|
|
ROSTUDEL demos
& samples
|
 |
| Thanks
to our new JAVA remote platform,
you will find soon here some interactive demos of ROSTUDEL
applications. Meanwhile, we provide here some code samples or
screenshots that show
the power of modeling using cutting-edge platforms we intensively
use at ROSTUDEL. |

|
OPL
and sparsity : a common pitfall in modeling
is to get stucked in tedious data structures. Imagine you have some
people in an organization that can work everyday but some prefixed
vacations and you would like to set up a constraint only on those
working days. Furthermore, let's assume you would also like to count
only those worked hours that correspond to "odd" days (Monday,
Wednesday, Friday and Sunday) and exclude the first person of your
employees from this set of person.
Well, coding corresponding data structures would be tedious in a usual
object-oriented language such as C++ or JAVA. Meanwhile, it is almost
straightforward as you could see in the code
sample. Note the beauty of collection filtering, |

|
OPTIMJ
and comprehension arrays: in this screenshot, you can see the
eclipse environment with a file with an .optimj extension. The code
illustrates the powerful comprehension notation in OPTIMJ that allows
to build arrays or maps with advanced filters. The developpers benefits
from a very high level of abstraction similar to HQL in the ORM world.
Note the bottom panel that shows the log including lp-solve logs. The
left panel shows the benefit of including directly .optimj files within
your eclipse project at no additional cost. |

|
Scheduling
in OPL6: Combinatorial problems involving tasks to
schedule under precedence constraints and resource availability are one
of the hardest problems to solve. They include for instance PERT,
Job-Shop, Open-Shop... OPL6 introduces high level constraint
programming concepts addressing this issue, including interval
variables and cumulative functions. The sample
shows a practical scheduling model minimizing the makespan for a set of
tasks linked by precedence constraints, and consuming two kind of
resources : a unary resource and a capacity one. Thanks to new default
search heuristics, OPL6 runs this model to optimality for 3000 tasks,
and achieves the same quality for a number of other benchmarks. |

|
What-if analysis with ODM:
Analysts and customers are very demanding in scenarios analysis : what
if I increase my inventory areas, what if prices rise in two months,
should I buy for stock now or just when the order falls ? Thanks to the
ODM platform, scenarios can be created on the fly from an OPL model,
duplicated for data changes, analysed on basis of several kpi's,
compared to a base scenario. On this screenshot,
we compare the inventory level of a given item subject to nominal
purchasing prices (lower curve) with an “inflation
scenario” where the prices increase from period 5. We clearly
notice the order anticipation at period 4 that reflects a
“purchase-to-stock” policy rather than buying at higher
prices. |
|