76
J. A. BURNS
in addition to our more specific interest in interpreting the arginine pathway
in Neurospora. A final requirement of "modelling" is that by its nature it
demands a turn round of at least twice per day, which in turn requiers, on a
shared machine, that the answers to a significant group of questions can be
computed in a time of order 15 min.
A general method for modelling must therefore make it easy for the
investigator to move from his "annotated diagram" to the computer and
also facilitate his consideration of the effects in the model of changes in
structure or parametric values. The method must also allow the investigator
to ask "pre-programmed" questions about the system in a flexible way, i.e.
in any order, and to carry out calculations, as required, on variables and
parameters in the model using their normal names. The ability to extend
the list of pre-programmed questions and operations as new ideas occur to
an investigator is most important and the overall programming technique
should make it easy to do this.
Programming Methods
The machine, an E.E.L.M. KDF9 had Core: 16K of 48 bit words Cycle
time: 8 micro-sees; Floating point multiply: 15 micro-sees. The KDF9 is
available to all university users and provides a turn round of the order
twice per 24 hrs for programs running less than 15 min. For longer runs
turn round is unpredictable.
The single pass "load and go" "Atlas Autocode" compiler commonly
used in Edinburgh [1] compiles very rapidly from an "Algol like" source
language taking of the order 4 min for the largest programs possible
with the 16K of core. In view of the availability of a fast Algol type compiler, the requirements mentioned in the introduction can be met most
simply using a program structure as follows:
Start of Program
A. Algebraic description of system under investigation, preceded by
list of all names used.
B. Routines which enable analysis to be carried out on system in A and
which facilitate setting variables and parameters; (for examples see later).
C. List of commands for the analysis of A subject to what is "preprogrammed" in B, simple calculations etc.
End of Program
The above layout is a program which does not require any external data
but has merely to be compiled and executed. Clearly sections A and C can
be very simple, amounting, respectively, to an algebraic description of the
J. A. BURNS
in addition to our more specific interest in interpreting the arginine pathway
in Neurospora. A final requirement of "modelling" is that by its nature it
demands a turn round of at least twice per day, which in turn requiers, on a
shared machine, that the answers to a significant group of questions can be
computed in a time of order 15 min.
A general method for modelling must therefore make it easy for the
investigator to move from his "annotated diagram" to the computer and
also facilitate his consideration of the effects in the model of changes in
structure or parametric values. The method must also allow the investigator
to ask "pre-programmed" questions about the system in a flexible way, i.e.
in any order, and to carry out calculations, as required, on variables and
parameters in the model using their normal names. The ability to extend
the list of pre-programmed questions and operations as new ideas occur to
an investigator is most important and the overall programming technique
should make it easy to do this.
Programming Methods
The machine, an E.E.L.M. KDF9 had Core: 16K of 48 bit words Cycle
time: 8 micro-sees; Floating point multiply: 15 micro-sees. The KDF9 is
available to all university users and provides a turn round of the order
twice per 24 hrs for programs running less than 15 min. For longer runs
turn round is unpredictable.
The single pass "load and go" "Atlas Autocode" compiler commonly
used in Edinburgh [1] compiles very rapidly from an "Algol like" source
language taking of the order 4 min for the largest programs possible
with the 16K of core. In view of the availability of a fast Algol type compiler, the requirements mentioned in the introduction can be met most
simply using a program structure as follows:
Start of Program
A. Algebraic description of system under investigation, preceded by
list of all names used.
B. Routines which enable analysis to be carried out on system in A and
which facilitate setting variables and parameters; (for examples see later).
C. List of commands for the analysis of A subject to what is "preprogrammed" in B, simple calculations etc.
End of Program
The above layout is a program which does not require any external data
but has merely to be compiled and executed. Clearly sections A and C can
be very simple, amounting, respectively, to an algebraic description of the