3 Wegematic 1000 Memories


Index - - Contents - - Previous chapter - Next chapter - - Previous page - Next page

Aimo Törn, cont.

After graduating in Applied Mathematics for Prof. Bertil Qvist in December 1961 I continued studies in Theoretical Physics for my Licentiate degree. Prof. Karl-Gustaf Fogel and his student Thorolf Dahlblom were engaged in energy calculations in Nuclear Physics.

The resulting mathematical problem consisted of estimating some many-dimensional integrals I(q) over complicated regions depending on the parameter q. With computers just available we were made aware of the possibility to use Monte Carlo calculations to estimate these integrals (Kurt Pörn).

In order to check the accuracy of the Monte Carlo computations two of the integrals, a 6-dimensional and a 9-dimensional were solved analytically by developing them into power series after q. These calculations involving a lot of integrations term by term and adding up for the coefficients. All this was done by hand utilizing a mechanical hand calculator for the arithmetical operations (Original Ödner). The skill in performing lengthy manual computations correctly, which was a must in those pre-computer days, is not learned anymore and I think that the students of today would feel uneasy in making such lengthy calculations tolerating no errors.

The Monte Carlo results were then compared with the analytically derived ones and it was found that with just 512 Monte Carlo points the relative error was a few %, and gave confidence in the Monte Carlo approach which was very easy to implement on Wegematic 1000. The thesis was accepted in May 1963.

During my post lic studies at Nordita in Copenhagen in the Autumn 1963 the Monte Carlo calculations were continued now on the Gier computer of the Metheorological Institute.

The writer is by October 2000 Professor in Computer Science at Åbo Akademi.