This document is the Tanner crab parallel to bbrkc_rtmb_port_steps.qmd. The current Tanner work is at the input-orientation stage: the repository now has a Tanner-specific reader for the available GMACS files in examples/tanners. This page also checks for Tanner ADMB report outputs and reports which parity targets are available for future RTMB objective and deterministic checks.
The Tanner reader uses:
| File | Role |
|---|---|
examples/tanners/gmacs_26_22_03d5_aEffExp1.Run1.dat |
GMACS main run file. |
examples/tanners/TannerCrab_Data202509.EffExp1.dat |
Tanner data file. |
examples/tanners/TannerCrab_26_22_03d5_a.Run1.ctl |
Tanner control file. |
examples/tanners/TannerCrab_26_22_03d5_a.prj |
Tanner projection file. |
examples/tanners/gmacs_26_22_03d5_aEffExp1.par |
Fitted parameter table. |
examples/tanners/Gmacsall.out or Tanner equivalent |
Optional ADMB likelihood and fit-summary output. |
examples/tanners/gmacs.rep or Tanner equivalent |
Optional ADMB report output for deterministic blocks. |
Optional ADMB parity targets are discovered if they are added to the same folder. Canonical names are Gmacsall.out and gmacs.rep; the reader also checks common Tanner-specific variants such as gmacs_26_22_03d5_aEffExp1.Gmacsall.out and gmacs_26_22_03d5_aEffExp1.rep.
The Tanner files use a different GMACS text layout from the BBRKC example. R/read_gmacs_tanner.R therefore reads the Tanner file set directly instead of forcing these inputs through BBRKC-specific parser assumptions.
source("R/read_gmacs_bbrkc.R")
source("R/read_gmacs_tanner.R")
tanner <- read_tanner_inputs("examples/tanners")
validate_tanner_inputs(tanner)
has_tanner_index_fit <- !is.null(tanner$admb_reference$index_fit)
has_tanner_size_fit <- !is.null(tanner$admb_reference$size_fit)data.frame(
report = names(tanner$admb_reference$files),
path = unlist(tanner$admb_reference$files),
available = file.exists(unlist(tanner$admb_reference$files)),
row.names = NULL
) report path available
1 all <NA> FALSE
2 report <NA> FALSECandidate ADMB report names checked by the reader:
candidates <- tanner_admb_report_candidates("examples/tanners")
data.frame(
report_type = rep(names(candidates), lengths(candidates)),
candidate = basename(unlist(candidates)),
present = file.exists(unlist(candidates)),
row.names = NULL
) report_type candidate present
1 all Gmacsall.out FALSE
2 all gmacsall.out FALSE
3 all GmacsAll.out FALSE
4 all TannerCrab_Gmacsall.out FALSE
5 all gmacs_26_22_03d5_aEffExp1.Gmacsall.out FALSE
6 all gmacs_26_22_03d5_aEffExp1.all.out FALSE
7 report gmacs.rep FALSE
8 report Gmacs.rep FALSE
9 report TannerCrab.rep FALSE
10 report TannerCrab_26_22_03d5_a.rep FALSE
11 report gmacs_26_22_03d5_aEffExp1.rep FALSEdims <- tanner$data$dimensions
data.frame(
quantity = names(dims),
value = as.integer(dims),
row.names = NULL
) quantity value
1 syr 1948
2 nyr 2024
3 nseason 6
4 nfleet 6
5 nsex 2
6 nshell 2
7 nmature 2
8 nclass 32data.frame(
component = c("catch", "survey", "size composition", "growth", "maturity"),
dataframes = c(
tanner$data$catch$n,
tanner$data$survey$n,
tanner$data$size_comp$n,
NA_integer_,
NA_integer_
),
rows = c(
sum(tanner$data$catch$rows),
sum(tanner$data$survey$rows),
sum(tanner$data$size_comp$rows),
tanner$data$growth$n,
tanner$data$maturity$n
)
) component dataframes rows
1 catch 6 282
2 survey 6 168
3 size composition 15 471
4 growth NA 214
5 maturity NA 136data.frame(
parameter_rows = nrow(tanner$parameters),
active_estimated_parameters = sum(tanner$parameters$active),
min_estimated_index = min(tanner$parameters$estimated_index, na.rm = TRUE),
max_estimated_index = max(tanner$parameters$estimated_index, na.rm = TRUE)
) parameter_rows active_estimated_parameters min_estimated_index
1 469 173 1
max_estimated_index
1 157head(tanner$parameters[tanner$parameters$active, c(
"parameter_index", "estimated_index", "value", "label"
)], 12) parameter_index estimated_index value
3 3 1 6.3256587
11 11 2 5.4245970
13 13 NA 32.3984669
15 15 NA 163.9410634
18 18 NA 33.3714519
20 20 NA 115.3816014
22 22 3 0.3174953
23 23 4 0.7395193
24 24 5 0.2510995
25 25 6 0.3191779
26 26 7 0.6503671
27 4 4 4.8327442
label
3 Log(Rbar) 1
11 Log(Rbar)_block_group_10_block_1 2
13 Alpha_male
15 Beta_male
18 Alpha_female
20 Beta_female
22 M_base_male_mature 3
23 M_male_mature_block_group_1_block_1 4
24 M_base_male_immature 5
25 M_base_female_mature 6
26 M_female_mature_block_group_1_block_1 7
27 The Tanner reader validates:
.par file.The same style of ADMB target extraction used for BBRKC is wired for Tanner when report outputs are present:
read_admb_likelihood_reference() for nloglike, nlogPenalty, and priorDensity;read_admb_catch_fit_summary(), read_admb_index_fit_summary(), and read_admb_size_fit_summary() for observed/predicted fit summaries;read_admb_selectivity_block() for slx_capture, slx_retaind, and slx_discard from the report file.admb_ref <- tanner$admb_reference
data.frame(
target = c(
"nloglike",
"nlogPenalty",
"priorDensity",
"catch fit summary",
"index fit summary",
"size fit summary",
"slx_capture",
"slx_retaind",
"slx_discard"
),
source_file = c(
rep(admb_ref$files$all, 6),
rep(admb_ref$files$report, 3)
),
parsed_rows = c(
if (!is.null(admb_ref$likelihood)) length(admb_ref$likelihood$nloglike) else 0L,
if (!is.null(admb_ref$likelihood)) length(admb_ref$likelihood$nlog_penalty) else 0L,
if (!is.null(admb_ref$likelihood)) length(admb_ref$likelihood$prior_density) else 0L,
if (!is.null(admb_ref$catch_fit)) nrow(admb_ref$catch_fit) else 0L,
if (!is.null(admb_ref$index_fit)) nrow(admb_ref$index_fit) else 0L,
if (!is.null(admb_ref$size_fit)) nrow(admb_ref$size_fit) else 0L,
if (!is.null(admb_ref$selectivity)) nrow(admb_ref$selectivity$capture) else 0L,
if (!is.null(admb_ref$selectivity)) nrow(admb_ref$selectivity$retained) else 0L,
if (!is.null(admb_ref$selectivity)) nrow(admb_ref$selectivity$discard) else 0L
),
available = c(
rep(!is.null(admb_ref$likelihood), 3),
!is.null(admb_ref$catch_fit),
!is.null(admb_ref$index_fit),
!is.null(admb_ref$size_fit),
rep(!is.null(admb_ref$selectivity), 3)
),
row.names = NULL
) target source_file parsed_rows available
1 nloglike <NA> 0 FALSE
2 nlogPenalty <NA> 0 FALSE
3 priorDensity <NA> 0 FALSE
4 catch fit summary <NA> 0 FALSE
5 index fit summary <NA> 0 FALSE
6 size fit summary <NA> 0 FALSE
7 slx_capture <NA> 0 FALSE
8 slx_retaind <NA> 0 FALSE
9 slx_discard <NA> 0 FALSEIf the parsed row counts are zero, add Gmacsall.out and gmacs.rep, or one of the Tanner-equivalent names listed above, to examples/tanners. Once those files are present, this page will expose the ADMB likelihood, fit, and selectivity targets needed to build Tanner parity checks.
if (!has_tanner_index_fit || !has_tanner_size_fit) {
data.frame(
diagnostic = c("index observed-vs-predicted", "size observed-vs-predicted"),
status = c(
if (has_tanner_index_fit) "available" else "missing ADMB Index_fit_summary",
if (has_tanner_size_fit) "available" else "missing ADMB Size_fit_summary"
),
required_file = "Tanner Gmacsall.out or equivalent",
row.names = NULL
)
} diagnostic status
1 index observed-vs-predicted missing ADMB Index_fit_summary
2 size observed-vs-predicted missing ADMB Size_fit_summary
required_file
1 Tanner Gmacsall.out or equivalent
2 Tanner Gmacsall.out or equivalentWhen Tanner ADMB fit summaries are present, the figure below shows observed and predicted index fits from Gmacsall.out. In the current repository, Tanner ADMB report output is not present, so this figure is observed-only.
if (!is.null(tanner$admb_reference$index_fit)) {
index_fit <- tanner$admb_reference$index_fit
series_ids <- sort(unique(index_fit$series))
panel_rows <- ceiling(length(series_ids) / 2)
panel_slots <- panel_rows * 2
old_par <- par(mar = c(3, 3, 2, 1), oma = c(0, 0, 0, 0))
layout(
rbind(matrix(seq_len(panel_slots), ncol = 2, byrow = TRUE), rep(panel_slots + 1, 2)),
heights = c(rep(1, panel_rows), 0.16)
)
for (series in series_ids) {
x <- index_fit[index_fit$series == series, ]
ylim <- range(c(x$obs, x$predicted), finite = TRUE)
plot(x$year, x$obs, pch = 16, col = "#1b6da8", ylim = ylim,
xlab = "", ylab = "", main = paste("Series", series), bty = "l")
line_keys <- unique(x[c("fleet", "season", "sex", "maturity")])
for (line_i in seq_len(nrow(line_keys))) {
keep <- rep(TRUE, nrow(x))
for (key_name in names(line_keys)) {
keep <- keep & x[[key_name]] == line_keys[[key_name]][line_i]
}
line_x <- x[keep, ]
line_x <- line_x[order(line_x$year), ]
lines(line_x$year, line_x$predicted, lwd = 2, col = "#c43c39")
points(line_x$year, line_x$predicted, pch = 1, col = "#c43c39")
}
grid(col = "grey88")
}
if (length(series_ids) < panel_slots) {
for (unused in seq_len(panel_slots - length(series_ids))) {
plot.new()
}
}
par(mar = c(0, 0, 0, 0))
plot.new()
legend("center", horiz = TRUE, bty = "n",
pch = c(16, NA), lty = c(NA, 1), lwd = c(NA, 2),
col = c("#1b6da8", "#c43c39"), legend = c("Observed", "Predicted"))
layout(1)
par(old_par)
} else {
index_obs <- tanner$data$survey_observations
series_ids <- sort(unique(index_obs$series))
old_par <- par(mfrow = c(ceiling(length(series_ids) / 2), 2), mar = c(3, 3, 2, 1), oma = c(0, 0, 2, 0))
for (series in series_ids) {
x <- index_obs[index_obs$series == series, ]
plot(x$year, x$obs, type = "b", pch = 16, col = "#1b6da8",
xlab = "", ylab = "", main = paste(x$fleet[1], x$sex[1], x$maturity[1]),
bty = "l")
grid(col = "grey88")
}
mtext("Observed only: Tanner ADMB fit summaries are not present", outer = TRUE, line = 0.4)
par(old_par)
}
When Tanner ADMB size-fit summaries are present, the panels compare observed and predicted compositions from Gmacsall.out. Otherwise they show observed input compositions for recent rows from the first four size-composition dataframes. In the current repository, Tanner ADMB report output is not present, so this figure is observed-only.
if (!is.null(tanner$admb_reference$size_fit)) {
size_fit <- tanner$admb_reference$size_fit
size_keys <- c("modified_series", "year", "fleet", "season", "sex", "type", "shell", "maturity")
size_key_rows <- unique(size_fit[size_fit$vector == "obs", size_keys])
size_key_rows <- size_key_rows[order(size_key_rows$modified_series, size_key_rows$year), ]
series_ids <- sort(unique(size_key_rows$modified_series))[1:4]
old_par <- par(mar = c(3, 3, 2, 1), oma = c(0, 0, 0, 0))
layout(rbind(matrix(1:4, ncol = 2, byrow = TRUE), c(5, 5)), heights = c(1, 1, 0.16))
for (series in series_ids) {
candidates <- size_key_rows[size_key_rows$modified_series == series, ]
key <- candidates[nrow(candidates), ]
key_match <- rep(TRUE, nrow(size_fit))
for (key_name in size_keys) {
key_match <- key_match & size_fit[[key_name]] == key[[key_name]]
}
obs <- size_fit[key_match & size_fit$vector == "obs", c("size_bin", "value")]
pred <- size_fit[key_match & size_fit$vector == "pred", c("size_bin", "value")]
x <- merge(obs, pred, by = "size_bin", suffixes = c("_observed", "_predicted"))
x <- x[order(x$size_bin), ]
observed <- x$value_observed / sum(x$value_observed)
predicted <- x$value_predicted / sum(x$value_predicted)
ylim <- range(c(observed, predicted), finite = TRUE)
mids <- barplot(observed, col = "#1b6da8", border = NA,
ylim = ylim, xlab = "", ylab = "",
main = paste("Series", series, key$year), bty = "l")
axis(1, at = mids, labels = x$size_bin)
lines(mids, predicted, lwd = 2, col = "#c43c39")
grid(col = "grey88")
}
if (length(series_ids) < 4) {
for (unused in seq_len(4 - length(series_ids))) {
plot.new()
}
}
par(mar = c(0, 0, 0, 0))
plot.new()
legend("center", horiz = TRUE, bty = "n",
fill = c("#1b6da8", NA), border = c(NA, NA),
lty = c(NA, 1), lwd = c(NA, 2), col = c(NA, "#c43c39"),
legend = c("Observed", "Predicted"))
layout(1)
par(old_par)
} else {
size_obs <- tanner$data$size_comp_observations
series_ids <- sort(unique(size_obs$series))[1:4]
old_par <- par(mfrow = c(2, 2), mar = c(3, 3, 2, 1), oma = c(0, 0, 2, 0))
for (series in series_ids) {
x <- size_obs[size_obs$series == series, ]
x <- x[order(x$year), ]
row <- x[nrow(x), ]
bins <- grep("^bin_", names(row), value = TRUE)
values <- as.numeric(row[1, bins])
values <- values / sum(values)
barplot(values, col = "#1b6da8", border = NA,
xlab = "", ylab = "", main = paste(row$fleet, row$sex, row$catch_type, row$year),
bty = "l")
grid(col = "grey88")
}
mtext("Observed only: Tanner ADMB size-fit summary is not present", outer = TRUE, line = 0.4)
par(old_par)
}