Changes between Initial Version and Version 1 of PDAF_alloc

Timestamp:

Jun 5, 2025, 12:54:13 PM (2 days ago)

Author:

lnerger

Comment:

--

PDAF_alloc

@@ +1 @@
+= PDAF_alloc =
+
+This page documents the routine `PDAF_alloc` which is part of the module `PDAF_utils`. 
+
+`PDAF_alloc` is used by each DA method in its routine `PDAF_X_alloc` to allocate the arrays provided by PDAF's framework infrastructure.
+
+The interface is:
+{{{
+SUBROUTINE PDAF_alloc(dim_p, dim_ens, dim_ens_task, dim_es, dim_bias_p, &
+     dim_lag, statetask, outflag)
+
+  INTEGER, INTENT(in) :: dim_p           !< Size of state vector
+  INTEGER, INTENT(in) :: dim_ens         !< Ensemble size
+  INTEGER, INTENT(in) :: dim_ens_task    !< Ensemble size handled by a model task
+  INTEGER, INTENT(in) :: dim_es          !< Dimension of error space (size of Ainv)
+  INTEGER, INTENT(in) :: dim_bias_p      !< Size of bias vector
+  INTEGER, INTENT(in) :: dim_lag         !< Smoother lag
+  INTEGER, INTENT(in) :: statetask       !< ID of model task forecasting a single state
+  INTEGER, INTENT(inout):: outflag       !< Status flag
+
+}}}
+
+Notes:
+* For consistency, PDAF's internal arrays are always allocated, but if they are not used they are allocated with minimum size
+ * If `dim_es=0`, the array `Ainv` will be allocated with size `(1,1)`
+ * If `dim_bias_p=0`, the array `bias` will be allocated with size `(1)`
+ * If `dim_lag=0`, the array `sens` will be allocated with size `(1,1,1)`
+* The dimensions are usually set in `PDAF_X_init` and then used in `PDAF_X_alloc`.