🌟 Our Vision

NEUROVEDIK represents a unique fusion of 5,000-year-old Vedic mathematical wisdom with cutting-edge hardware acceleration. We believe that the ancient mathematicians of India developed algorithms that were ahead of their time—algorithms that naturally map to the parallel processing capabilities of modern CPUs and GPUs.

"The Vedic sutras describe computational methods that seem almost prescient— cross-multiplication patterns that parallelize perfectly across SIMD lanes, as if the ancient sages could see centuries into the future of computing."

💡 Philosophy

🔬

Hardware-Aware Design

Every algorithm is designed with modern CPU architecture in mind—cache lines, SIMD registers, and branch prediction.

🔐

Security by Default

Cryptographic operations are constant-time by default, protecting against timing attacks without sacrificing performance.

Performance First

We optimize for the 95% case. The "Vedic sweet spot" (64-2048 bits) covers nearly all cryptographic operations.

🌍

Open & Accessible

Apache 2.0 licensed with comprehensive documentation. We want everyone to benefit from these optimizations.

🧮 The Vedic Connection

The Vedas are ancient Indian texts dating back to 1500-500 BCE. Within these texts, particularly the Atharva Veda, are mathematical sutras (concise rules) that describe remarkably efficient computational methods.

The 16 Vedic Sutras

NEUROVEDIK implements several of these sutras:

Urdhva Tiryakbhyam
"Vertically and Crosswise"
Used for multiplication—computes columns in parallel
Nikhilam Sutra
"All from 9 and the last from 10"
Fast division when divisor is near a power of 10
Anurupyena
"Proportionately"
Scaling operations and approximate arithmetic

🎯 Who Uses NEUROVEDIK?

🔐 Cryptography Engineers

Building TLS libraries, HSMs, and secure authentication systems that need high-throughput RSA and ECC operations.

🤖 AI/ML Researchers

Implementing efficient quantization, mixed-precision training, and hardware-accelerated inference engines.

⛓️ Blockchain Developers

Zero-knowledge proofs, ECDSA signatures, and on-chain computation where every cycle counts.

🧮 Scientific Researchers

Arbitrary-precision calculations for physics simulations, number theory research, and mathematical proofs.

🗺️ Roadmap

v1.0 - Current

Core arithmetic, cryptographic primitives, AVX-512 optimization, Python bindings

v1.1 - Q2 2026

ARM NEON support, WebAssembly build, improved documentation

v1.2 - Q3 2026

Multi-threaded Karatsuba, CUDA GPU acceleration, benchmarking suite

v2.0 - Q4 2026

Post-quantum cryptography (Kyber/Dilithium), homomorphic encryption primitives

🤝 Contributing

NEUROVEDIK is open source and we welcome contributions! Whether you're fixing bugs, improving documentation, or adding new features, your help is appreciated.

📜 License

NEUROVEDIK is released under the Apache License 2.0, which means:

  • ✅ Free for commercial and personal use
  • ✅ Can be modified and distributed
  • ✅ Patent rights granted
  • ✅ Attribution required

See the full LICENSE file for details.

📚 Additional Resources