuned by ad hoc parameter sweeps. The current design workflows are ill-suited for the systematic exploration of the high-dimensional space of shapes, compositions, and release behaviors. We present PILL-CoDe, a polypill co-design framework that simultaneously optimizes tablet geometry and excipient distribution to match prescribed drug-release kinetics. The framework couples a supershape parametrization of the pill geometry with a coordinate-based neural network representation of the excipient distribution, and governs dissolution through a coupled system of modified Allen-Cahn and Fickian diffusion equations. Implemented in JAX, the entire pipeline is end-to-end differentiable, with automatic differentiation providing exact sensitivities for gradient-based co-optimization of shape and composition under manufacturability constraints. We demonstrate the method through single-phase and multi-excipient case studies, showing accurate matching of both monotonic and non-monotonic target release profiles.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes PILL-CoDe, a co-design framework for polypills that parametrizes tablet geometry using supershapes and excipient distribution using coordinate-based neural networks. Dissolution is simulated via a coupled system of modified Allen-Cahn and Fickian diffusion equations. The entire pipeline is implemented in JAX to enable end-to-end differentiability, allowing gradient-based optimization of shape and composition to achieve prescribed drug-release kinetics under manufacturability constraints. The method is demonstrated on in-silico case studies for single-phase and multi-excipient polypills matching monotonic and non-monotonic target profiles.

Significance. This work introduces an innovative differentiable physics-informed approach to inverse design in pharmaceutical engineering, potentially streamlining the development of complex fixed-dose combination therapies. The use of automatic differentiation for exact gradient computation in a high-dimensional design space is a notable technical contribution. However, the significance is tempered by the purely computational nature of the results, as no experimental validation is provided to confirm that optimized designs translate to real-world performance.

major comments (2)
  1. [Case Studies] The reported demonstrations show the optimizer recovering designs that match synthetic target release profiles generated from the same PDE model. However, no quantitative metrics such as mean squared error, maximum deviation, or convergence behavior are provided to assess the accuracy and reliability of the optimization process.
  2. [Physical Model] The framework's validity hinges on the modified Allen-Cahn and Fickian diffusion model accurately capturing multi-material dissolution kinetics. The manuscript does not include any experimental comparison or parameter calibration against in-vitro data from 3D-printed polypills, which is a load-bearing assumption for claiming practical utility.
minor comments (2)
  1. [Abstract] The abstract claims 'accurate matching' of target profiles but does not specify any error measures; including brief quantitative results would strengthen the summary.
  2. [Notation] Ensure consistent definition of all parameters in the supershape and neural network representations across the text and figures.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the thoughtful review and constructive comments on our manuscript. We address each major comment point by point below. We agree that quantitative metrics will improve the clarity of the case studies and will incorporate them in the revision. For the physical model, we will expand the discussion of assumptions and limitations while clarifying the computational scope of the current work.

read point-by-point responses

  1. Referee: [Case Studies] The reported demonstrations show the optimizer recovering designs that match synthetic target release profiles generated from the same PDE model. However, no quantitative metrics such as mean squared error, maximum deviation, or convergence behavior are provided to assess the accuracy and reliability of the optimization process.

    Authors: We agree that quantitative metrics are essential for rigorously evaluating the optimization performance. In the revised manuscript, we will add tables and figures reporting mean squared error (MSE) and maximum absolute deviation between the achieved and target release profiles for all case studies. We will also include convergence plots showing the evolution of the loss function and design parameters over optimization iterations, along with statistics on run-to-run variability when using different initializations. These additions will directly quantify the accuracy and reliability of the gradient-based co-optimization. revision: yes

  2. Referee: [Physical Model] The framework's validity hinges on the modified Allen-Cahn and Fickian diffusion model accurately capturing multi-material dissolution kinetics. The manuscript does not include any experimental comparison or parameter calibration against in-vitro data from 3D-printed polypills, which is a load-bearing assumption for claiming practical utility.

    Authors: The current manuscript presents a computational framework for inverse design, with all demonstrations performed in-silico using synthetic targets generated from the same PDE system. We will revise the manuscript to include an expanded discussion of the model assumptions, citing the literature basis for the modified Allen-Cahn and Fickian components, and explicitly state the limitations regarding direct experimental validation. Parameter calibration against in-vitro data from 3D-printed polypills is an important next step but lies outside the scope of this work, which focuses on establishing the differentiable co-design methodology. We will add a dedicated limitations subsection outlining plans for future experimental studies. revision: partial

standing simulated objections not resolved
  • Direct experimental validation or parameter calibration of the optimized polypill designs against in-vitro dissolution data from 3D-printed samples cannot be provided in the current revision, as the work is purely computational and no such data were generated.

Circularity Check

0 steps flagged

No circularity: optimization recovers designs for externally prescribed targets via independent physics model

full rationale

The paper defines an end-to-end differentiable pipeline (supershape geometry + NN excipient field + modified Allen-Cahn/Fickian PDE) whose loss is the mismatch to a prescribed release profile supplied as an external input. Demonstrations recover parameters that reproduce synthetic targets generated from the same forward model, but this is standard validation of an inverse solver rather than a reduction of the claimed result to its own fitted values by construction. No self-citation load-bearing steps, no uniqueness theorems imported from prior author work, no ansatz smuggled via citation, and no renaming of known empirical patterns appear in the provided text. The central claim (gradient-based co-optimization under manufacturability constraints) remains independent of the target data and does not collapse to a tautology.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the assumption that the chosen PDE model is a faithful forward simulator and that the supershape plus neural-network representation is sufficiently expressive; no new physical constants or entities are introduced.

axioms (2)
  • domain assumption Modified Allen-Cahn equation accurately models the moving dissolution front in multi-material tablets
    Invoked to govern the phase-change aspect of dissolution; no derivation or validation supplied in abstract.
  • domain assumption Fickian diffusion governs drug transport inside the dissolving matrix
    Standard transport assumption used to couple with the phase-field model.

pith-pipeline@v0.9.0 · 5525 in / 1476 out tokens · 30967 ms · 2026-05-16T23:00:42.672805+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

65 extracted references · 65 canonical work pages · 2 internal anchors

  1. [1]

    Yasin, M

    H. Yasin, M. Al-Tabakha, S. Chan, Fabrication of polypill pharmaceutical dosage forms using fused deposition modeling 3d printing: A systematic review. pharm 2024 (2024)

    work page 2024

  2. [2]

    U.S. Food and Drug Administration, CDER, Discussion paper: Distributed manufacturing and point of care manufacturing of drugs,https: //downloads.regulations.gov/FDA-2022-N-2316-0002/content.pdf, federal Register notice: 87 FR 62416 (Oct 14, 2022) (2022)

    work page 2022

  3. [3]

    Food and Drug Administration, CDER, Distributed manufacturing of drugs: Stakeholder feedback and action plan,https://www.fda

    U.S. Food and Drug Administration, CDER, Distributed manufacturing of drugs: Stakeholder feedback and action plan,https://www.fda. gov/media/173449/download(2023)

    work page 2023

  4. [4]

    K. Sen, T. Mehta, S. Sansare, L. Sharifi, A. W. Ma, B. Chaudhuri, Pharmaceutical applications of powder-based binder jet 3d printing process–a review, Advanced drug delivery reviews 177 (2021) 113943

    work page 2021

  5. [5]

    Iqbal, Q

    H. Iqbal, Q. Fernandes, S. Idoudi, R. Basineni, N. Billa, Status of polymer fused deposition modeling (fdm)-based three-dimensional printing (3dp) in the pharmaceutical industry, Polymers 16 (3) (2024) 386

    work page 2024

  6. [6]

    Seoane-Via ˜no, P

    I. Seoane-Via ˜no, P. Januskaite, C. Alvarez-Lorenzo, A. W. Basit, A. Goyanes, Semi-solid extrusion 3d printing in drug delivery and biomedicine: Personalised solutions for healthcare challenges, Journal of controlled release 332 (2021) 367–389

    work page 2021

  7. [7]

    Charoo, C

    N. Charoo, C. Funkhouser, M. Kuttolamadom, M. Khan, Z. Rahman, Opportunities, and challenges of selective laser sintering 3d printing in personalized pharmaceutical manufacturing, Am Pharm Rev (2021)

    work page 2021

  8. [8]

    Deshmane, P

    S. Deshmane, P. Kendre, H. Mahajan, S. Jain, Stereolithography 3d printing technology in pharmaceuticals: a review, Drug Development and Industrial Pharmacy 47 (9) (2021) 1362–1372

    work page 2021

  9. [9]

    X. Xu, P. Robles-Mart ´ınez, C. M. Madla, F. Joubert, A. Goyanes, A. W. Basit, S. Gaisford, Stereolithography printing of an antihypertensive polyprintlet: case study of an unexpected photopolymer drug reaction, International Journal of Pharmaceutics 586 (2020) 119594.doi: 10.1016/j.ijpharm.2020.119594

    work page doi:10.1016/j.ijpharm.2020.119594 2020

  10. [10]

    FierceBiotech, Mit-based startup nabs $30m for first 3-d printed drug approved by fda, accessed 2025-10-26 (aug 2015)

    work page 2025

  11. [11]

    Scoutaris, M

    N. Scoutaris, M. R. Alexander, P. R. Gellert, C. J. Roberts, Inkjet printing as a novel medicine formulation technique, Journal of controlled release 156 (2) (2011) 179–185

    work page 2011

  12. [12]

    S. J. Trenfield, A. Awad, A. Goyanes, S. Gaisford, A. W. Basit, 3d printing pharmaceuticals: drug development to frontline care, Trends in pharmacological sciences 39 (5) (2018) 440–451

    work page 2018

  13. [13]

    Aprecia Pharmaceuticals, Fda approves the first 3d printed drug product,https://aprecia.com/resources/press/ fda-approves-the-first-3d-printed-drug-product/(2015)

    work page 2015

  14. [14]

    Adamo, R

    A. Adamo, R. L. Beingessner, M. Behnam, J. Chen, T. F. Jamison, K. F. Jensen, J.-C. M. Monbaliu, A. S. Myerson, E. M. Revalor, D. R. Snead, et al., On-demand continuous-flow production of pharmaceuticals in a compact, reconfigurable system, Science 352 (6281) (2016) 61–67

    work page 2016

  15. [15]

    Zhang, N

    P. Zhang, N. Weeranoppanant, D. A. Thomas, K. Tahara, T. Stelzer, M. G. Russell, M. O’Mahony, A. S. Myerson, H. Lin, L. P. Kelly, et al., Advanced continuous flow platform for on-demand pharmaceutical manufacturing, Chemistry–A European Journal 24 (11) (2018) 2776–2784. 18

    work page 2018

  16. [16]

    Defense Advanced Research Projects Agency, Battlefield medicine,https://www.darpa.mil/research/programs/ battlefield-medicine(2017)

    work page 2017

  17. [17]

    J. J. Lewin III, E. J. Choi, G. Ling, Pharmacy on demand: New technologies to enable miniaturized and mobile drug manufacturing, American Journal of Health-System Pharmacy 73 (2) (2016) 45–54

    work page 2016

  18. [18]

    URLhttps://globalbiodefense.com/2012/03/26/darpa-biologically-derived-medicines-on-demand-baa/

    Defense Advanced Research Projects Agency (DARPA), Biologically-derived medicines on demand (bio-mod): Broad agency announcement, Program announcement (archival report), accessed 2025-10-26 (2012). URLhttps://globalbiodefense.com/2012/03/26/darpa-biologically-derived-medicines-on-demand-baa/

    work page 2025

  19. [19]

    International Space Station National Laboratory, In space production applications,https://issnationallab.org/ research-and-science/space-research-overview/research-areas/in-space-production-applications/(2024)

    work page 2024

  20. [20]

    National Aeronautics and Space Administration, In space for earth!,https://www.nasa.gov/wp-content/uploads/2023/10/ in-space-production-applications-overview-october-2023.pdf(2023)

    work page 2023

  21. [21]

    Konta, S

    A. Konta, S. Ab Rahim, E. Pawlikowska, J. Mrosko, A. Kaczmarek-Szcz ´os, M. Paluch, Personalised printed medicines: Which techniques print what, Pharmaceutics 9 (3) (2017) 21.doi:10.3390/pharmaceutics9030021

    work page doi:10.3390/pharmaceutics9030021 2017

  22. [22]

    Panetta, H

    J. Panetta, H. Mohammadian, E. Luci, V . Babaei, Shape from release: Inverse design and fabrication of controlled release structures, ACM Transactions on Graphics 41 (6) (2022) 1–16.doi:10.1145/3550454.3555456

    work page doi:10.1145/3550454.3555456 2022

  23. [23]

    Altunay, J

    R. Altunay, J. Suuronen, E. Immonen, L. Roininen, J. H ¨am¨al¨ainen, Computational design of personalized drugs via robust optimization under uncertainty, arXiv preprint arXiv:2507.16470 (2025)

    work page arXiv 2025

  24. [24]

    B. Du, V . R. Daniels, Z. Vaksman, J. L. Boyd, C. Crady, L. Putcha, Evaluation of physical and chemical changes in pharmaceuticals flown on space missions, The AAPS journal 13 (2) (2011) 299–308

    work page 2011

  25. [25]

    Van der Merwe, J

    J. Van der Merwe, J. Steenekamp, D. Steyn, J. Hamman, The role of functional excipients in solid oral dosage forms to overcome poor drug dissolution and bioavailability, Pharmaceutics 12 (5) (2020) 393

    work page 2020

  26. [26]

    S. A. Khaled, J. C. Burley, M. R. Alexander, J. Yang, C. J. Roberts, 3d printing of five-in-one dose combination polypill with defined immediate and sustained release profiles, Journal of controlled release 217 (2015) 308–314

    work page 2015

  27. [27]

    D. M. Saylor, C.-S. Kim, D. V . Patwardhan, J. A. Warren, Diffuse-interface theory for structure formation and release behavior in controlled drug release systems, Acta biomaterialia 3 (6) (2007) 851–864

    work page 2007

  28. [28]

    Mohseni-Motlagh, R

    S.-F. Mohseni-Motlagh, R. Dolatabadi, M. Baniassadi, M. Karimpour, M. Baghani, Tablet geometry effect on the drug release profile from a hydrogel-based drug delivery system, Pharmaceutics 15 (7) (2023) 1917

    work page 2023

  29. [29]

    Goyanes, P

    A. Goyanes, P. R. Martinez, A. Buanz, A. W. Basit, S. Gaisford, Effect of geometry on drug release from 3d printed tablets, International journal of pharmaceutics 494 (2) (2015) 657–663

    work page 2015

  30. [30]

    Gielis, A generic geometric transformation that unifies a wide range of natural and abstract shapes, American journal of botany 90 (3) (2003) 333–338

    J. Gielis, A generic geometric transformation that unifies a wide range of natural and abstract shapes, American journal of botany 90 (3) (2003) 333–338

    work page 2003

  31. [31]

    R. K. Padhy, K. Suresh, A. Chandrasekhar, Tomas: topology optimization of multiscale fluid flow devices using variational auto-encoders and super-shapes, Structural and Multidisciplinary Optimization 67 (7) (2024) 119

    work page 2024

  32. [32]

    Chandrasekhar, K

    A. Chandrasekhar, K. Suresh, Multi-material topology optimization using neural networks, Computer-Aided Design 136 (2021) 103017

    work page 2021

  33. [33]

    S. M. Allen, J. W. Cahn, A microscopic theory for antiphase boundary motion and its application to antiphase domain coarsening, Acta metallurgica 27 (6) (1979) 1085–1095

    work page 1979

  34. [34]

    Z. Xu, H. Huang, X. Li, P. Meakin, Phase field and level set methods for modeling solute precipitation and/or dissolution, Computer Physics Communications 183 (1) (2012) 15–19

    work page 2012

  35. [35]

    Fick, Ueber diffusion, Annalen der physik 170 (1) (1855) 59–86

    A. Fick, Ueber diffusion, Annalen der physik 170 (1) (1855) 59–86

    work page

  36. [36]

    Goodacre, P

    B. Goodacre, P. Murray, A mathematical model of drug absorption, Journal of Clinical Pharmacy and Therapeutics 6 (2) (1981) 117–133

    work page 1981

  37. [37]

    Siepmann, F

    J. Siepmann, F. Siepmann, Mathematical modeling of drug dissolution, International journal of pharmaceutics 453 (1) (2013) 12–24

    work page 2013

  38. [38]

    Z. Xu, P. Meakin, Phase-field modeling of solute precipitation and dissolution, The Journal of chemical physics 129 (1) (2008)

    work page 2008

  39. [39]

    S. Yang, N. Ukrainczyk, A. Caggiano, E. Koenders, Numerical phase-field model validation for dissolution of minerals, Applied Sciences 11 (6) (2021) 2464

    work page 2021

  40. [40]

    Singer-Loginova, H

    I. Singer-Loginova, H. Singer, The phase field technique for modeling multiphase materials, Reports on progress in physics 71 (10) (2008) 106501

    work page 2008

  41. [41]

    R. Yang, K. L. Chong, H.-R. Liu, R. Verzicco, D. Lohse, Melting and solidification in periodically modulated thermal convection, Journal of Fluid Mechanics 998 (2024) A10

    work page 2024

  42. [42]

    Sleziona, D

    D. Sleziona, D. R. Ely, M. Thommes, Modeling of particle dissolution behavior using a geometrical phase-field approach, Molecular Phar- maceutics 19 (11) (2022) 3749–3756

    work page 2022

  43. [43]

    Bringedal, L

    C. Bringedal, L. V on Wolff, I. S. Pop, Phase field modeling of precipitation and dissolution processes in porous media: Upscaling and numerical experiments, Multiscale Modeling & Simulation 18 (2) (2020) 1076–1112

    work page 2020

  44. [44]

    Dokoumetzidis, P

    A. Dokoumetzidis, P. Macheras, A century of dissolution research: from noyes and whitney to the biopharmaceutics classification system, International journal of pharmaceutics 321 (1-2) (2006) 1–11

    work page 2006

  45. [45]

    Kumar Padhy, P

    R. Kumar Padhy, P. Thombre, K. Suresh, A. Chandrasekhar, Treetop: topology optimization using constructive solid geometry trees, Struc- tural and Multidisciplinary Optimization 68 (2) (2025) 1–13

    work page 2025

  46. [46]

    R. K. Padhy, A. Chandrasekhar, Photos: topology optimization of photonic components using a shape library, Engineering with Computers 41 (2) (2025) 1141–1153

    work page 2025

  47. [47]

    F. Wang, B. S. Lazarov, O. Sigmund, On projection methods, convergence and robust formulations in topology optimization, Structural and multidisciplinary optimization 43 (6) (2011) 767–784

    work page 2011

  48. [48]

    Tancik, P

    M. Tancik, P. Srinivasan, B. Mildenhall, S. Fridovich-Keil, N. Raghavan, U. Singhal, R. Ramamoorthi, J. Barron, R. Ng, Fourier features let networks learn high frequency functions in low dimensional domains, Advances in neural information processing systems 33 (2020) 7537–7547

    work page 2020

  49. [49]

    Chandrasekhar, K

    A. Chandrasekhar, K. Suresh, Approximate length scale filter in topology optimization using fourier enhanced neural networks, Computer- Aided Design 150 (2022) 103277. 19

    work page 2022

  50. [50]

    A. F. Agarap, Deep learning using rectified linear units (relu), arXiv preprint arXiv:1803.08375 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  51. [51]

    Glorot, Y

    X. Glorot, Y . Bengio, Understanding the difficulty of training deep feedforward neural networks, in: Proceedings of the thirteenth international conference on artificial intelligence and statistics, JMLR Workshop and Conference Proceedings, 2010, pp. 249–256

    work page 2010

  52. [52]

    Bradbury, R

    J. Bradbury, R. Frostig, P. Hawkins, M. J. Johnson, C. Leary, D. Maclaurin, G. Necula, A. Paszke, J. VanderPlas, S. Wanderman-Milne, Q. Zhang, JAX: composable transformations of Python+NumPy programs (2018). URLhttp://github.com/jax-ml/jax

    work page 2018

  53. [53]

    L. D. Dalcin, R. R. Paz, P. A. Kler, A. Cosimo, Parallel distributed computing using python, Advances in Water Resources 34 (9) (2011) 1124 – 1139, new Computational Methods and Software Tools.doi:10.1016/j.advwatres.2011.04.013

    work page doi:10.1016/j.advwatres.2011.04.013 2011

  54. [54]

    Balay, S

    S. Balay, S. Abhyankar, M. Adams, J. Brown, P. Brune, K. Buschelman, L. Dalcin, A. Dener, V . Eijkhout, W. Gropp, et al., Petsc users manual (2019)

    work page 2019

  55. [55]

    D. P. Kingma, J. Ba, Adam: A method for stochastic optimization, arXiv preprint arXiv:1412.6980 (2014)

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  56. [56]

    Kervadec, J

    H. Kervadec, J. Dolz, J. Yuan, C. Desrosiers, E. Granger, I. B. Ayed, Constrained deep networks: Lagrangian optimization via log-barrier extensions, in: 2022 30th European Signal Processing Conference (EUSIPCO), IEEE, 2022, pp. 962–966

    work page 2022

  57. [57]

    Nocedal, S

    J. Nocedal, S. J. Wright, Numerical optimization, Springer, 1999

    work page 1999

  58. [58]

    Chandrasekhar, A

    A. Chandrasekhar, A. Mirzendehdel, M. Behandish, K. Suresh, Frc-tounn: Topology optimization of continuous fiber reinforced composites using neural network, Computer-Aided Design 156 (2023) 103449

    work page 2023

  59. [59]

    R. K. Padhy, K. Suresh, A. Chandrasekhar, Toflux: A differentiable topology optimization framework for multiphysics fluidic problems, arXiv preprint arXiv:2508.17564 (2025)

    work page arXiv 2025

  60. [60]

    R. K. Padhy, K. Suresh, A. Chandrasekhar, Tomatoes: Topology and material optimization for latent heat thermal energy storage devices, arXiv preprint arXiv:2510.07057 (2025)

    work page arXiv 2025

  61. [61]

    Blondel, Q

    M. Blondel, Q. Berthet, M. Cuturi, R. Frostig, S. Hoyer, F. Llinares-L ´opez, F. Pedregosa, J.-P. Vert, Efficient and modular implicit differenti- ation, Advances in neural information processing systems 35 (2022) 5230–5242

    work page 2022

  62. [62]

    Q. Wang, P. Moin, G. Iaccarino, Minimal repetition dynamic checkpointing algorithm for unsteady adjoint calculation, SIAM Journal on Scientific Computing 31 (4) (2009) 2549–2567

    work page 2009

  63. [63]

    K. A. James, H. Waisman, Topology optimization of viscoelastic structures using a time-dependent adjoint method, Computer Methods in Applied Mechanics and Engineering 285 (2015) 166–187

    work page 2015

  64. [64]

    Alexandersen, M

    J. Alexandersen, M. Appel, Large-scale topology optimisation of time-dependent thermal conduction using space-time finite elements and a parallel space-time multigrid preconditioner, arXiv preprint arXiv:2508.09589 (2025)

    work page arXiv 2025

  65. [65]

    Appel, J

    M. Appel, J. Alexandersen, One-shot parareal approach for topology optimization of transient heat flow, SIAM Journal on Scientific Com- puting 47 (6) (2025) B1450–B1480. 20

    work page 2025