========================================================================
NUMERICAL EXPERIMENTS FOR SEMISMOOTH NEWTON SOLVER & APPLICATION PROBLEMS
========================================================================


Overview
--------

This project contains MATLAB scripts for conducting numerical
experiments on: 
	- Experiments for Elliptope, QAP with facial reduction (QAPFR), and QAP without facial reduction (QAPnoFR); 
	- Experiments using random instances and an instance from protein side-chain positioning problem;
	- A test on a randomly generated instance.

The folder is organized into three main directories:

 ── SolversAux/  :    Core solver implementations & auxiliary functions
 ── testMCQAP/   :    Workflow for MCQAP instances & LaTeX table generation
 ── testsSF/     :    Experiments on random & protein instances



------------------------------------------------------------------------
CORE SOLVER
-----------

`SolversAux/semismoothsolverK_preconditioner.m`

An inexact semismooth Newton algorithm with preconditioning designed to solve SDP nearest-point/projection problems:
    min  1/2 ||v - x||^2   s.t.   A(x) = b,  sMat(x) >= 0

The solver is designed to find the point y satisfying the optimality conditions F(y) = A(v + A'*y)_+ - b = 0; see the manuscript for more detail.

------------------------------------------------------------------------
MAIN EXPERIMENTS
----------------

1. `testMCQAP`

Main controller: `testMCQAP.m`

   - Problem Classes: Elliptope (`MC`), QAP with FR (`QAPFR`), QAP without FR (`QAPnoFR`).
   - Objective Settings: Rank-one optimal matrix W vs. Random matrix W.
   - Solvers Evaluated: Semismooth Newton vs. SDPT3 (`sqlp.m`).
   - Output: Formatted LaTeX tables via `tablegeneratorMCQAP.m` shown in Section 6.2 of the paper.

2. `testsSF`

   Main controller: `table_file.m`

   - Random instances: Evaluates convergence across dynamic problem sizes.
   - Protein instance: Loads `1CRNdata.mat` via `tempLoadData.m`.
   - Output: Eigenvalue convergence plots and LaTeX tables in Section 6.1.1 and 6.1.2 of the paper.

3. `TestSemismoothSplitting.m` (A random instance test)

   Standalone script testing the semismooth splitting variant on single random instances.

