Greater precision with optim.interv

[jhhughes256]

Goal

To improve the precision in which optim.interv determines its parameters.

Description

As seen in #7 there is a lack of precision of the algorithm. This could be due to poorly chosen intervals by optim.interv.

Ideas

Greater precision might be achieved with following:

• [x] Using the tmax to influence initial parameter generation
  • Investigated in #9
  • Problematic results, some intervals skipping the tmax entirely
  • Not included in algorithm
• [x] Use a repeat loop to test different initial parameters
  • Investigated in #9
  • Included in algorithm
• [x] Ordering final parameters received from optim.interv
  • Investigated in issue #16
  • Included in algorithm
• [x] Using the ga() function to generate initial parameters
  • Open issue #18
  • Has replaced the use of repeat loop
• [x] Incorporation of lin-log trapezoidal rule
  • Open issue #4
• [x] Use hessian matrix to determine standard error on parameter estimates
  • Investigated in #17
  • See .rmd provided by Richard Upton for explanation of what needs to be done
  • This could help in determining useful error messages
  • May be able to address some of the setbacks found in #18