betaFeedback

WorksComputer Science

OPERA: OPamp Design by an Expert-Taught Reasoning Agent, from Specifications to Layout

Abstract

Analog flows from specifications to a signed-off layout rarely run without manual iteration. Recent large language model (LLM) agents size circuits with optimizer loops over the device dimensions and stop at the schematic. We present OPERA (OPamp design by an Expert-taught Reasoning Agent), in which a general-purpose LLM agent designs CMOS opamps from specifications to post-layout sign-off with open-source tools. First, the designer teaches the agent the gm/ID method: map the specifications to transconductances and minimum currents, choose gm/ID per device role from characterized lookup tables, and size from equations, leaving a few knobs to search and an explainable design. Second, the agent learns the gLayout framework and taught layout practice, and writes a reusable layout generator that closes the layout and extraction loop at designer-approved gates. In SKY130, the agent designed a shared constant-gm bias and six opamps (5T, folded-cascode and telescopic OTAs; Miller, folded-cascode and telescopic two-stage) for fun >= 50 MHz, PM >= 60 degrees, SR 25 V/μs into 1 pF and a minimum dc gain of 30 to 60 dB. All six layouts are DRC-clean and LVS-matched, and their extracted netlists meet every target at all 15 PVT corners ( fun >= 53.7 MHz, PM >= 62.1 degrees, SR >= 27.9 V/μs). The benchmark netlists are released at https://github.com/AMPICLab/AMPICBench.

Introduction

An analog block reaches tape-out only after two nested loops close: the designer sizes, compensates and re-simulates the circuit until it meets its specifications over process, voltage and temperature (PVT), then draws, verifies and extracts the layout, and returns to the schematic when parasitics or layout adjustments move the results. Automating both loops end to end remains a long-standing goal of analog designers and EDA developers.

LLM-based flows now generate topologies and netlists [1, 2, 3], port amplifiers across processes [4] and assist layout [5, 6, 7], alongside algorithmic layout tools such as ALIGN [8] and MAGICAL [9]. For sizing, most flows place an optimizer in a loop: the LLM proposes or narrows a search space, and Bayesian optimization, trust-region search or the LLM itself iterates over the device dimensions [10, 11, 12, 13]. These loops spend many simulations rediscovering what a designer knows from first principles, usually run at the typical corner, and stop at the schematic.

The OPERA Flow

Figure 1 depicts our end-to-end flow. The agent is a general-purpose LLM (Claude, Anthropic) in a Linux sandbox with shell, file and simulation tools, without training or fine-tuning. It uses open-source tools, i.e., ngspice [14], Magic and Netgen [15], KLayout [16], SchemDraw [17], the open SKY130 models [18], and gLayout [6, 19]. We as the designer do not edit netlists or layouts; each design starts from one directive, such as “design a Folded-cascode OTA for 50 dB gain, fun≥50f_\mathrm{un} \ge50 MHz for a 1 pF load, and SR>>25 V/μ\mus for minimum static power consumption.” The designer’s lessons are kept in five skill files (sizing, compensation, gated layout flow, SKY130 layout rules, opamp layout guidelines).

Figure 1. The OPERA flow. (1) The designer gives the specifications, teaches skills and decides at the gates. (2) A general-purpose LLM agent applies the taught skills and (3) sizes the circuit with the six-step gm/IDg_m/I_D flow (specifications to gmg_m and minimum currents, per-role gm/IDg_m/I_D, LL, sizing from tables, checks, trade-offs), leaving a few knobs. (4) Simulation over 15 PVT corners returns worst-corner feedback to those knobs. (5) The layout generator builds the cell after the unit and floorplan gates, and (6) DRC in two decks, LVS, extraction and post-layout PVT simulation sign it off or trigger regeneration. (7) Each signed-off cell is distilled into a reusable, tested generator that the next opamp reuses.

Designer-taught skills and decision gates

We used Claude’s Opus 5.5 model for the LLM teaching phase and skill development, then switched to the lower-token-cost Sonnet 5.5 model to run the iterative loops. Table 1 lists what the designer taught through plain language, lecture slides, short recipes and worked examples. Each lesson became a skill file of rules and checks that later sessions reload. The designer intervenes only at defined gates: the per-role gm/IDg_m/I_D choices, any sizing change, any accepted corner failure and, in layout, the unit fingering (gate 1) and the floorplan (gate 2). At each gate the agent presents two or three simulated options.

SkillTaught by the designerUsed by the agent as
TopologySignal path, gain and pole nodes, stack headroom, replica bias, folding currentsHeadroom budget per stack
gm/IDg_m/I_DSix-step flow: specs to gmg_m and currents, gm/IDg_m/I_D per role, LL, size, check, exploreDensity, fTf_T, gmrog_mr_o from tables; sizes from equations; worst-corner knob iteration
CompensationTwo-stage Miller recipe: m=gm2/gm1m = g_{m2}/g_{m1}, CcC_c, RzR_z placementStarting CcC_c, RzR_z, mm; one knob per specification
LayoutFloorplans; common centroid, interdigitation, dummies; rings per stage; packing ≥70%\geq70\%; short parasitics sensitive nodesGenerator templates; matching and packing checkers

Table 1. SKILLS TAUGHT BY THE DESIGNER

Equation-First Sizing

Characterization

The agent studied the author’s lecture slides [20] and the gm/IDg_m/I_D literature [21], then characterized the SKY130 low-VthV_{\mathrm{th}} devices at the planned finger width (3.11 μ\mum NMOS, 2.48 μ\mum PMOS), and built lookup tables of current density, fTf_T, gmrog_mr_o, and capacitance against gm/IDg_m/I_D for several LL at five process corners.

The gm/IDg_m/I_D Design Flow

The designer taught the agent the six steps designers follow [21] (Figure 1, panel 3):

  1. Specifications to gmg_m and currents. Gain-bandwidth or unity-gain frequency (funf_{\mathrm{un}}) sets gm1=2πfun⋅CLg_{m1}=2\pi f_{\mathrm{un}}\cdot C_L for single-stage, and gm1=2πfun⋅CLg_{m1}=2\pi f_{\mathrm{un}}\cdot C_L for two-stage amplifier; the phase margin sets gm2g_{m2} through the non-dominant pole, and the slew rate sets the minimum currents. The first stage slews by steering its tail current into the compensating capacitance, so Itail≥SR⋅CLI_{\mathrm{tail}}\geq\mathrm{SR}\cdot C_L in a single-stage and Itail≥SR⋅CcI_{\mathrm{tail}}\geq\mathrm{SR}\cdot C_c in a two-stage amplifier. In a folded-cascode (FC) stage, folding branches must carry at least ItailI_{\mathrm{tail}}, and a class-A output stage must sink at least SR(CL+Cc)\mathrm{SR}(C_L+C_c). For 50 MHz and 25 V/μ\mus into 1 pF, a single stage needs gm1≥314 μSg_{m1}\geq314\ \mu\mathrm{S} and Itail≥25 μAI_{\mathrm{tail}}\geq25\ \mu\mathrm{A}; as gm1=(gm/ID)1Itail/2g_{m1}=(g_m/I_D)_1I_{\mathrm{tail}}/2, SR/fun=4π/(gm/ID)1\mathrm{SR}/f_{\mathrm{un}}=4\pi/(g_m/I_D)_1 caps (gm/ID)1(g_m/I_D)_1 at 25.

  1. gm/IDg_m/I_D per device role. High for the input pair (gmg_m per current, noise, offset), low for mirrors and current sources (for matching, as σΔI/I∝(gm/ID)σVT\sigma_{\Delta I/I}\propto(g_m/I_D)\sigma_{V_T}, noise and area considerations), moderate to high for cascodes (headroom), and capped where a device’s fTf_T sets a signal-path pole. Devices that copy a current share gm/IDg_m/I_D and LL, so the copy holds over corners.

  1. Channel length. LL follows from the gain budget (gmrog_mr_o against gm/IDg_m/I_D), traded against fTf_T and matching.

  1. Sizing from the tables. ID=gm/(gm/ID)I_D=g_m/(g_m/I_D), or the branch current, and W=ID/(ID/W)W=I_D/(I_D/W), rounded to whole fingers. About a dozen continuous widths, lengths and currents reduce to a few knobs: the gm/IDg_m/I_D of each role, the branch currents and, in a two-stage amplifier, m=gm2/gm1m=g_{m2}/g_{m1}, CcC_c and RzR_z.

  1. Checks and iteration. Headroom (VD,sat≈2/(gm/ID)V_{D,\mathrm{sat}}\approx2/(g_m/I_D)) and pole estimates come first. The equations using the characterization tables predicted gain and power within 2% but overestimate the phase margin by about 18°, so stability closes in simulation, one knob per specification at the worst corner, with secant updates for continuous knobs and bisection for integer ones.

  1. Trade-offs. Design charts (power, gain, swing, fTf_T against gm/IDg_m/I_D) yield two or three candidates for a gate.

Compensation recipe

For two-stage amplifiers the agent applies the designer’s recipe, based on the two-pole Miller model,

PM≈90∘−tan⁡−1(C2mCc)−tan⁡−1(1m),(1)\mathrm{PM}\approx90^\circ-\tan^{-1}\left(\frac{C_2}{mC_c}\right)-\tan^{-1}\left(\frac{1}{m}\right), \tag*{(1)}

with m=gm2/gm1m = g_{m2}/g_{m1}. With Rz=1/gm2R_z = 1/g_{m2} the zero term of (1) vanishes and 60∘60^\circ needs mCc/C2≥1.73mC_c/C_2 \ge1.73; a larger RzR_z moves the zero into the left half plane just above funf_{un}; the third pole 1/(2πRzC1)1/(2\pi R_z C_1) must stay well above funf_{un}. The agent uses (1) for starting values and knob directions and closes the phase margin in simulation.

Closing the Layout and Extraction Loop

Learning the layout framework

The agent learned gLayout [6] from its examples and documentation and found its gaps by running DRC on small pieces: gLayout’s layer names were offset by one from SKY130, so the agent wrote an explicit layer map and routes every net as a Manhattan path with via stacks. It then wrote a row generator for matched transistor arrays and uses Magic’s PDK generators for resistors and MIM capacitors.

Taught layout practice and gates

The designer taught layout from examples and know-how (Table I). For a two-stage opamp: the input pair as a two-dimensional common centroid; the load mirror and tail interdigitated with dummies; each stage in its own guard rings; the Miller capacitor between the second-stage PMOS and NMOS rings, low-impedance plate on the output; and the nulling resistor between the stages, so high-impedance nodes stay short. At gate 1 the agent proposes unit fingerings and simulates each over the corners, since finger widths move binned models more than parasitics do; at gate 2 it compares two or three floorplans by area, utilization and routing cost; gate 3 is the full layout and its verification.

A reusable layout generator

The agent then distilled its hand-scripted layouts into a generator for the two-stage family. Its inputs are the reference netlist and a layout intent file recording the designer’s decisions: unit sizes, matched-set patterns, passive placement and pin edges. One command regenerates the cell and reports PASS only if Magic and KLayout DRC are clean, Netgen LVS matches, the layout satisfies every item of the intent file, the matching and packing checks pass, and the extracted netlist meets the design targets over the corners. Finished blocks (bias, reference, opamp) are then assembled hierarchically: each block exposes its rails and pins as a fixed contract, neighbouring blocks abut and share supply rails, the remaining nets are routed and the flat cell is re-verified with DRC, LVS and extraction.

Case Study

Specifications and shared bias

All amplifiers target open-loop fun≥50f_{un} \ge50 MHz with CL=1C_L = 1 pF in a loop-gain analysis [24] testbench similar to Spectre’s stb analysis, PM ≥60∘\ge60^\circ, SR ≥25\ge25 V/μ\mus for a 0.7-to-1.1-V follower step (20–80%), and a minimum DC gain A0A_0 of 30 dB (5T), 50 dB (FC and telescopic OTAs) or 60 dB (two-stage opamps), VDD = 1.8 V and VCM=0.9V_{CM} = 0.9 V (testbenches and netlists released as a benchmark [25]), biased on chip by a constant-gmg_m beta-multiplier reference (BMR) feeding a wide-swing cascode bias [26], [27]. As the BMR transconductance tracks $1/R$, the agent chose an n+ diffusion resistor (+1435 ppm/K) over high-resistance poly. The short-channel BMR [20]` provides a 20.0 μ\muA reference at tt (19.1 to 21.3 over corners), with loop PM 72.4 to 74.6°.

Schematic results

The agent designed six topologies from the designer’s Cadence starting points in SKY130 PDK: 5T, folded-cascode (FC) and telescopic OTAs (the latter with a replica half-circuit for its cascode bias), a Miller opamp, and an FC or telescopic stage followed by a class-A stage. Figure 2(a) shows the FC two-stage. Table III lists the knobs and the results over 15 corners. Every design meets fun≥50f_{un} \ge50 MHz, PM ≥60∘\ge60^\circ, SR ≥25\ge25 V/μ\mus and its A0A_0 target at every corner. The lowest slew rates, 27.1 and 28.6 V/μ\mus, are in the two-stage opamps with cascode first stages, set by the first-stage current charging CcC_c. One limit remains: at fs, 70°C the weakest cascode is 23 to 63 mV into triode in five designs, yet all targets are met.

CircuitKnobsa^aA0A_0 (dB)funf_{un} (MHz)PM (°)SRb^b (V/μ\mus)IDDI_{DD} (μ\muA)
5T OTAnn31.1–31.868.1–76.082.4–84.135.4126–144
FC OTAn1,n2n_1,n_251.1–53.066.5–77.467.6–72.633.2282–358
Telescopicnn51.9–55.096.6–115.271.2–77.941.4243–307
Millern,m,Cc,Rzn,m,C_c,R_z60.8–64.082.2–100.869.1–75.746.1338–422
FC 2-stagen,m,Cc,Rzn,m,C_c,R_z63.2–68.257.5–72.870.2–78.928.6330–424
Tele. 2-stagen,m,Cc,Rzn,m,C_c,R_z68.7–78.075.5–95.866.8–75.627.1330–423

Table III. Opamp results over 15 PVT corners (schematic, CL=1C_L=1 pF, VDD=1.8 V, VCM=0.9V_{CM}=0.9 V, on-chip bias). a^a nn: pair fingers, n1/n2n_1/n_2: pair/folding fingers. b^b Worst edge, 0.7–1.1 V step, 20–80%. Shading: worst-corner margin. IDDI_{DD} includes the bias. Corners: tt, ff, ss, fs, sf at 0/27/70 °C. A0A_0 targets: 30, 50, 60, 60, 60 dB.

:::ેણ

Figure 2. Two-stage opamp with a folded-cascode first stage and a class-A second stage, designed by OPERA. (a) Schematic drawn by the agent from its sized netlist, devices colored by role; labels are finger counts. (b) Its automatically generated layout; DRC/LVS clean and meets specs in all corners.

AutoSizer [13] draws its optimizer from a pool that includes Bayesian optimization and simulates only the typical corner.

WorkTopo.Sizeb^bgm/IDg_m/I_DComp.PVTLayoutSign-offc^cReus. gen.Taught skills
AnalogCoder [1]✓L×××××××
ADO-LLM [10]OO×××××××
LEDRO [12]OO×××××××
AutoSizer [13]OO×××××××
Artisan [22]EE✓O×××××
AmpAgent [4]EL×O×××××
EEsizer [11]LL×××××××
White-box [23]LL✓O×××××
Analog.Ag [3]OO×××××××
LayoutCop. [5]×××××✓OO×
Glayout NL [6]×××××✓O××
ACDC [7]O××××××××
OPERA (ours)Oa^aE✓✓✓✓✓✓✓

Table II. Comparison with LLM-based analog design flows. ✓ Yes, O Partly, × No. a^a Topologies chosen by the designer and taught to the agent. b^b O: optimizer loop; L: LLM-driven iteration; E: equations, then a few knobs. c^c DRC, LVS and extraction over PVT.

CircuitOPERAᵃ A / BProtocol A (spec) RandomProtocol A (spec) TPEProtocol B (sat.) RandomProtocol B (sat.) TPE
S1 5T OTA54 / 545/5 (58)5/5 (48)5/5 (76)5/5 (48)
S1 FC OTA30 / 593/3 (250)3/3 (129)1/3 (368)1/3 (32)
S1 Telescopic30 / 593/3 (35)3/3 (45)2/3 (258)2/3 (41)
S2 Miller30 / 533/3 (40)3/3 (204)3/3 (40)2/3 (181)
S2 FC 2-stage30 / 552/3 (281)3/3 (141)0/30/3
S2 Tele. 2-stage30 / 553/3 (217)3/3 (166)1/3 (352)0/3

Table 4. Runs to first passing design: passing seeds (median runs)

CellArea (μm2\mu\mathrm{m}^{2})Util.DRCᵃ LVSA0A_{0} (dB) PEXPM (°) PEXfunf_{\mathrm{un}} (MHz) PEXSR (V/μ\mus)
S1 5T OTA49689.2%0/0 ✓31.1–31.881.8–83.566.5–74.334.3
S1 FC OTA348269.0%0/0 ✓51.0–53.063.5–68.663.5–74.031.5
S1 Telescopic301586.8%0/0 ✓51.9–54.966.2–72.992.4–109.941.4
S2 Miller235678.3%0/0 ✓60.8–64.064.6–71.078.8–95.646.1
S2 FC 2-stageᶜ357066.9%0/0 ✓63.2–68.362.1–70.753.7–68.027.9
S2 Tele. 2-stage289357.3%0/0 ✓68.5–78.063.1–72.771.1–89.530.5
B BMR109276.5%0/0 ✓—61.2–65.5ᵇ52.8–81.2ᵇ—

Table 5. Layouts and post-layout results (SKY130, extracted R and C, ideal RzR_z and CcC_c, 15 corners)

To see how such optimizers fare over PVT, random search and the tree-structured Parzen estimator (TPE) [28], a Bayesian method, in Optuna [29], sized the same six circuits with the same bias, testbenches and 15 corners (3 seeds, 5 for the 5T; 500-run cap). It uses the first-campaign spec (no gain target). The cost is ngspice runs, a candidate stops at its first failing corner, and a pass means all 15 corners; OPERA is counted the same way on its stored sweeps. Protocol A is the specification plus the OPERA IDDI_{\mathrm{DD}}; B adds a worst saturation margin ≥−50\ge-50 mV (5T: ≥0\ge0). Under A, TPE passes every circuit in 45 to 204 runs (median) against 30 (54 for the 5T) for OPERA, but with margins down to −169-169 mV. Under B it passes 5 of 15 runs on the five larger circuits and never on the FC and telescopic two-stage designs, where OPERA needs 55 (Table IV).

Layout results

Figure 2(b) shows the FC two-stage layout, and Table V summarizes the six layouts and the bias cell. All six are DRC clean in both decks and LVS matched, and their extracted netlists meet every target at all 15 corners. Extraction lowers funf_{\mathrm{un}} by 2 to 7% and PM by 1 to 8°, and moves A0A_{0} by 0.1 dB at most. Three first extractions missed the PM (FC 57.2, telescopic 53.3, FC two-stage 59.4°); shorter source LL (FC, 1.0 μ\mum), shorter cascode LL (telescopic, 0.6 μ\mum) and Rz=3.7R_z = 3.7 kΩ (FC two-stage) fixed them. Layout area utilization is 57 to 89%. Telescopic layouts hold the bias replica in

the same arrays as the devices it copies; pairs, sources and cascodes are centroided, single-unit devices are not.

Comparison with Prior Work

Table II positions this work. LEDRO [12] and Auto-Sizer [13] use the LLM to steer an optimizer over the device dimensions; Artisan [22] and [23] give the LLM gm/IDg_m/I_D data, but it still chooses each width. In OPERA the designer teaches the gm/IDg_m/I_D flow, so the agent sizes from equations and searches a few knobs. Layout agents [5, 6, 7] do not close the loop through extraction and PVT; OPERA does, and leaves a reusable generator. Its scope, one designer and one generator family, keeps results traceable, and extending to more designers and families is the next step.

Conclusion

OPERA shows that a general-purpose LLM agent, taught the gm/IDg_m/I_D flow and layout practice by a designer, can take CMOS opamps from specifications to post-layout sign-off with open-source tools. In SKY130, six opamps on a shared constant-gm bias meet their dc-gain targets, fun≥50f_{\mathrm{un}} \ge50 MHz, SR ≥25\ge25 V/μ\mus and 60° of phase margin (for two-stage) at all 15 PVT corners, in schematic and in extracted layouts. The benchmark circuit netlists are available at [25].

References

References

  1. [1]Y. Lai, S. Lee, G. Chen, S. Poddar, M. Hu, D. Z. Pan, and P. Luo, “AnalogCoder: Analog circuit design via training-free code generation,” in Proc. AAAI Conf. Artificial Intelligence, vol. 39, no. 1, 2025, pp. 379–387.
  2. [2]J. Gao, W. Cao, J. Yang, and X. Zhang, “AnalogGenie: A generative engine for automatic discovery of analog circuit topologies,” in Proc. Int. Conf. Learning Representations (ICLR), 2025.arxiv.org/abs/2503.00205
  3. [3]Z. Bao, Z. Lin, J. Wang, J. Hu, Y. Gao, Y. Wu, X. Li, and X. Xu, “AnalogAgent: Self-improving analog circuit design automation with LLM agents,” arXiv:2603.23910, 2026.arxiv.org/abs/2603.23910
  4. [4]C. Liu, W. Chen, A. Peng, Y. Du, L. Du, and J. Yang, “AmpAgent: An LLM-based multi-agent system for multi-stage amplifier schematic design from literature for process and performance porting,” arXiv:2409.14739, 2024.arxiv.org/abs/2409.14739
  5. [5]B. Liu, H. Zhang, X. Gao, Z. Kong, X. Tang, Y. Lin, R. Wang, and R. Huang, “LayoutCopilot: An LLM-powered multi-agent collaborative framework for interactive analog layout design,” arXiv:2406.18873, 2024.arxiv.org/abs/2406.18873
  6. [6]A. Hammoud, C. Goyal, S. Pathen, A. Dai, A. Li, G. Kielian, and M. Saligane, “Human language to analog layout using Glayout layout automation,” Google Research publication, 2024.
  7. [7]Y. Esfandjari, J. Rego, A. Meyer, J. Gallagher, and M. Levy, “LLMs for analog circuit design continuum (ACDC),” arXiv:2512.09199, 2025.
  8. [8]K. Kunal, M. Madhusudan, A. K. Sharma, W. Xu, S. M. Burns, R. Harjani, J. Hu, D. A. Kirkpatrick, and S. S. Sapatnekar, “ALIGN: Open-source analog layout automation from the ground up,” in Proc. ACM/IEEE Design Automation Conf. (DAC), 2019.
  9. [9]B. Xu, K. Zhu, M. Liu, Y. Lin, S. Li, X. Tang, N. Sun, and D. Z. Pan, “MAGICAL: Toward fully automated analog IC layout leveraging human and machine intelligence,” in Proc. IEEE/ACM Int. Conf. Computer-Aided Design (ICCAD), 2019, pp. 1–8.DOI
  10. [10]Y. Yin, Y. Wang, B. Xu, and P. Li, “ADO-LLM: Analog design Bayesian optimization with in-context learning of large language models,” in Proc. IEEE/ACM Int. Conf. Computer-Aided Design (ICCAD), 2024.DOI
  11. [11]C. Liu and D. Chitnis, “ECsizer: LLM-based AI agent for sizing of analog and mixed signal circuit,” arXiv:2509.25510, 2025.DOI
  12. [12]D. V. Kochar, H. Wang, A. P. Chandrakasan, and X. Zhang, “LEDRO: LLM-enhanced design space reduction and optimization for analog circuits,” 2024.arxiv.org/abs/2411.12930
  13. [13]X. Yu, D. Torbunov, S. Mandal, and Y. Ren, “AutoSizer: Automatic sizing of analog and mixed-signal circuits via large language model (LLM) agents,” 2026, to appear at ICML 2026.arxiv.org/abs/2602.02849
  14. [14]H. Vogt, G. Atkinson, and P. Nenzi, “Ngspice user’s manual, version 42,” https://ngspice.sourceforge.io, 2024.
  15. [15]R. T. Edwards, “Magic VLSI layout tool and netgen LVS,” http://opencircuitdesign.com.
  16. [16]M. Kӧfferlein, “KLayout: GDS2 and OASIS viewer and editor,” https://www.klayout.de.
  17. [17]C. J. Delker, “SchemDraw: Electrical circuit schematic drawing in Python,” https://github.com/cdelker/schemdraw, 2026, version 0.23, MIT license.
  18. [18]SkyWater Technology and Google, “SkyWater SKY130 open source process design kit,” https://github.com/google/skywater-pdk, 2020.
  19. [19]“OpenFASoC: Fully open-source autonomous SoC synthesis using customizable cell-based synthesizable analog circuits,” https://github.com/idea-fasoc/OpenFASOC, includes the gLayout generator.
  20. [20]V. Saxena, “CPEG 424/624: Analog integrated circuit design,” Course website, University of Delaware, Spring, 2020, accessed: Oct. 27, 2026. [Online]. Available: https://www.eecis.udel.edu/~vsaxena/courses/eleg624/f20/eleg624.htm
  21. [21]P. G. A. Jespers and B. Murmann, Systematic Design of Analog CMOS Circuits Using Pre-Computed Lookup Tables. Cambridge University Press, 2017.
  22. [22]Z. Chen, J. Huang, Y. Liu, F. Yang, L. Shang, D. Zhou, and X. Zeng, “Artisan: Automated operational amplifier design via domain-specific large language model,” in Proc. ACM/IEEE Design Automation Conf. (DAC), 2024.DOI
  23. [23]J. Chen, S. Li, and X. He, “White-box reasoning: Synergizing LLM strategy and gm/Id data for automated analog circuit design,” 2025.arxiv.org/abs/2508.13172
  24. [24]M. Tian, V. Visvanathan, J. H. Jangtan, and K. Kundert, “Striving for small-signal stability,” IEEE Circuits and Devices Magazine, vol. 17, no. 1, pp. 31–41, 2001.
  25. [25]V. Saxena, “AMPICBench: Analog Circuit Benchmarks for Sizing, Simulation, and Layout Tools,” Oct. 2026. [Online]. Available: https://github.com/ampiclab/AMPICBench
  26. [27]V. Saxena and R. J. Baker, “Indirect Feedback Compensation of CMOS Op-Amps,” in Proceedings of the IEEE/EDS Workshop on Microelectronics and Electron Devices (WMED), 2006, pp. 3–4.DOI
  27. [28]J. Bergstra, R. Bardenet, Y. Bengio, and B. Kégl, “Algorithms for hyper-parameter optimization,” in Advances in Neural Information Processing Systems (NIPS), vol. 24, 2011, pp. 2546–2554.
  28. [29]T. Akiba, S. Sano, T. Yanase, T. Ohta, and M. Koyama, “Optuna: A next-generation hyperparameter optimization framework,” in Proc. ACM SIGKDD Int. Conf. Knowledge Discovery & Data Mining, 2019, pp. 2623–2631.arxiv.org/abs/1907.10902

Paper details

vlnc:W_2yfcmwza
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Not recorded for this paper

View statement
Looking for related papers…
Contents