What is MatLogica AADC?

MatLogica AADC is a just-in-time compiler that transforms object-oriented code (C++, C#, Python) into highly optimized machine code with automatic adjoint differentiation. It achieves 6-100x speedup and adjoint factor less than 1, meaning original function plus all derivatives compute faster than the original alone.

How much performance improvement can I expect with AADC?

Typical improvements include: 6-100x faster analytics and simulations, 50% reduction in cloud costs, 3x faster new model development, 30x faster on-demand calculations, and 10x more throughput from existing hardware through AVX2/AVX512 vectorization and multi-threading.

What is Code Generation AAD and how is it different?

Code Generation AAD generates optimized machine code at runtime rather than using tape recording. This pioneering approach achieves adjoint factor less than 1 (faster than original code), accelerates scenarios and pricing (not just Greeks), requires minimal code changes, and delivers 20-50x better performance than tape-based AAD.

Can AADC reduce my cloud computing costs?

Yes, clients achieve 50-99% cloud cost reduction. Example: £12M+ annual savings through computational efficiency, reduced instance requirements, and optimized resource utilization. AADC's binary kernels enable secure cloud execution without exposing proprietary models.

MatLogica | MatLogica - Fast AAD Compiler for Quantitative Finance

Name: MatLogica AADC
Rating: 5 (15 reviews)
Author: MatLogica

MatLogica offers an award-winning toolkit that is 40x faster than PyTorch, JAX and TensorFlow, transforming complex model development.

By automatically enabling best-possible performance for simulations, repetitive calculations, and fastest automatic differentiation, it reduces architectural complexity, accelerates time-to-market, yields orders of magnitude performance improvements, and drives substantial cloud cost savings.

Game-Changing Results Our Clients Have Already Achieved:

50% cloud cost reduction

Over £12M annual cloud savings through computational efficiency

3x faster new model development

Up to 75% of modeling codebase eliminated through simplified development

30x faster on-demand simulations

New revenue streams accessed with faster intraday pricing & real-time risk

10x more from existing hardware

Improved operations and reduced risks without infrastructure investment

The Secret Sauce: Code Generation AAD™

MatLogica Developer Productivity: Object-Oriented Code to Data-Oriented Performance

MatLogica pioneered Code Generation AAD™ - the breakthrough approach that on-the-fly converts object-oriented code into bare-metal machine performance, designed for cross-platform execution.

The key innovation: A specialized JIT compiler that maximizes developer productivity while enabling the best possible runtime performance.

Adjoint factor less than 1: Original function plus all derivatives compute faster than original alone - impossible with traditional AAD approaches.

Whether developing new quant libraries or enhancing existing codebases, MatLogica enables groundbreaking team efficiencies and model execution speed!

Mix-mode execution: Seamlessly combine C++ and Python - define custom payoffs in Python, execute in optimized C++.

Live Risk

Example Use Case

Turn your batch-processing risk analytics into "always on" Live Risk server.

This example is based on Quantlib, supercharged by MatLogica AADC. Hit "Launch The Live Risk App" in the widget to see it live! You can access the code behind it using code examples on our website. You can read more about this using the link below.

Read now

Take Your Analytics to the Next Level

In as little as 1 month, MatLogica will boost simulation performance and enable Automatic Adjoint Differentiation with minimal code changes.

Custom-built JIT compiler enables AAD sensitivities computed faster than the original calculation!

Adjoint factor less than 1: Original function plus all derivatives faster than original alone.

The pioneering Code Generation AAD™ solution requires minimal refactoring, no template metaprogramming, and includes cutting-edge automated integration and debugging toolkits.

Key advantages:

Accelerates pricing, Greeks, AND scenarios
20-50x faster than tape-based AAD
Handles calibration automatically (AIFT)
Works with legacy code

Learn About Code Generation AAD

Our optimization layer unlocks:

Native CPU vectorization (AVX2 and AVX512) - 4-8 operations per cycle
Safe multi-threading - automatic parallelization
6-100x speedups on single core!
Linear scaling across multiple cores

Cross-platform execution: Choose the hardware most suited for your workloads with a flick of a switch.

CUDA users: Enable AAD for your CUDA code with minimal changes and run on CPU with comparable performance!

Result: GPU-like performance on CPU without GPU complexity or costs.

React to market conditions faster by simplifying your modeling codebase and accelerating prototyping and time-to-market.

Cloud cost reduction:

Use cloud resources more efficiently
Save up to 99% of cloud spend
Scale compute quickly and securely
Deploy encrypted binary kernels
Keep proprietary models on premises

Developer productivity: Focus on models, delegate optimization to AADC.

Developers Are More Productive with the Right Tools

Focus on models and delegate with confidence - our toolkit seamlessly handles performance optimizations and sensitivity computations, unlocking your full potential.

Quick & Easy
Integration

Simplified
Development

Past Initiative
Agnostic

Clean, Intuitive
Interface

Chartis QuantAnalytics50 2025 - MatLogica #13 Overall, Category Leader AAD

"MatLogica's suite of tools enables acceleration and scalability in both analytical and quantitative code. The transformational nature of MatLogica's offering is reflected in its top 15 placing in the QuantitativeAnalytics50, alongside several awards for innovation in computational frameworks and code acceleration."

"MatLogica's toolkit offers a groundbreaking approach to enabling analytics acceleration, developer productivity and hardware efficiencies. Moreover, with its simple but effective tools, performance optimizations can be flexibly implemented and integrated."

Antoine Savine - Quant and Mathematician

"MatLogica offers a unique AAD engine, bundled with a purposely developed just-in-time compiler, whereby banks can effortlessly optimize existing code with AAD, into machine code optimized for modern hardware. Because of this optimizing compiler, Matlogica's offering currently stands way above competition. I have seen the results and they are truly impressive."

"One notable feature of AADC is that complex model calibration procedures do not require any special attention as is often the case with other AAD approaches. For example, we "AADC-record" simple Dupire volatility calculations, regression-based continuation value calibration and even Monte-Carlo based Cheyette model calibrations, with no special care needed to back-propagate adjoints. This has allowed us to eliminate a large amount of complex and hard to maintain code."

Dr. Roland Olsson - University Researcher

"MatLogica offers a unique tool for developing machine learning applications in C++, C# or Python. It combines the simplicity of object-oriented design and the performance of graph-based languages. For applications like time series analysis, recursive neural nets, and similar, the performance of MatLogica AADC is impressive. Due to excellent vectorization and good support for multithreading, the execution on a CPU is very fast."

MatLogica AADC applications: automatic adjoint differentiation, code generation AAD, quantitative finance Greeks, XVA pricing CVA DVA, Live Risk real-time, Monte Carlo acceleration, model calibration automatic, cloud cost reduction, AVX2 AVX512 vectorization, multi-threading automatic, machine learning C++, PDE sensitivity analysis, FRTB sensitivities, stress testing acceleration, scenario analysis performance, derivative pricing Greeks, adjoint factor less than 1

Transform Analytics Performance with Code Generation AAD™