Imagine a pride of lions all sitting under a large tree. The Leader usually takes the best spot with the rest fitting in according to their hierarchy in the pride.
TimeDesign tries to simulate this environment with the groups each given an initial level of strength based on the number of classes, educators, and rooms in the group. This initial strength level, then allows the program to do a first pass in slotting in the groups. On the next run, the groups that have not been slotted in get an increase in strength, then attempt to force out any groups that are now weaker. To ensure that a really strong group does not get stuck indefinitely in a period slot that might cause the table to be unsolvable, the program lets these groups die from old age, which basically means the group has to start again fighting it's way to the top.
The ideal solving environment is when you are able to keep the current line on the graph moving in a saw tooth fashion. Before each loop, the groups are resorted based on their strength, with the strongest getting first options.
The Parameters.
No Change Max.
If after x number of major loops with no improvement, all group strengths are set to their base level.
Partial Death.
The weaker half of the pride is removed. Strengths are kept the same.
Full Death.
The whole pride is removed. Strengths are kept the same.
Mini Loops.
How many loops take place before a major loop takes place.
Reset to best.
TimeDesign tries to allow potential new solutions to develop, if after x number of loops with no improvement, the best solution is used to continue with. Strengths are not reset.
Initial Fights.
How many times per mini loop the groups will strengthen and attempt to oust another group.
Old Age.
The age a group must reach before it dies a 'natural death' and restarts.
Random Placement.
If at least x blank periods are available on a day, the group to be randomly placed on one of them otherwise TimeDesign takes the first available blank period.
Positional Win.
Any group is only allowed to fight with another group if it is within x number of steps within the hierarchy i.e. the low ranking groups have to fight their way up the ladder.
Positional Fights.
This is similar to Initial Fights, the tenacity of the fighting group.
Neighbour Win and Neighbour Fights.
These are the same as the two above except that they come into action if the group already has at least one period slotted in a day, and is now attempting to get the prior or following period as well.
Neighbour Kills.
Sometimes, where the group already has on period slotted, it is unable to get the second period to make a double period. This allows the slotted period to get killed, thus allowing it to find a new spot where possibly it is able to get the double easier.
Level One Cut off.
This is used with the 'slow but steady' option to decide when the multiple periods and min/day periods have been placed then the rest are allowed to be placed.
Strength Consideration.
The checked options are taken into account when the group needs to be strengthened, if none are ticked, a basic strength increment is added to the group.
Bias Settings.
This is expressed as a percentage here. It indicates how strongly the program must take into account various scenarios.
You have the option to load and save different settings according to the style of timetable that you're working with. If you load new settings, it will only come into effect when you apply them. this allows you to preview different settings first. To cancel these loaded settings, switch to one of the other screens (not one within the settings tab) then switch back again.
Feel free to change these values as the values given are the result of months of testing and timing with a spread of different styles of timetables, it is possible to find values that are even better for your style of timetable.
You have the ability to reset the values to the defaults, so there is no chance of messing things up beyond repair.