Keep writing your models in Python. AADC records the calculation once and compiles it to machine code, so Monte Carlo and risk run up to 1000x faster and every sensitivity comes with the price. Measured 10x+ faster than JAX, PyTorch and TensorFlow on quant workloads.
Up to 1000x faster simulations with derivatives, 10x+ faster than JAX, PyTorch and TensorFlow
AADC supports mixed-language computational graphs, combining C++ and Python code into a single optimised kernel. The solution enables direct interaction between Python and C++ components in quantitative libraries. Functions can be recorded across Python and C++ and used to accelerate Monte Carlo simulations and computing sensitivities (Greeks) with unprecedented efficiency using automatic adjoint differentiation. This tool not only represents a leap in computational capability but also a significant stride towards optimising developers' time and resources in financial modelling.
Mark inputs/outputs and record computational graphs that span both Python and C++ code
AADC's JIT compiler creates a single optimised kernel from the mixed-language graph
Call the unified kernel from any supported language with full performance
Python front-end calling C++ pricing engines with unified AAD
Prototype in Python, optimise critical paths in C++, deploy as unified kernel
Wrap existing C++ libraries for Python access without performance penalty
Python Monte Carlo up to 1000x faster, which turns an overnight batch into something you can run while the market is open.
Every Greek and sensitivity from the same recording, exact rather than bumped, using MatLogica's patented Code Generation AAD.
The same workload at a tenth of the compute. A compiled kernel can also be saved and shipped to a cloud node without the model source.
A calculation that takes a tenth of the CPU time takes a tenth of the power. For a bank running Monte Carlo every night, that is a measurable number.
Existing NumPy code records as it is. NumPy ufuncs and array functions are supported out of the box.
Derivatives pricing, Monte Carlo, risk, calibration, and any other numerical Python that runs the same calculation many times.
An existing model is usually running compiled within days
The change to your code is a type: inputs become AADC values, the function is recorded, and from then on it replays compiled with derivatives. The optimisation and the AAD that used to take a specialist months come with the recording.
Find the right implementation path for your Python projects
Accelerate existing C++ and Python codebases without rewrites
Build new applications with performance and AAD built-in
Supercharge QuantLib, ORE, and other open-source libraries
Reduce cloud costs by 50-99% with optimised compute
How the record-once replay-many paradigm accelerates Python and C++ analytics on Intel CPUs
Try it with pip install aadc, or book a demo and we run it on one of your models.
MatLogica Python Accelerator delivers over 1000x performance improvements for Python quantitative models and Monte Carlo simulations compared to vanilla Python. For existing projects, even those with tens of millions of lines of code, thousand-fold speed enhancements can be achieved in just days of integration effort.
Independent benchmarks show MatLogica Python Accelerator is 10x+ faster than JAX, PyTorch, and TensorFlow for quantitative finance workloads. While ML frameworks are optimised for large tensor operations, Python Accelerator is specifically designed for quantitative finance applications with scalar operations and complex computational graphs typical in derivatives pricing and Monte Carlo simulations.
Yes, Python Accelerator provides advanced NumPy support with compatibility for NumPy ufuncs and functions out of the box. You can integrate with existing Python/NumPy code, enhancing performance without a complete overhaul of your quantitative codebase.
Yes, Python Accelerator enables cross-language interaction between Python and C++ components. Functions can be recorded across both Python and C++ and used to accelerate simulations and compute sensitivities with automatic adjoint differentiation. This cross-language capability is unique and enables leveraging existing mixed codebases.
Organisations using MatLogica Python Accelerator achieve 90% or more cloud computing cost reductions for quantitative workloads. The dramatic performance improvements (1000x faster) mean you need significantly fewer cloud resources to run the same Python simulations, directly translating to reduced infrastructure costs.
Yes, Python Accelerator includes built-in Automatic Adjoint Differentiation (AAD) straight out of the box, leveraging MatLogica's patented Code Generation AAD™ technology. This enables computing Greeks and sensitivities for derivatives pricing without manual implementation, a capability that vanilla Python lacks and that traditionally required extensive expertise.
Python Accelerator can be integrated into existing Python projects in just days of effort, even for codebases with tens of millions of lines. This stands in stark contrast to traditional optimisation approaches that require weeks or months of work to achieve only modest 10% performance gains. The accelerator provides thousand-fold speed enhancements with minimal code changes.