|
The Leibniz System is a software package created by Klaus Truemper
for the
development and implementation of logic-based intelligent
systems. The package covers various aspect of the construction
of such systems with modules for
- logic computation
- learning logic formulas from data
- discretization of data
- subgroup discovery from data
- data estimation by a lazy learner
- dimension reduction of models
- decomposition of graphs and matrices
- solution of constrained optimization problems involving single or
multiple objective functions
|
Downloading and Installation of Leibniz System
- Get the files
leibniz.complete.zip and
installation.pdf.
- Install the Leibniz System using the instructions in the file
installation.pdf.
- Optional: Send email to klaus@utdallas.edu with
"Leibniz System Download" in the subject line. We will inform you
about updates of the system. No other use will be made of the
information.
- More mirror sites: Are being developed.
The Derivative-Free Library (DFL) is available
here.
The library offers a variety of codes (freely available under the GPL) to
solve optimization problems when first order information on the objective
or constraint functions are not available. Such problems are also known as
black-box (BB) or simulation-based (SB) optimization problems. They are
ubiquitous in the real world and especially in industrial design and
production. In particular, the library offers:
|
- Local optimization solvers:
- SDBOX, an algorithm for bound constrained optimization problems
- SDPEN, an algorithm for general (inequality) constrained optimization problems
- DFN, a linesearch-based algorithm for nonsmooth constrained optimization problems
- SDMINMAX, an algorithm for finite minimax optimization problems
- Global optimization solvers:
- ACRS, an Adaptively Controlled Random Search algorithm for bound constrained global optimization problems
- DDFSA, a Simulated Annealing algorithm for bound constrained global optimization problems
- DFSA, a Simulated Annealing algorithm for general constrained global optimization problems
- DIRMIN, a DiRect algorithm with derivative-free local searches for bound constrained global optimization problems
- DIRDFN, a DiRect algorithm with derivative-free local searches for general constrained global optimization problems
- DIRECT, an implementation of the DiRect algorithm for bound constrained global optimization problems
- Mixed-integer optimization solvers:
- DFL box, a linesearch program for bound constrained Mixed Integer NonLinear Programming
- DFL gen, a linesearch program for general (inequality) constrained Mixed Integer NonLinear Programming
- Multiobjective optimization solvers:
- DFMO, a linesearch program for Multiobjective Optimization
|
The SWIM - A
Software
Suite to Unveiling Crucial Nodes in Complex Networks
SWItchMiner (SWIM) is a software suite for the identification of a small
pool of genes, called switch genes, which are likely to be critically
associated with drastic changes in many biological settings. This
procedure was set in studying grapevine genome [see Plant Cell],
where switch genes were found to be master regulators of the
previously reported transcriptome remodeling that marks the
developmental shift from immature to mature growth in grapevine.
In another study switch genes have been investigated in
different human cancer types and the results strongly support the
hypothesis of their key role in cancer development [see Nature Scientific Reports].
|
- Software requirements:
SWIM has been developed in MATLAB ®1 (version R2013a including the Bioinformatics and Statistics
Toolboxes) and tested on the following operative systems:
- OSX 10.9.5
- GNU/Linux Ubuntu 14.04
- Windows 10 Pro
- Setting up:
- Install MATLAB ® and the Bioinformatics and Statistics Toolboxes
- Download and unzip the compressed file SWIM.zip that is available here. This will create a folder named SWIM in the current directory.
- The User Guide can downloaded from here.
|
A new branch&bound algorithm for Standard Quadratic Programming (StQP)
problems is available here
|
StQP_BB is a branch&bound algorithm developped by G. Liuzzi (CNR-IASI),
M. Locatelli (University of Parma) and V. Piccialli (University of Rome
"Tor Vergata"), to solve difficult standard quadratic programming problems
to global optimality. StQP_BB main features are:
- implicit enumeration of all the KKT (stationary) points of the problem
- use of an efficient polyhedral bounding technique
- customizable B&B tree exploration (best-bound or depth-first policies)
- possibility to use binary or n-ary node generation
- use the power of Gurobi to efficiently solve the LP subproblems
|
