The independent verification of results is a critical step in the scientific process, and it would seem that achieving reproducibility should be much easier for computational scientists than for experimentalists. However, problems with reproducibility have received wide attention in recent years in many fields of computational inquiry, and some refer to the inability to routinely achieve reproducibility as a true crisis in computational science. Reproducibility in computational neuroscience requires descriptions of complex models that are precise and unambiguous. In this talk, I will discuss some recent neuroinformatics approaches aimed toward achieving reproducibility including model sharing databases, suggestions for publication standards, software for tracking computational "experiments", and simulator independent model description languages. In addition, I will provide an overview of the current status of efforts in our community and their trajectories.

The talk will demonstrate how a constructive version of the description logic ALC (Attribute Logic with Complement) can serve as a semantic type system for an extension of the simply typed lambda-calculus to express computations in knowledge bases. This cALculus embodies a functional core language which provides static type checking of semantic information processing of data whose structure is organised under a relational data model as used in description logics. The cALculus arises from a natural interpretation of the tableau rules for constructive ALC following the Curry-Howard-Isomorphism. We will argue that this modal type theory, due to its strong constructive interpretation, can deal with incomplete and dynamically changing data. This includes knowledge states of mobile agents which would be inconsistent in classical or even intuitionistic modal logics.

Autonomic systems are ones that exhibit the properties of self-organization, self-healing, self-optimization and self-protection. This talk explores some of the issues surrounding autonomic computing; namely the automatic distribution of an application across available resources. The goal is to provide the benefits of autonomic computing while at the same time relieving the applications programmer from the need to interact with complex middleware interfaces and deal with the mechanics of manual distribution.

This talk will examine 1) AdJava, an extension to Java to provide support to convert a multi-threaded application into a distributable one and inject autonomic properties, and 2) an agent-based approach to convert legacy Java systems (byte code) into distributed autonomic systems.

tba.

The Internet protocol stack has a layered architecture that resembles an hourglass. The lower and higher layers tend to see frequent innovations, while the protocols at the waist of the hourglass appear to be 'ossified'. The authors of this work have proposed a model by which this might occur, suggesting even that this ossification may be inevitable. This work is particularly interesting for the questions it tries to answer, as well as the strong feelings it evoked.

tba.

No-one seriously doubts that the brain is the site of the mind. But what is the nature of this sited-ness? What are the candidates? Von der Malsburg has a much-cited paper entitled "Am I thinking Assemblies" in 1988: but what does that imply? We are tempted to assume that what we can measure is what matters (whether that be local field potentials, spikes, fMRI measurements, or others. Indeed, much has been made of mind-reading machines based on fMRI in some parts of the press. But what are the alternatives, and why and how might we pursue this issue?