The inverse problems encountered in geosciences (seismic tomography,characterisation of reservoirs with production data, 4D seismic,stratigraphic inversion…) often require to solve a large scale nonlinearproblem subject to nonlinear constraints with cpu time consuming evaluationof the cost function. A Sequential Quadratic Programming approach in whichthe QP’s are solved by an augmented Lagrangian method has been developed andproven to be very efficient for seismic tomography.The objectives of the postdoc is to adapt and extend this algorithm forother applications in geosciences. We face different difficulties : thelarge size of the problems to be considered (large dataset, large number ofconstraints …), the non-linearity of the function to be minimized and ofthe constraints, the gradients are often not available …Different applications will be studied: the characterisation of reservoirsfrom production data at well locations, the stratigraphic inversion toestimate the location of reservoirs or to quantify petrophysic properties ofrocks.

Education and qualification:
With an interest and motivation for nonlinear optimization and convexanalysis applied to geosciences, the candidate has a PhD in appliedmathematics. Good knowledge in C and/or Fortran is required.

Location:
Institut Fran�ais du P�trole (IFP, http://www.ifp.fr) is situated at about10 kilometres from Paris and offer excellent working conditions in a livelyresearch environment.

Start: 2006
Duration: 1 year.
Salary: ~1900EUR (taxes excluded).

Interested candidates are encouraged to send an application with c.v., apublication list and referee letters to:
Dr. Delphine Sinoquet, IFP
Delphine.Sinoquet@ifp.fr