R/rescomp.R

In glotaran/TIMP: Fitting Separable Nonlinear Models in Spectroscopy and Microscopy

"rescomp" <-
  function(theta = vector(), d = vector(), currModel = currModel, currTheta = vector()) {
    # browser()
    if (currModel@modelspec[[1]]@lscalpar) {
      thetascal <- currModel@modellist[[1]]@thetascal
      theta <- theta * thetascal
    }
    if (length(currTheta) == 0) {
      currTheta <- getThetaCl(theta, currModel)
    }
    groups <- currModel@groups
    m <- currModel@modellist
    resid <- clpindepX <- list()
    nexp <- length(m)
    for (i in 1:nexp) {
      clpindepX[[i]] <- if (!m[[i]]@clpdep || m[[i]]@getX) {
        getClpindepX(
          model = m[[i]], theta =
            currTheta[[i]], multimodel = currModel,
          returnX = FALSE, rawtheta = theta, dind = 0
        )
      } else {
        matrix()
      }
    }
    for (i in 1:length(groups)) {
      resid[[i]] <- residPart(
        model = m[[1]],
        group = groups[[i]], multimodel = currModel,
        thetalist = currTheta, clpindepX = clpindepX,
        finished = currModel@finished,
        returnX = FALSE, rawtheta = theta
      )
      if (currModel@finished) {
        currModel <- fillResult(
          group = groups[[i]],
          multimodel = currModel, thetalist = currTheta,
          clpindepX = clpindepX, rlist = resid[[i]],
          rawtheta = theta
        )
      }
    }
    if (currModel@finished) {
      currModel@fit@nlsres$onls$nclp <- currModel@nclp
      if (currModel@optlist[[1]]@sumnls) {
        if (is(currModel@fit@nlsres$onls, "nls")) {
          class(currModel@fit@nlsres$onls) <- "timp.nls"
        } else if (is(currModel@fit@nlsres$onls, "nls.lm")) {
          class(currModel@fit@nlsres$onls) <- "timp.nls.lm"
        } else {
          class(currModel@fit@nlsres$onls) <- "timp.optim"
        }
        currModel@fit@nlsres$sumonls <- summary(currModel@fit@nlsres$onls,
          currModel = currModel,
          currTheta = currTheta
        )
      }
      if (currModel@stderrclp) {
        for (i in 1:length(groups)) {
          currModel <- getStdErrClp(
            group = groups[[i]],
            multimodel = currModel, thetalist = currTheta,
            clpindepX = clpindepX, rlist = resid[[i]],
            rawtheta = theta
          )
        }
      }
    }
    ## if using a trilinear type model, we have cp=AE; so separate A out.
    if (currModel@finished && currModel@trilinear) {
      trires <- triResolve(currModel, currTheta)
      currModel <- trires$currModel
      currTheta <- trires$currTheta
    }
    if (currModel@finished && m[[1]]@mod_type == "kin") {
      if (m[[1]]@fullk) {
        for (i in 1:nexp) {
          nocolsums <- length(m[[1]]@lightregimespec) > 0 # lightdiff (see compModel.R)
          eig <- fullKF(
            currTheta[[i]]@kinpar,
            currTheta[[i]]@kinscal,
            m[[1]]@kmat,
            currTheta[[i]]@jvec,
            m[[1]]@fixedkmat,
            m[[1]]@kinscalspecial,
            m[[1]]@kinscalspecialspec,
            nocolsums
          )
          currTheta[[i]]@eigenvaluesK <- eig$values
        }
      }
    }
    if (currModel@finished) {
      return(list(currModel = currModel, currTheta = currTheta))
    }
    if (currModel@algorithm == "optim") { ## minimize this sum
      retval <- sum(unlist(resid))
    } else {
      retval <- unlist(resid)
    } ## nls and nls.lm want the residuals,
    ## to minimize the sum of their squares
    retval
  }