## Do not edit this file manually. ## It has been automatically generated from *.org sources. library(sarima) context("Fitting Sarima models with estimated unit roots") ## 2018-08-23 wrapping the tests in test_that() and adding an expectation ## in the hope that this will speed up valgrind. test_that("Sarima and Arma models work ok", { expect_true(TRUE) # for now ## the air plane model (1-B)(1-B^12) + MA(1) + SMA(12, 1) sarima(log(AirPassengers) ~ 0 | ma(1, c(-0.3)) + sma(12,1, c(-0.1)) + i(1) + si(12,1), ss.method = "sarima") ## two ways to estimate the parameters of the unit root operator ## to see if the result is similar (it is) ## 2025-03-18 wrap in try() to avoid CRAN 'Additional issues' on OpenBLAS: ## Error in optim(flat_par[nonfixed], ss_sarima, use.symm = use.symm, method = "BFGS", : ## non-finite finite-difference value [1] ## An object of class "SarimaModel" ## [ FAIL 1 | WARN 10 | SKIP 0 | PASS 217 ] ## ## ══ Failed tests ════════════════════════════════════════════════════════════════ ## ── Error ('test-sarima-uar.R:17:1'): Sarima and Arma models work ok ──────────── ## Error in `optim(flat_par[nonfixed], loglik_sarima, use.symm = use.symm, ## method = "BFGS", hessian = TRUE)`: non-finite finite-difference value [1] ## Backtrace: ## ▆ ## 1. └─sarima::sarima(...) at test-sarima-uar.R:17:1 ## 2. └─sarima:::sarimat(...) ## 3. └─stats::optim(...) ## ## [ FAIL 1 | WARN 10 | SKIP 0 | PASS 217 ] ## Error: Test failures ## Execution halted ## ## TODO: check the code to make it more robust (but this works on numerical limits) try( sarima(log(AirPassengers) ~ 0 | ma(1, c(-0.3)) + sma(12,1, c(-0.1)) + uar(13, c(rep(0,12), 1), fixed = 13, atanh.tr = TRUE), ss.method = "sarima") ) sarima(log(AirPassengers) ~ 0 | ma(1, c(-0.3)) + sma(12,1, c(-0.1)) + i(2) + uar(11, c(rep(0,10), 1), fixed = 11), ss.method = "sarima") })