Finite Temperature Lanczos Method (FTLM)¶

Overview¶

The Finite Temperature Lanczos Method (FTLM) computes thermodynamic observables such as partition function, internal energy, and specific heat without explicit full diagonalization. It uses stochastic sampling over random initial vectors combined with Lanczos iteration.

API¶

import qkrylov as qk

result = qk.ftlm(H, beta=10.0, n_random=50, n_steps=100)
print(f"Z = {result.partition_function:.4f}")
print(f"E = {result.internal_energy:.4f}")
print(f"Cv = {result.specific_heat:.4f}")

auto result = ftlm(H, /*beta=*/10.0, /*n_random=*/50, /*n_steps=*/100);
std::cout << "Z = " << result.partition_function << "\n";
std::cout << "E = " << result.internal_energy << "\n";

Coming Soon¶

• Parameter guide: choosing n_random and n_steps
• Error estimation and convergence
• Temperature sweeps and thermodynamic curves