Skip to content

Secure numerical simulations using fully homomorphic encryption

Authors: Arseniy Kholod; Yuriy Polyakov; Michael Schlottke-Lakemper
Year: 2026
Venue: Comput. Phys. Commun. 318 (2026) 109868

type: application collection: sok-references status: auto

Links: DOI · arXiv:2410.21824

Summary

Auto-extracted summary (unreviewed)

Aimed at computational scientists, this work explores privacy-preserving numerical simulation of PDEs with CKKS, introducing the Julia packages OpenFHE.jl and SecureArithmetic.jl that wrap the OpenFHE C++ library. The authors benchmark the accuracy and runtime of individual CKKS operations (bootstrapping being by far the least accurate and slowest), add a CKKS-compatible circshift, and implement first-order upwind and second-order Lax-Wendroff finite-difference schemes for 1D/2D linear advection on encrypted data. Secure simulations match unencrypted results with relative errors around 1e-6 after bootstrapping but are roughly six orders of magnitude slower, and achieve less than single-precision accuracy.

Abstract

Data privacy is a significant concern when using numerical simulations for sensitive information such as medical, financial, or engineering data -- especially in untrusted environments like public cloud infrastructures. Fully homomorphic encryption (FHE) offers a promising solution for achieving data privacy by enabling secure computations directly on encrypted data. Aimed at computational scientists, this work explores the viability of FHE-based, privacy-preserving numerical simulations of partial differential equations. The presented approach utilizes the Cheon-Kim-Kim-Song (CKKS) scheme, a widely used FHE method for approximate arithmetic on real numbers. Two Julia packages are introduced, OpenFHE$.$jl and SecureArithmetic$.$jl, which wrap the OpenFHE C++ library to provide a convenient interface for secure arithmetic operations. With these tools, the accuracy and performance of key FHE operations in OpenFHE are evaluated, and implementations of finite difference schemes for solving the linear advection equation with encrypted data are demonstrated. The results show that cryptographically secure numerical simulations are possible, but that careful consideration must be given to the computational overhead and the numerical errors introduced by using FHE. An analysis of the algorithmic restrictions imposed by FHE highlights potential challenges and solutions for extending the approach to other models and methods. While it remains uncertain how broadly the approach can be generalized to more complex algorithms due to CKKS limitations, these findings lay the groundwork for further research on privacy-preserving scientific computing.
Source: arxiv

Extracted details

Field Value Source Confidence
contribution OpenFHE.jl and SecureArithmetic.jl Julia packages wrapping OpenFHE for secure arithmetic llm 0.85
contribution Accuracy and performance evaluation of key CKKS operations in OpenFHE, including bootstrapping llm 0.85
contribution CKKS-compatible circshift operation and encrypted finite-difference solvers (upwind, Lax-Wendroff) for linear advection in 1D and 2D llm 0.85
contribution Analysis of algorithmic restrictions imposed by FHE for extending to other models and methods llm 0.85
schemes CKKS llm 0.85
libraries OpenFHE llm 0.85
libraries OpenFHE.jl llm 0.85
libraries SecureArithmetic.jl llm 0.85
tasks privacy-preserving numerical simulation of PDEs (linear advection, finite differences) llm 0.85
security level 128-bit llm 0.85
threat model untrusted public cloud performing outsourced computation on encrypted data llm 0.85
interaction non-interactive FHE llm 0.85
code url https://github.com/hpsc-lab/paper-2024-secure_numerical_simulations (URL found in the full text) regex 0.50
code url https://github.com/Microsoft/SEAL (URL found in the full text) regex 0.50
code url https://github.com/hpsc-lab/SecureArithmetic.jl (URL found in the full text) regex 0.50
code url https://github.com/hpsc-lab/OpenFHE.jl llm 0.85
limitations Less than single-precision floating-point accuracy achievable with FHE llm 0.85
limitations Six orders of magnitude computational overhead llm 0.85
limitations Bootstrapping is the least accurate and slowest CKKS operation llm 0.85
limitations Uncertain generalization to more complex algorithms due to CKKS limitations llm 0.85
key result runtime overhead of secure vs unencrypted simulation: about six orders of magnitude (linear advection, upwind and Lax-Wendroff, 1D/2D) llm 0.85
key result relative L-infinity error introduced by CKKS: O(1e-13) before first bootstrapping, O(1e-6) after (first bootstrapping at ~40 steps (1D) / ~25 steps (2D)) llm 0.85
key result runtime ratio 2D vs 1D: 2-4x (encrypted simulations) llm 0.85

SoK reference [59]

A. Kholod, Y. Polyakov, and M. Schlottke-Lakemper, “Secure numerical simulations using fully homomorphic encryption,” Computer Physics Communications, vol. 318, p. 109868, 2026. [Online]. Available: https://www.sciencedirect.com/science/article/ pii/S0010465525003704

Listed on the reference page of the site.

Keyword mentions

Counts of vocabulary terms in the full text (first page of occurrence in brackets).

  • scheme: CKKS 117 [p.1], TFHE 8 [p.2], Scheme switching 6 [p.31], BFV 3 [p.2], BGV 3 [p.2], FHEW 3 [p.2], MPC 3 [p.1], Leveled / somewhat HE 3 [p.3], RSA 2 [p.37]
  • library: OpenFHE 75 [p.1], Microsoft SEAL 3 [p.8], HElib 3 [p.8], Phantom 1 [p.38]
  • model: Logistic regression 4 [p.27], Diffusion model 2 [p.35]
  • hardware: GPU 3 [p.34], FPGA 1 [p.34], ASIC 1 [p.34], CUDA 1 [p.38]
  • technique: Bootstrapping 127 [p.4], Rotations 43 [p.4], Key switching 14 [p.6], Rescaling 14 [p.4], Relinearization 4 [p.7], Functional bootstrapping 2 [p.34], Goldschmidt 2 [p.4], Newton iteration 2 [p.4], RNS 2 [p.5], Lookup table 2 [p.31]
  • assumption: RLWE 8 [p.3], Lattice-based 4 [p.2], LWE 3 [p.3]
  • task: Language modeling / generation 10 [p.5], Medical / healthcare 10 [p.1], Genomics 7 [p.2], Finance 6 [p.1]
  • topic: Homomorphic encryption 160 [p.1], Open source 9 [p.2], Privacy-preserving machine learning 2 [p.2]

Files

Path (under papers/) Collection Pages Size SHA-256 Note
references/sok-references/059_secure-numerical-simulations-using-fully-homomorphic-encrypt.pdf sok-references 38 2.2 MB f0e6a74cb934

Database key: secure-numerical-simulations-using-fully-homomorphic-encryption · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set secure-numerical-simulations-using-fully-homomorphic-encryption COLUMN VALUE