# Initialise -------------------------------------------------------------- # North sea params <- newMultispeciesParams(NS_species_params_gears, inter, n = 2/3, p = 0.7, lambda = 2.8 - 2/3) no_gear <- dim(params@catchability)[1] no_sp <- dim(params@catchability)[2] no_w <- length(params@w) no_w_full <- length(params@w_full) sim <- project(params, effort = 1, t_max = 20, dt = 0.5, t_save = 0.5) # Rescaled params_r <- params volume <- 1e-13 params_r@initial_n <- params@initial_n * volume params_r@initial_n_pp <- params@initial_n_pp * volume for (res in names(params@initial_n_other)) { params_r@initial_n_other[[res]] <- params@initial_n_other[[res]] * volume } params_r@species_params$gamma <- params@species_params$gamma / volume params_r <- setSearchVolume(params_r) params_r@species_params$R_max <- params_r@species_params$R_max * volume # Random abundances set.seed(0) n <- abs(array(rnorm(no_w * no_sp), dim = c(no_sp, no_w))) * 1e9 n_full <- abs(rnorm(no_w_full)) * 1e9 params2 <- params params2@initial_n <- params2@initial_n / 2 params2@initial_n_pp <- params2@initial_n_pp / 2 params2@initial_n_other <- list(test = 1) params2@initial_effort <- params2@initial_effort / 2 # getRates ---- test_that("getRates works", { r <- getRates(params) expect_identical(names(r), c("encounter", "feeding_level", "e", "e_repro", "e_growth", "pred_rate", "pred_mort", "f_mort", "mort", "rdi", "rdd", "resource_mort")) # test that the optional parameters take the correct defaults expect_identical(r, getRates(params, n = params@initial_n, n_pp = params@initial_n_pp, n_other = params@initial_n_other, effort = params@initial_effort, t = 0)) # test that getRates actually uses its optional arguments expect_identical(getRates(params2), getRates(params, n = params2@initial_n, n_pp = params2@initial_n_pp, n_other = params2@initial_n_other, effort = params2@initial_effort, t = 0)) }) # getEncounter -------------------------------------------------------------- test_that("getEncounter returns with correct dimnames", { enc <- getEncounter(params) expect_identical(dimnames(enc), dimnames(params@initial_n)) }) test_that("mizerEncounter is independent of volume", { enc <- getEncounter(params) enc_r <- getEncounter(params_r) expect_equal(enc, enc_r) }) test_that("External encounter is included", { enc <- getEncounter(params) # add something of the right dimension extra_enc <- params@mu_b ext_encounter(params) <- ext_encounter(params) + extra_enc expect_identical(getEncounter(params), enc + extra_enc) }) # getFeedingLevel ----------------------------------------- test_that("getFeedingLevel for MizerParams", { fl <- getFeedingLevel(params, n, n_full) # test dim expect_identical(dim(fl), c(no_sp, no_w)) expect_identical(dimnames(fl), dimnames(params@initial_n)) # A crap test - just returns what's already in the function encounter <- getEncounter(params, n = n, n_pp = n_full) f <- encounter / (encounter + params@intake_max) expect_identical(fl, f) # test value expect_known_value(fl, "values/getFeedingLevel") }) test_that("getFeedingLevel for MizerSim", { time_range <- 15:20 fl <- getFeedingLevel(sim, time_range = time_range) expect_length(dim(fl), 3) # because t_save is 0.5, there should be 11 time steps in the range 15:20 expect_equal(dim(fl), c(11, dim(params@initial_n))) expect_identical(dimnames(fl)$sp, dimnames(params@initial_n)$sp) expect_identical(dimnames(fl)$w, dimnames(params@initial_n)$w) time_range <- 20 expect_length(dim(getFeedingLevel(sim, time_range = time_range)), 3) expect_identical( getFeedingLevel(sim, time_range = time_range)[1, , ], getFeedingLevel(sim@params, sim@n[as.character(time_range), , ], sim@n_pp[as.character(time_range), ]) ) }) test_that("getFeedingLevel passes correct time", { # Here we will check that when getFeedingLevel() is called with # a sim object, it passes the correct values of t and n at each time step. # To do this we replace mizerFeedingLevel() with a simpler function that # just returns t * n time_range <- 15:20 time_elements <- get_time_elements(sim, time_range) times <- as.numeric(dimnames(sim@effort)$time[time_elements]) e <- globalenv() # We need to define the following functions in the # global environment so that mizer can find them e$testFeedingLevel <- function(params, n, t, ...) { n * t } sim@params <- setRateFunction(sim@params, "FeedingLevel", "testFeedingLevel") expect_identical(getFeedingLevel(sim, time_range = time_range), sweep(sim@n[time_elements, , ], 1, times, "*")) }) test_that("getFeedingLevel is independent of volume", { fl <- getFeedingLevel(params) fl_r <- getFeedingLevel(params_r) expect_equal(fl, fl_r) }) # getPredRate ------------------------------------------------------------- test_that("getPredRate for MizerParams", { pr <- getPredRate(params, n, n_full) # test dim expect_identical(dim(pr), c(no_sp, no_w_full)) expect_identical(dimnames(pr)$sp, dimnames(params@initial_n)$sp) expect_identical(dimnames(pr)$w_prey, as.character(signif(params@w_full, 3))) # test value expect_known_value(pr, "values/getPredRate") }) test_that("getPredRate is independent of volume", { pr <- getPredRate(params) pr_r <- getPredRate(params_r) expect_equal(pr, pr_r) }) # getPredMort ------------------------------------------------------------------- test_that("getPredMort for MizerParams", { # Randomize selectivity and catchability for proper test set.seed(0) params@catchability[] <- runif(prod(dim(params@catchability)), min = 0, max = 1) params@selectivity[] <- runif(prod(dim(params@selectivity)), min = 0, max = 1) m <- getPredMort(params, n, n_full) expect_known_value(m, "values/getPredMort") # Look at numbers in a single prey w_offset <- no_w_full - no_w m2temp <- rep(NA, no_w) pred_rate <- getPredRate(params, n, n_full) sp <- runif(1, min = 1, max = no_sp) for (i in 1:no_w) { m2temp[i] <- sum(params@interaction[, sp] * pred_rate[, w_offset + i]) } expect_equal(m2temp, m[sp, ], check.names = FALSE) }) test_that("getPredMort for MizerSim", { time_range <- 15:20 expect_length(dim(getPredMort(sim, time_range = time_range)), 3) time_range <- 20 expect_length(dim(getPredMort(sim, time_range = time_range)), 2) ##expect_that(getPredMort(sim, time_range=time_range), equals(getPredMort(sim@params, sim@n[as.character(time_range),,], sim@n_pp[as.character(time_range),]))) aq1 <- getPredMort(sim, time_range = time_range) aq2 <- getPredMort(sim@params, sim@n[as.character(time_range), , ], sim@n_pp[as.character(time_range), ]) ttot <- 0 for (i in seq_len(dim(aq1)[1])) { ttot <- ttot + sum(aq1[i, ] != aq2[i, ]) } expect_equal(ttot, 0) }) test_that("getPredMort passes correct time", { # Here we will check that when getPredMort() is called with # a sim object, it passes the correct values of t and n at each time step. # To do this we replace mizerFeedingLevel() with a simpler function that # just returns t * n times <- as.numeric(dimnames(sim@effort)$time) e <- globalenv() # We need to define the following functions in the # global environment so that mizer can find them e$testPredMort <- function(params, n, n_pp, t, ...) { n * t } sim@params <- setRateFunction(sim@params, "PredMort", "testPredMort") expect_identical(unname(getPredMort(sim)), unname(sweep(sim@n, 1, times, "*"))) }) test_that("interaction is right way round in getPredMort function", { inter[, "Dab"] <- 0 # Dab not eaten by anything params <- newMultispeciesParams(NS_species_params_gears, inter) m2 <- getPredMort(params, get_initial_n(params), params@cc_pp) expect_true(all(m2["Dab", ] == 0)) }) test_that("getPredMort is independent of volume", { pr <- getPredMort(params) pr_r <- getPredMort(params_r) expect_equal(pr, pr_r) }) # getResourceMort --------------------------------------------------------- test_that("getResourceMort", { m2 <- getResourceMort(params, n, n_full) # test dim expect_length(m2, no_w_full) # Check number in final prey size group m22 <- colSums(getPredRate(params, n, n_full)) expect_equal(m22, m2, check.attributes = FALSE) m2b1 <- getResourceMort(params, n, n_full) # test value expect_known_value(m2b1, "values/getResourceMort") }) test_that("getResourceMort is independent of volume", { pm <- getResourceMort(params) pm_r <- getResourceMort(params_r) expect_equal(pm, pm_r) }) # getFmortGear ------------------------------------------------------------ test_that("getFmortGear", { # Two methods: # MizerParams + numeric # MizerParams + matrix # Randomize selectivity and catchability for proper test set.seed(0) params@catchability[] <- runif(prod(dim(params@catchability)), min = 0, max = 1) params@selectivity[] <- runif(prod(dim(params@selectivity)), min = 0, max = 1) no_gear <- dim(params@catchability)[1] no_sp <- dim(params@catchability)[2] no_w <- length(params@w) # Single numeric effort_num1 <- runif(1, min = 0.1, max = 1) # Numeric vector effort_num2 <- runif(no_gear, min = 0.1, max = 1) # Matrix (or 2D array) - here with 7 timesteps effort_mat <- array(runif(no_gear * 7, min = 0.1, max = 1), dim = c(7, no_gear)) # Call both methods with different effort inputs f1 <- getFMortGear(params, effort_num1) f2 <- getFMortGear(params, effort_num2) f3 <- getFMortGear(params, effort_mat) # Check dimnames are right expect_named(dimnames(f1), c("gear", "sp", "w")) expect_named(dimnames(f2), c("gear", "sp", "w")) expect_named(dimnames(f3)[2:4], c("gear", "sp", "w")) expect_identical(dim(f3), c(dim(effort_mat)[1], no_gear, no_sp, no_w)) # check fails if effort is not right size bad_effort <- rep(effort_num1, no_gear - 1) expect_error(getFMortGear(params, bad_effort)) # Check contents of output widx <- round(runif(1, min = 1, max = no_w)) sp <- round(runif(1, min = 1, max = no_sp)) gear <- round(runif(1, min = 1, max = no_gear)) expect_identical(f1[gear, sp, widx], effort_num1 * params@catchability[gear, sp] * params@selectivity[gear, sp, widx]) expect_identical(f2[gear, sp, widx], effort_num2[gear] * params@catchability[gear, sp] * params@selectivity[gear, sp, widx]) expect_identical(f3[, gear, sp, widx], effort_mat[, gear] * params@catchability[gear, sp] * params@selectivity[gear, sp, widx]) expect_known_value(f3, "values/getFMortGear") expect_equal(getFMortGear(sim)[1, 1, 1, ], getFMortGear(sim@params, effort = sim@effort[1, ])[1, 1, ]) }) # getFMort ---------------------------------------------------------------- test_that("getFMort", { effort1 <- 0.5 effort2 <- rep(effort1, no_gear) effort3 <- array(effort1, dim = c(7, no_gear)) f1 <- getFMort(params, effort1) f2 <- getFMort(params, effort2) f3 <- getFMort(params, effort3) # check that length of dims is right expect_identical(dim(f1), c(no_sp, no_w)) expect_identical(dim(f2), c(no_sp, no_w)) expect_identical(dim(f3), c(dim(effort3)[1], no_sp, no_w)) # Check dimnames are right expect_named(dimnames(f3)[2:3], c("sp", "w")) # check fails if effort is not right size expect_error(getFMort(params, c(1, 2))) # check contents of output fmg1 <- getFMortGear(params, effort1) fmg2 <- getFMortGear(params, effort2) fmg3 <- getFMortGear(params, effort3) fmg11 <- array(0, dim = c(no_sp, no_w)) fmg22 <- array(0, dim = c(no_sp, no_w)) fmg33 <- array(0, dim = c(dim(effort3)[1], no_sp, no_w)) for (i in 1:no_gear) { fmg11 <- fmg11 + fmg1[i, , ] fmg22 <- fmg22 + fmg2[i, , ] fmg33 <- fmg33 + fmg3[, i, , ] } expect_equal(f1, fmg11) expect_equal(f2, fmg22) expect_equal(f3, fmg33) expect_known_value(f1, "values/getFMort") }) test_that("getFMort passes correct time", { # Here we will check that when getFMort() calls mizerFMort(). # it passes the correct time. To implement the test we write simple # replacement for mizerFMort() that puts the time # argument into the returned array. times <- as.numeric(dimnames(sim@effort)$time) e <- globalenv() # We need to define the following functions in the # global environment so that mizer can find them e$testFMort <- function(params, n, t, ...) { n * t } sim@params <- setRateFunction(sim@params, "FMort", "testFMort") expect_identical(getFMort(sim), sweep(sim@n, 1, times, "*")) }) test_that("getFMort passes correct time", { # Here we will check that when getFMort() calls getEGrowth(). # it passes the correct time. To implement the test we write simple # replacements for mizerFMort() and mizerEGrowth that put the time # argument into the returned array. times <- as.numeric(dimnames(sim@effort)$time) e <- globalenv() # We need to define the following functions in the # global environment so that mizer can find them e$testEGrowth <- function(params, n, t, ...) { n * t } e$testFMort <- function(params, e_growth, pred_mort, t, ...) { e_growth } sim@params <- setRateFunction(sim@params, "EGrowth", "testEGrowth") sim@params <- setRateFunction(sim@params, "FMort", "testFMort") expect_identical(getFMort(sim), sweep(sim@n, 1, times, "*")) # Now we do the same for when getFMort() calls getPredMort() e$testPredMort <- function(params, n, t, ...) { n * t } e$testFMort <- function(params, e_growth, pred_mort, t, ...) { pred_mort } sim@params <- setRateFunction(sim@params, "PredMort", "testPredMort") sim@params <- setRateFunction(sim@params, "FMort", "testFMort") expect_identical(getFMort(sim), sweep(sim@n, 1, times, "*")) }) # getMort -------------------------------------------------------------------- test_that("getMort", { no_gear <- dim(params@catchability)[1] effort1 <- 0.5 effort2 <- rep(effort1, no_gear) z <- getMort(params, n, n_full, effort = effort2) # test dim expect_identical(dim(z), c(no_sp, no_w)) expect_identical(dimnames(z)$prey, dimnames(params@initial_n)$sp) expect_identical(dimnames(z)$w_prey, dimnames(params@initial_n)$w) # Look at numbers in species 1 f <- getFMort(params, effort2) m2 <- getPredMort(params, n, n_full) z1 <- f[1, ] + m2[1, ] + params@species_params$z0[1] expect_equal(z1, z[1, ], check.names = FALSE) expect_known_value(z, "values/getMort") }) test_that("getMort is independent of volume", { m <- getMort(params, effort = 1) m_r <- getMort(params_r, effort = 1) expect_equal(m, m_r) }) # getEReproAndGrowth ------------------------------------------------------ test_that("getEReproAndGrowth", { erg <- getEReproAndGrowth(params, n, n_full) # test dim expect_identical(dim(erg), c(no_sp, no_w)) expect_identical(dimnames(erg), dimnames(params@initial_n)) # Check number in final prey size group f <- getFeedingLevel(params, n = n, n_pp = n_full) e <- (f[1, ] * params@intake_max[1, ]) * params@species_params$alpha[1] e <- e - params@metab[1, ] expect_equal(e, erg[1, ]) # Can be used with infinite intake_max params@intake_max[] <- Inf expect_true(!any(is.na(getEReproAndGrowth(params, n = n, n_pp = n_full)))) erg[erg <= 0] <- 0 expect_known_value(erg, "values/getEReproAndGrowth") }) test_that("getEReproAndGrowth is independent of volume", { g <- getEReproAndGrowth(params) g_r <- getEReproAndGrowth(params_r) expect_equal(g, g_r) }) # getERepro ------------------------------------------------------------ test_that("getERepro", { es <- getERepro(params, n, n_full) # test dim expect_identical(dim(es), c(no_sp, no_w)) expect_identical(dimnames(es), dimnames(params@initial_n)) e <- getEReproAndGrowth(params, n = n, n_pp = n_full) e_repro <- params@psi * e e_repro[e_repro <= 0] <- 0 expect_identical(es, e_repro) e_growth <- getEGrowth(params, n, n_full) e_growth_man <- e - es e_growth_man[e_growth_man <= 0] <- 0 expect_identical(e_growth, e_growth_man) expect_known_value(es, "values/getERepro") }) # getRDI ------------------------------------------------------------------ test_that("getRDI", { sex_ratio <- 0.5 rdi <- getRDI(params, n, n_full) # test dim expect_length(rdi, no_sp) expect_named(rdi, as.character(params@species_params$species)) # test values e_repro <- getERepro(params, n = n, n_pp = n_full) e_repro_pop <- apply(sweep(e_repro * n, 2, params@dw, "*"), 1, sum) rdix <- sex_ratio * (e_repro_pop * params@species_params$erepro) / params@w[params@w_min_idx] expect_equal(rdix, rdi, tolerance = 1e-15, check.names = FALSE) expect_known_value(rdi, "values/getRDI") }) test_that("getRDI is proportional to volume", { rdi <- getRDI(params) rdi_r <- getRDI(params_r) expect_equal(rdi * volume, rdi_r) }) # getRDD ------------------------------------------------------------------ test_that("getRDD", { rdd <- getRDD(params, n, n_full) expect_length(rdd, no_sp) expect_named(rdd, as.character(params@species_params$species)) rdi <- getRDI(params, n, n_full) rdd2 <- getRDD(params, n, n_full, rdi = rdi) expect_identical(rdd, rdd2) rdd2 <- do.call(params@rates_funcs$RDD, list(rdi = rdi, species_params = params@species_params)) expect_identical(rdd, rdd2) expect_known_value(rdd, "values/getRDD") }) test_that("getRDD is proportional to volume", { rdd <- getRDD(params) rdd_r <- getRDD(params_r) expect_equal(rdd * volume, rdd_r) }) # getEGrowth -------------------------------------------------------------- test_that("getEGrowth is working", { eg1 <- getEGrowth(params, n = n, n_pp = n_full) # test dim expect_identical(dim(eg1), c(no_sp, no_w)) expect_identical(dimnames(eg1), dimnames(params@initial_n)) expect_known_value(eg1, "values/getEGrowth") }) # fft ---- test_that("Test that fft based integrator gives similar result as old code", { # Make it harder by working with kernels that need a lot of cutoff species_params <- NS_species_params_gears species_params$pred_kernel_type <- "truncated_lognormal" species_params$sigma[3] <- 3 species_params$beta[4] <- species_params$beta[4] * 100 species_params$beta[5] <- species_params$beta[5] / 1000 # and use different egg sizes species_params$w_min <- seq(0.001, 1, length.out = no_sp) params <- newMultispeciesParams(species_params, inter, no_w = 30, min_w_pp = 1e-12) # create a second params object that does not use fft params2 <- setPredKernel(params, pred_kernel = getPredKernel(params)) # Test encounter rate integral efft <- getEncounter(params, params@initial_n, params@initial_n_pp) e <- getEncounter(params2, params@initial_n, params@initial_n_pp) # Only check values at fish sizes fish <- outer(1:no_sp, 1:no_w, function(i, a) a >= params@w_min_idx[i]) expect_equivalent(efft[fish], e[fish], tolerance = 3e-14) # Test available energy integral prfft <- getPredRate(params, params@initial_n, params@initial_n_pp) pr <- getPredRate(params2, params@initial_n, params@initial_n_pp) expect_equivalent(prfft, pr, tolerance = 1e-15) }) # One species only ---- test_that("project function returns objects of correct dimension when community only has one species", { params <- newCommunityParams(z0 = 0.2, f0 = 0.7, alpha = 0.2) t_max <- 50 sim <- project(params, t_max = t_max, effort = 0) n <- array(sim@n[t_max + 1, , ], dim = dim(sim@n)[2:3]) dimnames(n) <- dimnames(sim@n)[2:3] n_pp <- sim@n_pp[1, ] no_w <- length(params@w) no_w_full <- length(params@w_full) # MizerParams functions expect_equal(dim(getEncounter(params, n, n_pp)), c(1, no_w)) expect_equal(dim(getFeedingLevel(params, n, n_pp)), c(1, no_w)) expect_equal(dim(getPredRate(params, n, n_pp)), c(1, no_w_full)) expect_equal(dim(getPredMort(params, n, n_pp)), c(1, no_w)) expect_length(getResourceMort(params, n, n_pp), no_w_full) expect_equal(dim(getFMortGear(params, 0)), c(1, 1, no_w)) # 3D time x species x size expect_equal(dim(getFMortGear(params, matrix(c(0, 0), nrow = 2))), c(2, 1, 1, no_w)) # 4D time x gear x species x size expect_equal(dim(getFMort(params, 0)), c(1, no_w)) # 2D species x size expect_equal(dim(getFMort(params, matrix(c(0, 0), nrow = 2))), c(2, 1, no_w)) # 3D time x species x size expect_equal(dim(getMort(params, n, n_pp, effort = 0)), c(1, no_w)) expect_equal(dim(getEReproAndGrowth(params, n, n_pp)), c(1, no_w)) expect_equal(dim(getERepro(params, n, n_pp)), c(1, no_w)) expect_equal(dim(getEGrowth(params, n, n_pp)), c(1, no_w)) expect_length(getRDI(params, n, n_pp), 1) expect_length(getRDD(params, n, n_pp), 1) # MizerSim functions # time x species x size expect_equal(dim(getFeedingLevel(sim)), c(t_max + 1, 1, no_w)) # time x species x size - default drop is TRUE, if called from # plots drop = FALSE expect_equal(dim(getPredMort(sim)), c(t_max + 1, no_w)) # time x species x size expect_equal(dim(getPredMort(sim, drop = FALSE)), c(t_max + 1, 1, no_w)) # time x gear x species x size expect_equal(dim(getFMortGear(sim)), c(t_max + 1, 1, 1, no_w)) # time x species x size - note drop = TRUE expect_equal(dim(getFMort(sim)), c(t_max + 1, no_w)) # time x species x size expect_equal(dim(getFMort(sim, drop = FALSE)), c(t_max + 1, 1, no_w)) })