Numerical discretisation of stochastic (partial) differential equations

Dnr:

SNIC 2017/1-135

Type:

SNAC Medium

Principal Investigator:

David Cohen

Affiliation:

Umeå universitet

Start Date:

2017-03-30

End Date:

2018-04-01

Primary Classification:

10105: Beräkningsmatematik

I am interested in the numerical discretisation of
stochastic (partial) differential equations (SPDEs).
To test the convergence, or other properties,
of the numerical solutions, one often has
to compute the expectations of some
quantities. One thus need to simulate
many many times the solution of such
a stochastic problem. In the SPDE
context, one thus has to solve M times
a partial differential equation, where M should be
of the order of 10^3 or more, in order
to have M samples of the numerical solutions.
Once this is done, one can use this information
to approximate the above expectations and
thus confirm the good behaviour of the numerical
methods.
Since I realise that I need more CPU time than expected, I kindly ask if I you could upgrade my SMALL SNIC project to a MEDIUM one with 40000 CPU hours on Abisko or Kebnekaise.
Thank you for your understanding.