L-BFGS

Software for Large-scale Unconstrained Optimization by Jorge Nocedal

L-BFGS is a limited-memory quasi-Newton code for unconstrained
optimization. The code has been developed at the Optimization Center,
a joint venture of Argonne National Laboratory and Northwestern
University.  Downloading and Installing L-BFGS

The original package includes the following clause:

  Condition for Use: This software is freely available for educational
  or commercial purposes. We expect that all publications describing
  work using this software quote at least one of the references given
  below. This software is released under the BSD License.

References

J. Nocedal. Updating Quasi-Newton Matrices with Limited Storage
(1980), Mathematics of Computation 35, pp. 773-782.

D.C. Liu and J. Nocedal. On the Limited Memory Method for Large Scale
Optimization (1989), Mathematical Programming B, 45, 3, pp. 503-528.

---

This is a repackaging of the original code [1] that includes a C/C++
header file and a CMake-based build system.

[1] http://users.iems.northwestern.edu/~nocedal/lbfgs.html