Wednesday, May 16, 2007

Computer simulations and testing: A small clarification


The issues of using computers to test nuclear weapons designs has been brought up in some contexts. Some of this needs to be clarified.

Computers can be used to carry out some simulations which can aid in weapons design, but there are aspects of weapons design that cannot be simulated on a computer, and these simply have to be tested.

Computers can be used to solve certain equations. As long as the behaviour of the weapon can be described by those equations, the computer can in theory be used to address that design issue, if computational resources are available.

There are other aspects of a weapon (for example age related deterioration of warheads, or the sort of stuff that makes the American W76 warhead seem like a near total waste of money) which cannot be described by a "nice" equation and these problems can then only be crudely approximated on an exceptionally powerful computer which India has not yet built. ... a powerful computer that can also be used to solive "nice" equations very very fast.

Like the nuclear reactor itself, a very high performance computer is itself a "dual use" technology and the development of such systems eg. computer arrays for parallel processing usually makes the NPA's see red.

Computations can be used to optimise the yeild of a weapon. However in order to make this claim with credibility, a weapon has to be optimised with the computer aided design process and then field tested.

As I had said earlier, India can have an arsenal of a few weapons if it is allowed to optimise and test new designs with better yeild or it can have larger arsenal with smaller yeild but then it will have to vastly expand its high performance computing ability and it will still have to carry out periodic field tests to make sure that the computations are giving the right answers.

The Americans will also need to carry out such tests to ensure that their stockpile preservation programs and simulations are yeilding the right answers and that is why they have not signed any test-ban treaties. The prospect of a W76 going off when it is not supposed to has kept many of Americans up at night and a lot of money has been spent on understanding what happens when a weapon ages. This kind of activity has remained a largely hidden cost in the maintenance of such a large arsenal.

If India carries out more tests, the CTBT will suffer extremely unpleasant shocks. If India goes in for a larger arsenal, the FMCT will lose relevance *and* India will build a very large (essentially dual use) high perfomance computation capability essentially hollowing out the entire framework of computation related technology control regimes.

Either way we play this game, the NPA world view is going to have to be adjusted.

People have to make that choice what they prefer, an India with a larger arsenal and vastly expanded high performance computational capability or an India with a smaller arsenal but testing higher yeild weapons.

No one in India can make that choice for them.


Post a Comment

<< Home