Natural selection can work on differences to select optima {evolution theories}.
In fitness landscapes, starting from any genotype, change one allele randomly to go to neighboring genotype, and then stay there if it has higher fitness {adaptive walk}|. Adaptive walks go uphill to local peaks.
steps
After uphill steps, number of possibly higher steps becomes half or less and finding a higher step takes twice as long or longer.
If starting at low fitness, large steps work best. If starting at high fitness, small steps work best.
recombination
Recombination allows leaving local maximum to look for global maximum, if landscape is not too random.
few alleles
With two alleles for genes, number of steps to local peak is natural logarithm of gene number. Many local peaks are nearby. Because changes are random and neighboring fitnesses do not correlate, selection makes little improvement.
multiple gene changes
If several alleles change, steps go farther, and fitness-change range is greater.
gene number
If gene number increases, fitness increase per step is less.
Gene fitness can depend on other-gene fitness in same or other species {coupled fitness landscape} {fitness landscape, coupled}.
If species dependency is low, same-species gene dependency is high, or species number is low, system is stable. If species dependency is high, same-species gene dependency is low, or species number is high, system is unstable.
If species number or species-dependency level can change, species dependency tends to middle, to maximize average fitness and minimize average extinction rate. Middle level is just before point where too many changes happen.
Genes have alleles, which contribute to survival and reproduction {fitness, allele}.
fitness landscape
Graphs {fitness landscape, evolution} {genetic graph} show all genes and alleles in a genotype. Genotypes have height. Genotypes differ from neighboring genotypes by one allele.
correlations
Fitness landscapes can have neighbors with similar fitness {correlated fitness}, so allele fitness depends on other-gene fitness. If allele fitness greatly depends on other genes, allele change affects whole system, fitnesses are similar, and landscape has many local peaks. If genes affect all others, fitnesses are equal, as in random case. If allele fitness depends slightly on other genes, allele change affects small region, fitnesses are different, and landscape has few local peaks. Highest peaks have widest bases.
patch
Genetic graphs can divide into local maximum-fitness regions. Patch boundaries are at local fitness minima.
search
In fitness landscapes, search for maximum can stop at local maximum. To escape local maximum or minimum, methods can look at patches or ignore input from neighbors. Instead of moving from fitness-landscape point to point, trajectories can move from patch to patch, seeking highest maximum. If patch size is small, system takes many steps to reach maximum fitness, can move away from true maximum, and can change over time, so it never reaches maximum fitness. If patch size is large, patches can have several maximum-fitness peaks. Moderate patch size balances fitness and time. If fitness landscape is smooth, patches have no affect.
Systems can ignore input from neighbors and use input from farther away. If jumps are small, system takes many steps to reach maximum fitness, can move away from true maximum, and can change over time, so it never reaches maximum fitness. If jumps are large, system can jump several maximum-fitness peaks. Moderate jump size balances fitness and time.
Systems {genetic algorithm} can perform similar processes varying by one or more parameters, in response to environment situations that require learning. Some processes are more correct or successful and have better fitness. More-successful processes can continue by selection and/or initiate similar processes by reproduction. Systems thus move toward higher percentage of more-successful processes. Systems have evolution.
In Boolean systems {genetic circuit, mathematics} {genetic network model}, repressors can turn off gene expression. Enhancers can turn on gene expression. Genes are on or off. Repressors and enhancers are either present or absent. Boolean canalyzing functions model gene regulation. Without repressor, gene expresses. With derepressor, gene expresses. Different repressor or derepressor amounts can modulate gene expression. Promoter, derepressor, or other gene sites can regulate gene expression.
Outline of Knowledge Database Home Page
Description of Outline of Knowledge Database
Date Modified: 2022.0225