74
L. Rondoni
becuase of the large dimnsionality of the problems for which this approach is usually
invoked, i.e. those that are too complex to be tackled in a theory that starts from
an understanding of the fundamental mechanisms. The problem is enhanced in the
case of chaotic dynamics, that exponentially magnify in time the deviations from the
analog initial conditions. In principle, D equal to 6 or 7 is already out of reach in
those cases.
One could add further considerations to refute the idea that the scientific method
is made obsolete by the deluge of data. For instance, the tale attributed to Bertrand
Russell of a turkey before Thanksgiving day, shows how naive induction is doomed
to failure. The story goes like this: a turkey is fed every day. Every single feeding will
firm up the bird’s belief that it is the general rule of life to be fed every day by friendly
members of the human race “looking out for its best interests,” as a politician would
say. On the afternoon of the Wednesday before Thanksgiving, something unexpected
will happen to the turkey. It will incur a revision of belief …
Nevertheless, there are successes in using data that cannot be denied, and that can
certainly be integrated within a scientific framework, rather than opposing it; there
still is a lot to understand about what can be extracted from data, and proceeding by
examples is probably the most promising way to go. In fact, once again, it may be
beneficial to distinguish sufficient from necessary conditions, noting that negative
results, such as those about recurrence mentioned above do not prevent the method
to succeed. For instance, in big data practice there might be implicit use of a priori
knowledge, that needs to be made explicit.
1.11 Concluding Remarks
This paper gives an introduction to the field of nonequilbrium statistical mechanics,
a branch of physics that is in rapid and even tumultuous growth, given the variety of
possible applications, that go well beyond its original purpose: providing a microscopic foundation to equilibrium and nonequilbrium thermodynamics. Necessarily,
we have just scratched the surface, starting from it foundations and its connection
with thermodyamics. We have tried to explain why knowledge of atomic dynamics,
nuclear properties, subnuclear ingredients are not enough to produce a macroscopic
description of the world, and further ideas are needed when passing from one level
of description to another. The problem is not that we should doubt that macroscopic
objects are made of molecules, which are made of atoms, which are made of electrons and nuclei etc. The problem is what we know about these microscopic objects,
which is then expressed by our theories; indeed, such theories do not always match
in a smooth fashion [3]. The fact is that physics is a realm of measurements, and
like our senses, measurement instruments only reveal certain aspects of the object of
investigation, and is not uncommon that on different scales the resulting descriptions
may even look contradictory: “orthogonal” to each other. It is one of the purposes of
statistical physics to harmonize the different points of view, explaining what makes
them different. This should also help us understand why our present knowledge of
L. Rondoni
becuase of the large dimnsionality of the problems for which this approach is usually
invoked, i.e. those that are too complex to be tackled in a theory that starts from
an understanding of the fundamental mechanisms. The problem is enhanced in the
case of chaotic dynamics, that exponentially magnify in time the deviations from the
analog initial conditions. In principle, D equal to 6 or 7 is already out of reach in
those cases.
One could add further considerations to refute the idea that the scientific method
is made obsolete by the deluge of data. For instance, the tale attributed to Bertrand
Russell of a turkey before Thanksgiving day, shows how naive induction is doomed
to failure. The story goes like this: a turkey is fed every day. Every single feeding will
firm up the bird’s belief that it is the general rule of life to be fed every day by friendly
members of the human race “looking out for its best interests,” as a politician would
say. On the afternoon of the Wednesday before Thanksgiving, something unexpected
will happen to the turkey. It will incur a revision of belief …
Nevertheless, there are successes in using data that cannot be denied, and that can
certainly be integrated within a scientific framework, rather than opposing it; there
still is a lot to understand about what can be extracted from data, and proceeding by
examples is probably the most promising way to go. In fact, once again, it may be
beneficial to distinguish sufficient from necessary conditions, noting that negative
results, such as those about recurrence mentioned above do not prevent the method
to succeed. For instance, in big data practice there might be implicit use of a priori
knowledge, that needs to be made explicit.
1.11 Concluding Remarks
This paper gives an introduction to the field of nonequilbrium statistical mechanics,
a branch of physics that is in rapid and even tumultuous growth, given the variety of
possible applications, that go well beyond its original purpose: providing a microscopic foundation to equilibrium and nonequilbrium thermodynamics. Necessarily,
we have just scratched the surface, starting from it foundations and its connection
with thermodyamics. We have tried to explain why knowledge of atomic dynamics,
nuclear properties, subnuclear ingredients are not enough to produce a macroscopic
description of the world, and further ideas are needed when passing from one level
of description to another. The problem is not that we should doubt that macroscopic
objects are made of molecules, which are made of atoms, which are made of electrons and nuclei etc. The problem is what we know about these microscopic objects,
which is then expressed by our theories; indeed, such theories do not always match
in a smooth fashion [3]. The fact is that physics is a realm of measurements, and
like our senses, measurement instruments only reveal certain aspects of the object of
investigation, and is not uncommon that on different scales the resulting descriptions
may even look contradictory: “orthogonal” to each other. It is one of the purposes of
statistical physics to harmonize the different points of view, explaining what makes
them different. This should also help us understand why our present knowledge of