2048: Difference between revisions
Content added Content deleted
m (→Output: match revised source message.) |
m (→The Plan) |
||
| Line 2,169: | Line 2,169: | ||
Except that WAY and YAW are ''two'' dimensional arrays (rather than a one-dimensional array of complex number pairs, alas) so that the expression is in fact <code>RC*WAY(W,1:2)</code> and the calculations for ''both'' indices are done together. Because Fortran uses the "column-major" ordering of elements in storage, successive elements of a multi-dimensional array have the leftmost index varying most rapidly so that the order is WAY(1,1), WAY(2,1), WAY(3,1), WAY(4,1), WAY(1,2), ''etc'' and statements such as DATA or PARAMETER whereby values can be listed employ that ordering. So that the list of values for WAY and YAW can be aligned in the source code with the similar lists for the arrays specifying the loop parameters for each direction, the ordering is WAY(4,2) rather than WAY(2,4) even though this means that the two values for a given direction are not in adjacent storage, unlike the two parts of a complex number. |
Except that WAY and YAW are ''two'' dimensional arrays (rather than a one-dimensional array of complex number pairs, alas) so that the expression is in fact <code>RC*WAY(W,1:2)</code> and the calculations for ''both'' indices are done together. Because Fortran uses the "column-major" ordering of elements in storage, successive elements of a multi-dimensional array have the leftmost index varying most rapidly so that the order is WAY(1,1), WAY(2,1), WAY(3,1), WAY(4,1), WAY(1,2), ''etc'' and statements such as DATA or PARAMETER whereby values can be listed employ that ordering. So that the list of values for WAY and YAW can be aligned in the source code with the similar lists for the arrays specifying the loop parameters for each direction, the ordering is WAY(4,2) rather than WAY(2,4) even though this means that the two values for a given direction are not in adjacent storage, unlike the two parts of a complex number. |
||
Arrays WAY and YAW are reminiscent of "truth tables" in Boolean logic, and it is tempting to imagine that YAW = ¬WAY, but alas, a NOT operation applied to an integer variable will flip not just the lowest bit. Trying a .NOT. operation on LOGICAL variables instead will work as desired, except that their integer interpretations may not be as hoped for. |
Arrays WAY and YAW are reminiscent of "truth tables" in Boolean logic, and it is tempting to imagine that YAW = ¬WAY, but alas, a NOT operation applied to an integer variable will flip not just the lowest bit. Trying a .NOT. operation on LOGICAL variables instead will work as desired, except that their integer interpretations may not be as hoped for. Yet another ploy might be based on W being even/odd or odd/even, and similar trickery might be applied to the other arrays of constants, but, enough. The devious juggling of arrays is traditional in Fortran. |
||
===Source=== |
===Source=== |
||