Skip to content

Add nablatensor to Financial - #1326

Open
petrpravda wants to merge 1 commit into
akullpp:mainfrom
petrpravda:add-nablatensor
Open

petrpravda wants to merge 1 commit into
akullpp:mainfrom
petrpravda:add-nablatensor

Conversation

@petrpravda

@petrpravda petrpravda commented Sep 15, 2026

Copy link
Copy Markdown

Suggestion type

  • Project
  • Resource

Checklist

  • I searched the list and existing issues for duplicates.
  • I changed README_SOURCE.md, not the generated README.md.
  • This pull request contains one suggestion.
  • The suggestion is relevant to Java or the JVM and fits its chosen category.
  • I used the canonical project or resource link.
  • The suggestion is current and maintained.
  • The concise, neutral description explains its distinguishing value and ends with punctuation.
  • Licensing is clear and any restrictive terms are disclosed where applicable.

Adds nablatensor under Financial.

It's an adjoint-automatic-differentiation library for quantitative finance on the JVM: a valuation is recorded once against a scalar type, then replayed from the same tape to get a price and every first-order Greek from one reverse sweep, on CPU, SIMD, or GPU (CUDA/Vulkan/ROCm).

  • Apache-2.0, single repo
  • On Maven Central (com.nablatensor:nablatensor-core)
  • mvn -o test is green with no GPU/native lib/incubator flag (plain-Java cpu-jit backend by default)

Disclosure: I'm the author of nablatensor.

Closest existing neighbor in the list is OpenGamma Strata (market-risk analytics), though nablatensor's focus is the AAD engine + multi-backend replay rather than a market-data/product model library.


Summary by cubic

Adds nablatensor to the Financial section of README_SOURCE.md. It's an adjoint-automatic-differentiation library for quantitative finance: a valuation is recorded once against a scalar type, then replayed from the same tape to get a price and every first-order Greek from one reverse sweep, on CPU, SIMD, or GPU. Apache-2.0, available on Maven Central as com.nablatensor:nablatensor-core, and mvn -o test passes with the default plain-Java cpu-jit backend. Disclosure: the submitter is the author of nablatensor.

Written for commit 491bbca. Summary will update on new commits.

Review in cubic

Adjoint automatic differentiation library for quantitative finance
(price + all Greeks from one reverse sweep), targeting CPU, SIMD and
GPU backends.

@cubic-dev-ai cubic-dev-ai Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No issues found across 1 file

Re-trigger cubic

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant