Growth transition code evolution for FUNCTION calc_growth_transition

[jimianelli]

From comment in code (moved to issues)

• • Modified Dec 11, 2014. Diagonal of the matrix now represents probability
  • of not molting, and upper triangle is the probability of growing to the next
  • size interval given you molted.
  • Dec 20. Undid the above modification after correspondence with Jack Turnock.
  • He rightly pointed out that it is possible to molt and remain in the same bin
  • interval (if the intervals are sufficiently large).
  • Jan 11, 2015. Checked cumd_gamma function in ADMB with R. This is the same
  • function as pgamma with the rate parameter set at its default value 1.0. The
  • mean value of the function is the second argument of cumd_gamma, and the vector
  • of quantiles is the first argument. Both arguments are scaled by gscale.
  • Jan 20, 2015. Jim reported that cumd_gamma was not converging. Required large
  • number of iterations to solve the cumd_gamma function in gser in the ADMB libs.
  • Soln, either increase the MAXIT in gser, or rescale the problem to the maximum
  • of the size breaks. The latter seems to work. */