Index

References are by chapter and section number.

!$acc parallel loop — 35.2
-c flag (compile only) — 2.3, 8.5
-fast (nvfortran) — 30.2
-fbacktrace — 13.3, 30.3
-fcheck=all — 13.3
-fcheck=all (removed for production) — 30.3
-fcheck=all (run-time checks) — 2.4
-fcheck=bounds — 13.3, 13.6
-fcoarray (single / lib) — 32.5
-fexternal-blas — 16.1
-ffast-math — 30.1
-ffast-math (FP reordering) — 27.1
-ffpe-trap — 13.3, 20.4, 30.3
-finit-real=snan — 13.3, 13.6, 30.3
-flto (link-time optimization) — 30.1
-fno-protect-parens — 30.1
-fopt-info / -fopt-info-vec / -fopt-info-vec-missed — 27.5
-fopt-info-optimized / -fopt-info-inline — 27.5
-fp-model precise / fast (Intel) — 30.2
-fprofile-generate / -fprofile-use — 30.4
-funroll-loops — 27.1
-g (debug info) — 2.4
-I flag (module search path) — 8.5
-ipo (Intel interprocedural optimization) — 30.2
-llapack / -lblas — 21.5
-march=native — 27.1, 27.5, 30.1, 30.5
-Minfo (nvfortran) — 30.2
-O0 (no optimization) — 30.1, 30.3
-O1 — 30.1
-O2 — 30.1, 30.3
-O2 (optimization) — 2.4
-O2 (production default) — 27.1
-O3 — 30.1, 30.3
-O3 (auto-vectorization) — 27.1, 27.5
-Ofast — 30.1
-Ofast (and -ffast-math danger) — 27.1, Spaced Review
-Os (optimize for size) — 30.1
-qopt-report (Intel) — 30.2
-qopt-report (Intel) / -Minfo (nvfortran) — 27.5
-std=f2018 — 2.4
-Wall — 2.4
-xHost (Intel) — 30.2
.and. / .or. / .not. — 4.1, 4.4
.eqv. / .neqv. — 4.1
.f90 file extension — 2.2, 2.5
.mod file — 8.5
.smod file — 8.5
0.1 + 0.2 problem — 20.1
80/20 rule (profiling) — 36.5
@assertEqual — 37.1
@test annotation — 37.1
_FillValue — 25.5
!$omp atomic — 33.4
!$omp critical — 33.4
!$omp do — 33.2
!$omp parallel do — 33.2, Project Checkpoint
!$omp parallel` / `!$omp end parallel — 33.1
!$omp sections` / `!$omp section — 33.2
!$omp single — 33.2, 33.4
!$omp task` / `!$omp taskwait — 33.5
!$omp workshare — 33.2
!f2py directive — 15.2, 15.3
-fopenmp (compile flag) — 33.1
-std=legacy (gfortran flag) — 17.1, 17.6
.f2py_f2cmap (custom kind mapping) — 15.5, Project Checkpoint
assert np.allclose (same-answer check) — 15.5
bind(c) (for FFIs) — 15.4
default(none) — 33.3
firstprivate clause — 33.3
intent(hide) / depend — 15.2
intent(inout) (F-contiguity requirement) — 15.3
lastprivate clause — 33.3
np.isfortran / np.asfortranarray — 15.3
omp_get_num_threads — 33.1
omp_get_thread_num — 33.1
omp_lib module — 33.1
OMP_NUM_THREADS — 33.1
order='F' — 15.3, Project Checkpoint
private clause — 33.3
reduction clause — 33.3
schedule(dynamic) — 33.4
schedule(guided) — 33.4
schedule(static) — 33.4
shared clause — 33.3
time.perf_counter — 15.5

A

A edit descriptor (character) — 7.1
A-stability — 23.5
abs — 3.5
absolute stability / stability region — 23.5
absorption — 20.3
abstract interface (for a deferred binding) — 10.4
abstract interface (for the RHS) — 23.1, 23.2, 23.4
abstract interface (pure) — 22.2, 22.3
abstract solver_t (framework) — 10.6
abstract type — 10.4
accelerator — 31.3, 35.1
action (open specifier) — 7.3
actual argument — 6.1
adaptive integration — 22.3
adaptive step size — 23.3
AddressSanitizer / -fsanitize=address — 13.4
adjustl — 12.2
adjustr — 12.2
aerospace (Fortran employers) — 40.1
aimag — 3.1
aliasing (pointers/targets) — 29.4
aliasing (rigorous) — 27.3
allocatable array — 5.5
allocatable array (replacing over-dimensioned static) — 19.1
allocatable component — 9.4
allocatable components self-deallocate — 10.4
allocatable vs pointer — 11.4
allocate / deallocate — 5.5
allocated — 5.5
allocated (for strings) — 12.1
alpha (thermal diffusivity) — Project Checkpoint
Amdahl's Law — 31.2, 31.4, 31.5, 38.4
amplification factor — 23.5, 24.4
amplification factor (G) — 38.3
analytical solution — 38.3
analytical solution (as oracle) — 37.2
app/ src/ test/ layout — 16.3
architectural role — 36.2
argument association — 6.1, 6.3
arithmetic IF — 17.4
arithmetic IF, translating — 19.2
arithmetic intensity — 16.1, 21.2, 28.3, 29.2, 35.4, 35.5, 38.4
arithmetic intensity (flop/byte) — Case Study 2
arithmetic operators — 3.3
array (first-class object) — 5.1
array bounds (custom) — 5.1
array constructor — 5.3
array descriptor — 6.6
array layout contract — 14.5
array of derived types — 9.1, 9.5
array section — 5.2
Array-of-Structures / Structure-of-Arrays — 9.5
asind / acosd / atand / atan2d — 39.1
assembled solver architecture — 38.1
assert harness (hand-rolled) — 37.1
assertion — 13.5
assertion (in testing) — 37.1
associated (intrinsic) — 11.2
associated (two-argument form) — 11.2
association status — 11.2
assumed-length character (len=*) — 12.1, 12.5
assumed-shape argument (requires explicit interface) — 8.2
assumed-shape array — 6.6
assumed-shape array (replacing assumed-size) — 19.3
assumed-size array — 6.6
assumed-size array, translating — 19.3
assumed-size to assumed-shape — 18.1
asterisk fill (field overflow) — 7.1
atomic operations (atomic_add) — 32.3
attributes(global) — 35.3
auto-vectorization — 29.3
automatic deallocation — 5.5, 9.4, 11.4
automatic reallocation on assignment — 12.1, 12.3
axpy (BLAS Level-1) — 28.3
axpy / daxpy kernel — 27.3, 27.4

B

backward compatibility — 39.2
backward difference — 22.1
backward GOTO (loop) — 19.2
barrier (!$omp barrier) — 33.4
base case — 6.5
baseline -march (conservative target) — 30.5
benchmark methodology — 28.5
big-bang rewrite — 18.2
binary tree — 11.3
bind(c) — 14.2, 14.3, 14.4, 14.5
binding label — 14.2
bit-for-bit — 37.2
bit-for-bit reproducibility — 18.3
BLAS — 1.2, 16.1, 21.2, 21.5
BLAS (tuned; dgemm) — 29.2, 29.5
BLAS levels (1/2/3) — 16.1, 21.2
BLOCK DATA — 17.2
BLOCK DATA (mentioned) — 8.4
BLOCK DATA, translating — 19.1
blocked matrix multiply — 29.2
boundary vs interior points — Project Checkpoint
breakpoint — 13.4
build configuration — 36.3
build provenance / manifest — 30.4
build system — 36.3
building on the target node — 30.5
Butcher tableau — 23.2
by-reference passing — 14.4, 14.5
by-value passing — 14.3

C

C ABI — 15.4
C application binary interface (ABI) — 14.1, 14.2
C interoperability (Fortran 2023 additions) — 39.1
C linkage — 14.2
C struct interoperability — 14.5
C-contiguous — 15.3
C-interoperable kind — 14.1
c_associated — 14.1
c_bool — 14.1, 14.5
c_char — 14.1, 14.5
c_double — 14.1, 14.3, 14.4
c_f_pointer — 11.5, 14.1, 14.5
c_float — 14.1
c_funptr — 14.1
c_int — 14.1, 14.4
c_int64_t — 14.1
c_loc — 14.1, 14.5
c_null_char — 14.1, 14.5
c_null_ptr — 14.1
c_ptr — 14.1, 14.5
c_size_t — 14.1
c_sizeof — 14.1, 14.5
cache blocking — 21.2
cache blocking (tiling) — 29.2, Project Checkpoint
cache line — 5.6, 27.2, 29.1, 29.2, 33.5
cache miss rate — 28.4
cachegrind — 28.4
call graph — 28.2
call hoisting / common-subexpression elimination — 27.4
callback — 11.5
calling C from Fortran — 14.3
calling Fortran from C — 14.4
calling the parent's method — 10.1
Cannot open module file (error) — 8.5
capstone (heat solver) — 38.1, Project Checkpoint
career spectrum (science ↔ software) — 40.2
carriage control (history) — 7.1
case default — 4.2
case range — 4.2
catastrophic cancellation — 20.3
catastrophic cancellation (in differentiation) — 22.4
CELL_DATA (contrast) — 26.2
central difference — 22.1, 22.4
central second difference — 24.2
CESM (climate model) — 1.2, 36.1
CF conventions — 25.5
cffi — 15.4
CFL condition — 24.4, 38.2
CFL condition (as an ODE stability limit) — 23.4
cg_annotate — 28.4
character (type) — 3.1
character type — 12.1
characterization test — 18.2
chemical kinetics (reaction ODEs) — 23.6
chunking — 25.3
CI with two compilers — 30.5
circular module dependency — 8.5, 8.6
class (passed object) — 9.2
class (polymorphic declaration) — 10.2, 10.5, 10.6
class default (guard) — 10.3
class is (guard) — 10.3
classof — 39.1
clock speed plateau — 31.1
close statement — 7.3
close-enough (tolerance) comparison — 18.3
CMake — 16.5, 36.3
co_broadcast — 32.4
co_max / co_min — 32.4
co_reduce — 32.4
co_sum — 32.4
coalesced memory access — 35.5 (spaced review)
coarray — 32.1, 32.2, 32.3, 32.4, 32.5
coarrays (as parallel model) — 31.3
coarse-grained vs fine-grained polymorphism — 10.5, 10.6
codimension — 32.2
coindexed access — 32.2
collapse clause — 35.2 (project checkpoint)
collective communication — 34.3
collective subroutine — 32.4
colormap — 26.5
column-major / row-major (Python view) — 15.3
column-major loop order — 29.1, Project Checkpoint
column-major loop order (payoff of Ch.5) — 27.2, Project Checkpoint
column-major order — 1.3, 5.6
column-major order (as used by EQUIVALENCE) — 17.3
column-major storage (matrices) — 21.1
column-major vs row-major (interop) — 14.5
column-major write order — 26.2
comment (!) — 2.5
comment (C in column 1 vs !) — 19.4
comment line (C/* in column 1) — 17.1
commit hash (tie a result to) — 37.4
COMMON block — 1.4, 17.2
COMMON block (legacy) — 8.4
COMMON block to module — 18.1, 18.4
COMMON block, translating — 19.1
COMMON-to-module conversion — 8.4
communicator — 34.1
comparing floats with == (pitfall) — 20.1
compilation order — 8.5, 8.6
compile-link-run cycle — 2.3
compiler — 1.1
compiler (installing) — 2.1
compiler error vs linker error — 2.3
Compiler Explorer (godbolt) — 27.5, Further Reading
Compiler Explorer (godbolt.org) — 16.5
compiler flag — 2.4
compiler matrix — 37.3
compiler_options (intrinsic) — 37.5
compiler_options() — 30.4
compiler_version (intrinsic) — 37.5
compiler_version() — 30.4
complex (type) — 3.1
component — 9.1
component section (array%comp) — 9.1
compression — 25.3
computational scientist — 40.2
computational-science paper — 38.6
compute-bound — 28.3
compute-bound kernel — 29.2
computed GOTO — 17.4
computed GOTO, translating — 19.2
concatenation (//) — 12.1, 12.3
condition number — 20.5
conditional expression — 39.1, Project Checkpoint
conditional-compilation sentinel (!$) — 33.1
conditioning — 20.5
conditioning (linear solve) — 21.3
conformable / conformance — 5.3, 5.5
conjugate trick — 20.3
constructor (user-defined) — 9.4
contains — 6.5
contiguous (attribute) — 11.5
contiguous (foreshadowed) — 6.6
contiguous attribute — 29.4, Project Checkpoint
continuation (&) — 2.5
continuation (column 6 vs trailing &) — 19.4
continuation column (column 6) — 17.1
continuous integration — 37.3
Conventions attribute — 25.5
convergence — 22.4
convergence order (as test) — 37.2
convergence plot (log-log, slope 2) — 38.5
convergence study — 38.3
coordinate variable — 25.5
copy-in/copy-out — 29.4
corank — 32.2
coshape — 32.2
count — 5.4, 5.7
count_max (clock wrap-around) — 28.1
count_rate — 28.1
counted do loop — 4.3
Courant number — 24.4
cProfile (Python) — 28.2
CPU frequency scaling (turbo) — 28.5
CPU time — 28.1
cpu_time (intrinsic) — 28.1
critical construct — 32.3
ctags / tags index — 36.4
ctypes — 15.4
CUDA — 35.5
CUDA Fortran — 35.3
curse of dimensionality — 22.5
cycle — 4.4
cycle (replacing GOTO) — 19.2

D

D1 / LLd cache misses — 28.4
dangling pointer — 11.2
dangling pointer (revisited; owned Ch. 11) — 13.4
data clause (copyin/copyout/copy/create/present) — 35.2, 35.4
data flow — 36.5
data parallelism — 31.4
data race — 33.3
data race (previewed) — 31.3
data region (!$acc data) — 35.2, 35.4
data scoping — 33.3
DATA statement — 17.5
DATA statement, translating — 19.1
data-sharing attribute — 33.3
DataArray (VTK XML) — 26.2
dataset (HDF5) — 25.3
dead-code elimination (benchmark trap) — 28.5
dead-code elimination (checksum guard) — Case Study 2
deadlock (send/recv) — 34.2
declaration initializer — 19.1
declared type — 10.2
deep copy — 9.4
deep copy on assignment — 11.4
default initialization (component) — 9.1
defensive programming — 13.5
deferred binding — 10.4
deferred shape — 5.5
deferred-length character — 12.1, 12.3
deferred-length character component — 9.4
deferred/assumed type parameter — 9.3
definable entity — 6.2
deflate — 25.3
degree-valued trig (sind, cosd, tand) — 39.1, Project Checkpoint
deleted/removed feature — 19.5
Dennard scaling — 31.1
dependency graph — 8.6
dependency pinning (tag) — 16.3
derived type — 9.1
derived type (replacing EQUIVALENCE / COMMON) — 19.1
determinism (result vs. schedule) — Project Checkpoint
development build vs release build — 30.3
development vs release build — 2.4
device — 35.1
device array (device attribute) — 35.3
dgecon — 21.3
dgemm — 21.2
dgemm / tuned BLAS (compute-bound) — 28.3
dgesv — 16.1, 21.3
dgesvd — 16.1, 21.4
dgtsv (tridiagonal solve) — 21.6
diffusion equation — 24.1
diffusion number (mesh Fourier number) — 24.3, 24.4
diffusion number (r) — 38.2
diffusion number limit (r <= 1/4) — 24.4
digits (intrinsic) — 20.2
DIMENSIONS (VTK) — 26.2
diminishing returns — 29.5
direct access — 7.3
direct sparse solvers (SuperLU/MUMPS/PETSc) — 21.6
directive-based offload — 35.2
directory layout — 36.1
Dirichlet boundary condition — 24.5
disassociated pointer — 11.2
discrete maximum principle (convex-combination bound) — Project Checkpoint
dispatch cost / inlining barrier — 10.5
distributed memory — 31.3
distributed memory (MPI as) — 34.1
do concurrent — 4.6
do concurrent (performance use) — 29.3, Project Checkpoint
do concurrent (requires purity) — 27.4
do concurrent reduce — 39.4
do loop — 4.3
do while — 4.3
doc comment (!> / !!) — 37.4
domain decomposition — 34.4
domain decomposition (previewed) — 31.3, 31.5
domain decomposition (previewed; owned Ch.34) — 32.5, Project Checkpoint
domain scientist — 40.2
dot_product — 5.4, 21.1
double precision — 3.1, 3.2
double precision (binary64) — 20.1, 20.6
dp (double-precision kind) — 3.2
DPI (figure resolution) — 26.5
driver program — 36.2
dsyev — 16.1, 21.4
dtype — 15.3
dtype matching (NumPy ↔ Fortran kind) — 15.3
dummy argument — 6.1
dummy-argument association restriction — 27.3
dx (grid spacing) — Project Checkpoint
dynamic dispatch — 10.2, 10.5
dynamic type — 10.2

E

E edit descriptor (scientific) — 7.1
eager vs rendezvous protocol — 34.2
edit descriptor — 3.7, 7.1
edit descriptor i0 — 12.4
edit descriptor i6.6 — 12.4
eigenvalue problem — 21.4
eigenvector property (stencil) — 38.3
EISPACK — 16.1
elemental — 6.4
elemental (optimization implications) — 27.4
elementwise operation — 5.3
else if — 4.1
elsewhere — 4.5
embarrassingly parallel — Project Checkpoint
embedded Runge-Kutta pair (RKF45, Dormand-Prince) — 23.3
encapsulation — 8.2
end-of-file — 7.6
endianness — 7.5
endianness (portability) — 25.1, 25.3
energy drift / symplectic integrator — 23.4, 23.6
energy sector (seismic, reservoir, reactor codes) — 40.1
ENTRY — 17.4
entry point — 36.4
enum, bind(c) (Fortran 2003) — 39.1
enumeration type — 39.1, Project Checkpoint
epsilon (intrinsic) — 20.2
EQUIVALENCE — 17.3
EQUIVALENCE (modernizing) — 18.1
EQUIVALENCE, translating — 19.1
errmsg (allocation) — 13.1
error estimate (compare resolutions) — 22.3
error stop — 13.1, 13.2
error stop (I/O guard) — 7.6
error stop (in modernization) — 18.1, 18.4
error termination — 13.2
error_unit (standard error) — 13.1, 13.5
ES edit descriptor (scientific, mantissa 1-9) — 7.1
Euler's method — 23.1, 23.4, 23.5
Euler's method, backward (implicit) — 23.5
exact-zero comparison (arithmetic IF trap) — 19.2
exactness (test on polynomials) — 37.1
executable — 2.3
Exercise 1.28 (the sealed answer) — What's Next
exit — 4.3, 4.4
exit (replacing GOTO) — 19.2
exit code — 13.2
exit code (test failure) — 37.1, 37.3
explicit interface — 6.5, 6.6, 8.2
explicit scheme — 24.3
explicit-shape array — 6.6
exponent bias — 20.1
exponentiation — 3.3, 3.5
extends (attribute) — 10.1, 10.6
extension module — 15.2
extension type / child type — 10.1
extent — 5.1
external procedure — 6.5

F

F edit descriptor (fixed-point) — 7.1
F-contiguous — 15.3
f0.2 edit descriptor — 2.7
f2py — 15.2, 15.3, 15.5
f2py build command (-c -m) — 15.2
fall-through (absence of) — 4.2, 4.7
false sharing — 33.5
fast Fourier transform — 16.2
feature paper (committee) — 39.2
FFTW — 16.2
field width — 7.1
field-width overflow (asterisks) — 12.4
field_t (project) — 9.5, Project Checkpoint
final procedure (destructor) — 10.4
finalization — 10.4
findent / fprettify (conversion tools) — 18.1
finite difference — 22.1, 22.4
finite-difference method — 24.2
five-point stencil — 17.6, 24.2, 38.2
five-step reading method — 17.6
fixed-form source — 2.5, 17.1
fixed-form to free-form — 19.4
fixed-form to free-form conversion — 18.1
fixed-length character — 12.1, 12.6
flat profile — 28.2
floating-point exception — 13.3
floating-point exception trap — 13.3
floating-point representation — 20.1
forall (obsolescent) — 4.6
FORD — 16.5, 37.4
fork–join model — 33.1
form team / change team — 32.4
format string — 3.7, 7.1
formatted output — 3.7
fortls (Fortran Language Server) — 16.5, 39.3
fortls (Fortran language server) — 36.4
Fortran 2018 — 1.1, 1.6
Fortran 2023 — 1.1
Fortran 2023 (ISO/IEC 1539-1:2023) — 39.1, 39.2
Fortran 202Y (informal next revision) — 39.4
FORTRAN 77 — 1.1
FORTRAN 77 fixed-length strings — 12.6
Fortran 90 — 1.1
Fortran Discourse — 16.4
FORTRAN I (1957) — 1.1
Fortran Playground — 39.3
Fortran programmer shortage — 40.3
fortran-lang — 16.4, 39.3
forward difference — 22.1
forward-Euler (time stepping) — 24.3
fpm — 16.3
fpm (derives compile order) — 8.5, 8.6
fpm (Fortran Package Manager) — 36.3, 39.3
fpm new / build / run / test — 16.3
fpm.toml — 36.3
fpm.toml (manifest) — 16.3
frame_name (indexed filenames) — 26.3, Project Checkpoint
free lunch (end of) — 31.1
free-form source — 1.1, 1.4, 2.5
frozen interface (stable signature) — 38.1
FTCS (Forward-Time Centred-Space) — 24.3
FTCS scheme — 38.2
function — 6.1
fused multiply-add (FMA) — 30.1

G

Gauss-Legendre rule — 22.3
Gaussian quadrature — 22.3
gdb — 13.4
generics — 39.4
get (remote read) — 32.2
gfortran — 2.1, 2.2, 2.4
gfortran --version — 2.1
ghost cell — 24.5, 34.4
git dependency — 16.3
git tag (release / submission) — 37.4
GitHub Actions — 37.3
global order of accuracy (integrator) — 23.1, 23.2
global shared state — 17.2
global state — 36.5
gnuplot — 26.4
GO TO (unconditional) — 17.4
golden file / reference — 37.2
golden output / reference capture — 18.3
GOTO to structured control — 18.1, 18.4
GOTO, translating — 19.2
governing equation (heat equation) — 38.2
GPGPU (general-purpose GPU computing) — 35.5
gprof — 28.2
GPU (graphics processing unit) — 31.3, 35.1, 35.5
GPU offload (do concurrent) — 39.4
grep (navigation) — 36.4
grid cell (data structure) — 9.5
group (HDF5) — 25.3
Gustafson's Law — 31.2, 31.4

H

h5dcreate_f / h5dopen_f / h5dwrite_f / h5dread_f / h5dclose_f (dataset) — 25.3, 25.4
h5dump (header inspection) — 25.3
H5F_ACC_TRUNC_F / H5F_ACC_RDONLY_F — 25.3, 25.4
h5fc (compile helper) — 25.3
h5fcreate_f / h5fopen_f / h5fclose_f — 25.3, 25.4
h5gcreate_f / h5gopen_f / h5gclose_f / h5gn_members_f (group) — 25.3
h5open_f / h5close_f (library init) — 25.3
H5P_DATASET_CREATE_F — 25.3
h5pcreate_f / h5pset_chunk_f / h5pset_deflate_f / h5pset_shuffle_f / h5pclose_f (property list) — 25.3
h5screate_simple_f / h5sclose_f (dataspace) — 25.3
H5T_NATIVE_DOUBLE / H5T_IEEE_F64LE — 25.3
halo (halo cell) — 34.4
halo exchange — 32.2, 34.4, Project Checkpoint
halving test (measuring order) — 22.4
hand-tuning (SIMD intrinsics, manual unroll) — 29.4
harmonic oscillator (as a system) — 23.4
HDF5 — 16.2, 25.3, 25.4
hdf5 module — 25.3
heat (program / driver) — Project Checkpoint
heat equation — 1.6, 24.1, 24.3
heat_io module — 8.6, Project Checkpoint
heat_solver module — 8.6, Project Checkpoint
heat_types module — Project Checkpoint
Hello, world — 2.2
heterogeneous collection (array of boxes idiom) — 10.2
hid_t / hsize_t (HDF5 kinds) — 25.3
hidden bit — 20.1
high-level language — 1.1
high-performance computing (HPC) — 1.2
history attribute (provenance) — 25.5
Hollerith constant — 17.5
Hollerith constant, translating — 19.5
Homebrew (macOS) — 2.1
honest benchmarking — 15.5
honest performance claims — 38.6
host — 35.1
host association — 6.5
host association (submodule) — 8.3
host/device model — 35.1
host–device transfer — 35.4
hot kernel / hot loop — 15.1, 15.5
hot loop — 28.3
hot spot — 28.3
HPC engineer — 40.2
huge (intrinsic) — 20.2
hybrid MPI+OpenMP — 34.5

I

I edit descriptor (integer) — 7.1
I/O module — 36.2
I/O-bound — 28.1
IEEE 754 — 20.1
IEEE 754, relaxed by -Ofast — 30.1
ieee_arithmetic module — 20.4
ieee_is_finite — 20.4
ieee_is_nan — 20.4
ierr (error-code argument) — 34.1
if construct — 4.1
if, logical (single-line) — 4.1
ifort (Intel Fortran, deprecated) — 30.2
ifx (Intel Fortran, LLVM-based) — 30.2
illegal instruction (from -march=native) — 30.1, 30.5
image — 32.1, 32.2
image_index (intrinsic) — 32.2
ImageData — 26.1, 26.2
implicit heat step (backward Euler) — Project Checkpoint
implicit interface — 8.2
implicit method — 23.5
implicit none — 2.6
implicit none (in modernization) — 18.1
implicit none (replacing implicit typing) — 19.4
implicit save — 13.6
implicit SAVE — 17.5
implicit SAVE (initializer trap) — 19.1
implicit save (module variable) — 8.2
implicit scheme — 24.3
implicit typing — 2.6
implicit typing (I–N rule) — 17.5
implicit typing, translating — 19.4
implied-do — 5.3
implied-DO — 17.5
import (in an interface body) — 10.4, 10.6
import statement — 14.3
imshow (matplotlib) — 26.4, 26.5
INCLUDE — 17.2
INCLUDE (legacy) — 8.4
incremental modernization — 18.2
index (intrinsic) — 12.2, 12.3
index-based structure — 11.3
indexed output filename (heat_NNNNNN.vtk) — 12.4, Project Checkpoint
Inf (infinity) — 20.4
infinite do — 4.3
info — 21.3
initial-value problem (IVP) — 23.1
inlining — 27.1, 27.4
inquire statement — 7.3
int64 clock counters — 28.1
integer (type) — 3.1
integer division — 3.3, 3.4
integer loop counter — 4.3
integer overflow — 13.6
integer-division trap — 3.4
Intel MKL — 16.1, 21.5
intent — 6.2
intent (adding to legacy args) — 18.1
intent → Python signature — 15.2
intent(in) — 6.2
intent(inout) — 6.2
intent(out) — 6.2
intent, adding to legacy procedures — 19.3
interactive Fortran / REPL — 39.3
interface block (for C) — 14.3
interface checking (argument matching) — 8.2
interface vs implementation separation — 8.3
internal file — 12.4
internal procedure — 6.5
internal procedure (contrasted) — 8.1
internal procedure (replacing statement function) — 19.3
internal read (text to number) — 12.4
internal write (number to text) — 12.4
interoperable derived type — 14.5
intrinsic function — 3.2, 3.5
intrinsic module — 14.1
intrinsic type — 3.1
invariant (as test) — 37.1
iomsg — 7.6
iostat — 7.6
iostat_end — 7.6
iostat_eor — 7.6
ipiv (pivot array) — 21.3
is_contiguous — 11.5
ISO/IEC 1539-1 — 39.2
iso_c_binding — 14.1, 14.2, 14.3
iso_fortran_env — 3.2
iso_fortran_env / real64 — 2.7
iterative solvers (CG/GMRES) — 21.6

J

J3 committee — 39.2
Jacobi relaxation — 17.6
Jacobi relaxation (legacy PLATE) — 18.4
jump table — 4.2

K

Kahan (compensated) summation — 20.3, case study 2
Karp-Flatt metric — 31.5 (case study 1)
kernel — 35.3
keyword argument — 6.3
keyword argument (requires explicit interface) — 8.2
kind (intrinsic) — 3.2
kind parameter — 3.2
kind parameter (PDT) — 9.3
kinds module (dp) — 8.1, 8.6
kinds module (project) — Project Checkpoint

L

labeled DO / shared CONTINUE, translating — 19.2
labeled field (data structure) — 9.4, 9.5
Language Server Protocol (LSP) — 16.5
LAPACK — 1.2, 16.1, 16.2, 21.1, 21.2, 21.3, 21.4, 21.5, 21.6
LAPACK (used) — 29.2, 29.5
LAPACK routine naming — 16.1, 21.3, 21.4
Laplace's equation — 24.1, 24.6
Laplace's equation (steady state) — 17.6
laplacian (procedure) — 24.2, Project Checkpoint
Laplacian operator — 24.1, 24.2
launch configuration (<<>>) — 35.3
layered design (depend downward) — 8.6
ldb — 21.3
leading dimension (lda) — 21.3
leaked identifiers (HDF5 handles) — 25.3
least squares (dgels) — 21.4, case-study-02
legacy code — 1.4
Legendre polynomial (roots as nodes) — 22.3
len — 12.1, 12.2
len_trim — 12.1, 12.2
length parameter (PDT) — 9.3
LFortran — 16.4, 39.3
libgfortran (linking) — 14.4
link-time optimization — 30.1
linked list — 11.3
linker — 2.3
linking LAPACK/BLAS — 21.5
LINPACK — 16.1
LINPACK benchmark — 16.1
list-directed I/O — 7.2
list-directed internal read — 12.4
LLVM Flang — 39.3
local truncation error — 23.1
lock / unlock (lock_type) — 32.3
logical (type) — 3.1
logical operator — 4.1
longer source lines (Fortran 2023) — 39.1
loop fission (distribution) — 29.1
loop fusion (jamming) — 29.1, Project Checkpoint
loop interchange (compiler) — 27.2
loop order (cache locality) — 28.4
loop order (column-major habit) — 4.4
loop order (column-major) — 5.6
loop reordering (for column-major) — 29.1
loop trip count — 4.3
loop unrolling — 27.1
loop-carried dependency — 15.1, 15.5, 29.3
loop-invariant hoisting — 29.1, Project Checkpoint
LU factorization — 16.1, 21.3

M

machine balance — 29.5, 38.4
machine epsilon — 20.2
Make — 16.5
Make (build system) — 36.3
Makefile — 36.3
manufactured solution — 38.3
mask — 5.7
masked reduction — 5.7
masked whole-array assignment — 4.5
master thread — 33.1
matmul — 5.4
matplotlib (plotting the field) — Project Checkpoint
matplotlib post-processing — 26.5
matrix ASCII format — 26.4
matrix multiply (hand-rolled) — 21.2
matrix product (matmul) — 21.1
max / min (intrinsics) — 3.5
maximum principle (as test) — 37.1, 37.2
maxloc / minloc — 5.4
maxval / minval — 5.4
meaning-preserving optimization — 27.1
memory aliasing — 17.3
memory bandwidth / traffic — 27.2
memory overlay — 8.4
memory-bound — 27.2, 28.3, Project Checkpoint
memory-bound kernel — 29.2, 29.5, 38.4, Project Checkpoint
merge — 5.7
merge intrinsic (contrast) — 39.1
Meson — 16.5, 36.3
Meson build backend (f2py, Python 3.12+) — 15.2
message passing — 34.1
method — 9.2
method of lines — 23.4, 24.1, 24.3, Project Checkpoint
midpoint method (RK2) — 23.2
minimum vs median (statistic) — 28.5
mixed-language compile and link — 14.3, 14.4
mixed-mode arithmetic — 3.3
mod — 3.5
modern vs legacy string handling — 12.6
modernization recipe (eight steps) — 18.1
module — 1.4, 8.1, 8.2, 8.4, 8.6
module (replacing COMMON) — 19.1
module function / module subroutine prefix — 8.3
module hierarchy — 8.6, 36.1
module load (HPC) — 2.1
module map — 36.4
module map (final solver) — 38.1
module namespacing (f2py) — 15.2
module procedure — 8.1
module procedure shorthand (end procedure) — 8.3
module variable — 8.1, 8.2
module variable (as hidden input) — 36.5
modulo — 3.5
Monte Carlo integration — 22.5
Moore's Law — 31.1
move_alloc — 11.4
MPI — 16.2
MPI (Message Passing Interface) — 34.1, 34.5
MPI-IO — 34.5
mpi_allreduce — 34.3
mpi_bcast — 34.3
mpi_comm_rank — 34.1
mpi_comm_size — 34.1
MPI_COMM_WORLD — 34.1, 34.2, 34.3, 34.4
MPI_DOUBLE_PRECISION (datatype) — 34.2
mpi_finalize — 34.1
mpi_gather — 34.3, 34.4
mpi_init — 34.1
mpi_irecv — 34.4
mpi_isend — 34.4
MPI_PROC_NULL — 34.4
mpi_recv — 34.2
mpi_reduce — 34.3
mpi_scatter — 34.3
mpi_send — 34.2
mpi_sendrecv — 34.2, 34.4
MPI_STATUS_SIZE (status) — 34.2
MPI_SUM / MPI_MAX (reduce ops) — 34.3
mpi_wait / mpi_waitall — 34.4
mpif90 (compiler wrapper) — 34.5
mpirun (-np) — 34.5
MSYS2 (Windows) — 2.1
multi-GPU (one MPI rank per GPU) — project checkpoint
multicore (shift to) — 31.1
multidimensional array — 5.2
mutual exclusion — 33.4

N

name mangling — 14.2
name mangling (trailing underscore) — 15.4
name mangling (trailing underscore, MAIN__) — 28.2
name= clause — 14.2
named constant — 3.6
named constant in a module — 8.1
named construct — 4.4
named integer constant idiom ("poor man's enum") — 39.1
named loop / named construct — 19.2
namelist — 7.4
namelist group — 7.4
NaN — 20.4
NaN / Inf (used; owned Ch. 20) — 13.3, 13.6
national laboratories (as Fortran employers) — 40.1
ncdump (CDL header) — 25.2, 25.5
nearest (intrinsic) — 20.2
nested loops — 4.4
nested type — 9.1
NetCDF — 16.2, 25.2, 25.4, 25.5
netcdf module — 25.2
netcdf-fortran (Fortran interface) — 16.2
Neumann boundary condition — 24.5
never rewrite what you can refactor — 18.2
Newton's method (for implicit steps) — 23.5
newunit — 7.3
nf-config (compile helper) — 25.2
nf90_close — 25.2
nf90_create — 25.2
nf90_def_dim — 25.2
nf90_def_var — 25.2, 25.5
NF90_DOUBLE / NF90_GLOBAL / NF90_CLOBBER / NF90_NETCDF4 / NF90_UNLIMITED — 25.2, 25.4
nf90_enddef (define mode vs data mode) — 25.2
nf90_get_att — 25.4
nf90_get_var — 25.4
nf90_inq_dimid — 25.4
nf90_inq_varid — 25.4
nf90_inquire_dimension — 25.4
nf90_noerr / nf90_strerror (status check) — 25.2
nf90_open — 25.4
nf90_put_att — 25.2, 25.5
nf90_put_var — 25.2
nine-point stencil — 24.2
no-aliasing advantage — 1.3, 11.4
no-aliasing advantage (used) — 29.3, 29.4
node (quadrature) — 22.2, 22.3
node (self-referential derived type) — 11.3
non-associativity (floating point, in testing) — 37.2
non-associativity of addition — 20.3
non-blocking communication — 34.4
nopass — 9.2
normal equations — case-study-02
null-terminated string — 14.5
nullify — 11.2
num_images (intrinsic) — 32.1
numerical equivalence — 18.3
numerical stability — 20.5
numpy.loadtxt — 26.4, 26.5
nvfortran (-acc / -cuda / -Minfo=accel) — 35.2, 35.3
nvfortran (NVIDIA HPC SDK) — 30.2

O

Oak Ridge / Livermore / Los Alamos / Argonne / Sandia — 40.1
object file — 2.3
object file (.o) — 8.5
obsolescent feature — 4.6, 19.5, 39.2
offload — 35.1, 35.2
one module per file convention — 36.1, 36.4
one responsibility per module — 8.6
one-based indexing — 5.1
one-based thread index — 35.3
open statement — 7.3
open-source contribution — 40.5
OpenACC — 35.2
OpenBLAS — 16.1, 21.5
OpenCoarrays (caf / cafrun) — 32.5
OpenMP — 33.1, 33.2, 33.3, 33.4, 33.5
operator precedence — 3.3
optimal step size — 22.4
optimization exposes latent bugs — 30.3
optimization level — 30.1
optimization level (-O0/-O1/-O2/-O3) — 27.1
optimization license — 27.4
optimization report — 27.5
optimizing compiler — 27.1
optional argument — 6.3
optional argument (requires explicit interface) — 8.2
optional output argument — 6.3
oracle problem — 37.1
orbital motion (two-body / Kepler) — 23.6
order of accuracy — 22.4
order of accuracy (observed) — 38.3
order of accuracy (second-order) — 24.2
ORIGIN / SPACING (VTK) — 26.2
out-of-bounds access — 13.6
overflow — 20.4
overflow / NaN (from instability) — 24.4
overriding a type-bound procedure — 10.1

P

packed vs scalar instructions (…pd vs …sd) — 27.5
paper structure (IMRaD for simulation) — 38.6
parallel efficiency — 31.4
parallel overhead — 31.4
parallel region (!$omp parallel) — 33.1
parallelizable fraction — 31.2
parallelization plan — 31.5
parameter (named constant) — 2.7, 3.6
parameterized derived type (PDT) — 9.3
parametric polymorphism — 39.4
ParaView — 26.3, 26.5
parent component (self%parent_t) — 10.1
parent type / base type — 10.1
partial differential equation (PDE) — 24.1
partial pivoting — 21.3
particle (data structure) — 9.1, 9.5
partitioned global address space (PGAS) — 31.3, 32.1, 32.5
pass attribute — 9.2
pass by reference (Fortran) — 15.4
pass(name) — 9.2
passed-object dummy argument — 9.2
per-step transfer (pitfall) — 35.4 (case study 1)
percent operator (%) — 9.1
perceptually uniform colormap (viridis/inferno) — 26.5
perf (hardware counters) — 28.2, 28.4
performance analysis — 38.4
periodic boundary condition — 24.5
pFUnit — 16.5, 37.1
physics module — 36.2
PLATE (legacy kernel, modernized) — 19.1, 19.2, 19.3, Project Checkpoint
PLATE kernel — 17.6
PLATE kernel (modernized) — 18.4
playground (play.fortran-lang.org) — 16.4
pm3d (gnuplot) — 26.4
point ordering (x fastest) — 26.2
point-to-point communication — 34.2
POINT_DATA — 26.2
pointer — 11.1, 11.2, 11.4
pointer aliasing — 1.3
pointer assignment (=>) — 11.1
pointer chasing — 11.3
pointer component (vs allocatable) — 9.4
polyglot profile (Fortran + Python + HPC) — 40.4
polymorphic array (single dynamic type) — 10.2
polymorphic container — 11.5
polymorphism — 10.2
population dynamics (logistic, Lotka-Volterra) — 23.6
portability across compilers — 30.5
portable precision — 3.2
portfolio piece — 40.5, Project Checkpoint
positional association — 17.2
postcondition — 13.5
power density (thermal limit) — 31.1
precision (intrinsic) — 3.2, 20.2
precision loss — 13.6
precision vs accuracy — 20.6
precondition — 13.5
premature optimization — 28.5
premature optimization (Knuth) — 29.5, Overview
present (intrinsic) — 6.3
print '(a)' (formatted) — 2.2
print * (list-directed) — 2.2
print statement — 2.2, 7.2
private attribute — 8.2
private-by-default idiom — 8.2
problem statement (paper) — 38.2
procedure argument (integrand) — 22.2
procedure argument (RHS to a solver) — 23.1, 23.2, 23.4
procedure pointer — 11.5
process (MPI) — 34.1
product — 5.4
product rule (multidimensional) — 22.5
profile-guided optimization (PGO) — 30.4
profiling — 28.1, 28.2
profiling instrumentation (-pg) — 28.2
program / end program — 2.2
provenance — 37.5
public attribute — 8.2
publication-quality figure — 26.5
punch card (80 columns) — 17.1
pure — 6.4
pure (optimization implications) — 27.4
pure function (from statement function) — 19.3
put (remote write) — 32.2
PVD collection file (.pvd) — 26.3
Python orchestration layer — 15.1

Q

QUADPACK — 22.2, 22.3
quadratic formula (stable) — 20.3
quadrature — 22.2
quadruple precision (binary128) — 20.1, 20.6
Quantum ESPRESSO — 36.1
quick-reference table (Appendix E) — 19.5

R

race condition — 32.3
radians vs degrees — 3.5
rainbow / jet colormap (avoid) — 26.5
random seed (recording) — 37.5
range (intrinsic) — 3.2
rank — 5.1, 5.2, 34.1
rank (intrinsic) — 5.2
read statement — 7.2, 7.3
read_config (namelist) — Project Checkpoint
reading tracks — 1.6
README / LICENSE / abstract (packaging) — Project Checkpoint
real (type) — 3.1, 3.2
real do counter (deleted) — 4.3, 4.6
real equality pitfall — 4.1, 4.3
real(dp) / _dp suffix — 2.7
real64 — 3.2
reassociation (compiler) — 37.2
reassociation (floating-point) — 30.1
recl (record length) — 7.3
recompilation cascade — 8.3, 8.5
record — 7.3
record marker — 7.5
recording the flags — 30.4
RECTILINEAR_GRID / STRUCTURED_GRID / UNSTRUCTURED_GRID — 26.1
recursion — 6.5
recursive (attribute) — 6.5
reduction — 5.4
reduction (OpenACC clause) — 35.2 (case study 2)
reduction identity element — 33.3
reduction to first-order system — 23.4
refactoring — 18.1, 18.2
reference BLAS — 21.5
reference BLAS (Netlib) — 16.1
reference page (reading) — 21.4
register pressure — 29.1
regression reference (expected output) — 17.6
regression test — 18.3, 37.2
regression test (vs analytical) — 38.3
relational operator — 4.1
relational operators (.LT. vs <) — 19.4
rename on use (=>) — 8.1
repeat count (format) — 7.1
repeat count (in DATA) — 17.5
repeating binary fraction — 20.1
repetition (timing) — 28.5
reproducibility — 37.5, 38.6
reproducibility (metadata) — 25.5
reproducible build — 30.4
request handle — 34.4
research software engineer (RSE) — 40.2
reshape — 5.4
reshape (order=[2,1]) — 21.1
residency (keeping data on the device) — 35.4
residual (linear system) — 21.3
restrict (C keyword) — 1.3
restrict (C99) — 27.3
result clause — 6.1
result_image — 32.4
results and visualization — 38.5
revision cadence (five-year) — 39.2
Richardson extrapolation — 22.3
Richardson extrapolation (for integrators) — 23.3
right-hand-side function (RHS) — 23.1, 23.4
RK4 (classical fourth-order) — 23.2, 23.4, Project Checkpoint
Robertson problem (stiff benchmark) — 23.6
Romberg integration — 22.3
roofline intuition — Case Study 2
roofline model — 28.3, 29.2, 29.5, 38.4
round-off floor — 22.4
row-major vs column-major (tool display) — 25.2
Runge-Kutta method — 23.2
runtime aliasing check / loop versioning (C) — 27.3
résumé, presenting Fortran on — 40.4

S

save_every (output cadence) — 26.3, Project Checkpoint
SCALARS / LOOKUP_TABLE — 26.2
scaled speedup — 31.2
scan (intrinsic) — 12.2, 12.5
scarcity as leverage — 40.3
scheduling — 33.4
second difference (three-point) — 22.1, Project Checkpoint
second-difference (spatial) operator — 23.4, Project Checkpoint
segment (execution) — 32.3
segment ordering — 32.3
select case — 4.2
select case (replacing computed GOTO) — 19.2
select type — 10.3
selected_int_kind — 3.2
selected_real_kind — 3.2, 20.6
self time vs total time — 28.2
self-describing format — 7.5, 25.1, 25.3
separable solution (heat equation) — 37.2
separate memory spaces — 35.1, 35.4
separate module procedure — 8.3
separation of concerns (numerical) — 16.1
sequential access — 7.3
serial baseline — 38.4
serial fraction — 31.2, 31.4, 31.5
shallow copy / aliasing — 9.4
shape — 5.1, 5.2
shape (intrinsic) — 5.2, 5.4
shared memory — 31.3
short-circuit evaluation — 39.1
shuffle filter — 25.3
side effect — 6.4
sign / exponent / fraction fields — 20.1
significand (mantissa) — 20.1
silent copy (C-to-F) — 15.3
SIMD — 27.1, 27.5
SIMD (!$omp simd`, `!$omp do simd) — 33.5
SIMD (Single Instruction, Multiple Data) — 29.3
SIMD instruction set (AVX-512, native tuning) — 30.1
Simpson's rule — 22.2, 22.3
Simpson's-rule weights (in RK4) — 23.2
sin / cos / tan — 3.5
sind — 39.1, Project Checkpoint
single precision — 3.2
single precision (binary32) — 20.1, 20.6
singular value decomposition (SVD) — 21.4
size (intrinsic) — 5.1, 5.2, 5.4
slash edit descriptor (new record) — 7.1
slope field / direction field — 23.1
solver module — 36.2
source tree — 36.1, 36.4
sourced allocation (allocate source=) — 10.2
sp / dp / qp kinds — 20.6
spacing (intrinsic) — 20.2
sparse format (COO/CSR/CSC) — 21.6
sparse matrix — 21.6
speedup (10–100×, illustrative) — 15.5
speedup ceiling — 31.2
speedup ceiling (size of the prize) — 28.5
SPMD (Single Program, Multiple Data) — 32.1
SPMD (single program, multiple data) — 34.1
sqrt — 3.5
src/ app/ test/ layout — Project Checkpoint
stability (numerical) — 24.4
stable_dt (procedure) — 24.4, Project Checkpoint
standard_name / long_name — 25.5
standards process — 39.2
stat (allocation) — 13.1
statement function — 17.4
statement function (modernizing) — 18.4
statement function, translating — 19.3
statement label (columns 1–5) — 17.1
statistical sampling (profiler) — 28.2
status (open specifier) — 7.3
stdlib — 16.4
stdlib (Fortran standard library) — 39.3
stdlib_io — 16.4
stdlib_linalg (solve wrapper) — 21.3
stdlib_stats — 16.4
steady state — 24.1, 24.6, 38.5
steady state (as fixed-point test) — 37.2
steady-state hand-check (4x4 plate, 37.5/12.5) — Project Checkpoint
steady-state plate solution (37.5 / 12.5) — 18.4
stencil (1-D Laplacian) — 22.1, Project Checkpoint
stencil (procedure, optimized) — 29.1, 29.3, Project Checkpoint
step (procedure) — 24.3, Project Checkpoint
step doubling — 23.3
step size — 23.1, 23.3
step-size controller / safety factor — 23.3
stiffness — 23.5
stop code — 13.2
stop vs error stop — 13.2
storage_size (intrinsic) — 25.1
stream I/O — 7.5
stride — 5.2, 27.2
strong scaling — 31.4, 38.4
struct padding / alignment — 14.5
structure constructor — 9.1, 9.4
structure constructor for an extended type — 10.2
structured grid — 24.6
structured grid output — 26.1
STRUCTURED_POINTS dataset — 26.1, 26.2
submodule — 8.3
subnormal (denormal) — 20.4
subroutine — 6.1
subscript triplet — 5.2
substring — 12.3
sum — 5.4, 5.7
surface-to-volume ratio — 34.4
symmetry (as test) — 37.1
sync all — 32.3
sync images — 32.3
system of ODEs (first-order) — 23.4
system_clock (intrinsic) — 28.1

T

tag (message) — 34.2
target (attribute) — 11.1
task parallelism — 31.4
Taylor series (for error analysis) — 22.1
team — 32.4
team_type — 32.4
temporal blocking — 29.2
ternary operator (see conditional expression) — 39.1
test oracle — 37.1
test-drive — 16.5
text diff pitfall (compare numbers not text) — 18.3
text vs binary I/O (at scale) — 25.1
the 80/20 rule (Pareto principle) — 28.5
the no-aliasing advantage — 27.3
thermal diffusivity (alpha) — 24.1
this_image (cosubscripts form) — 32.2
this_image (intrinsic) — 32.1, 32.2
thread (team) — 33.1
threadIdx / blockIdx / blockDim — 35.3
threads per block — 35.3
throughput vs latency — 35.1
time-loop dependency — 31.5
time-series output — 26.3
time-stepping loop (heat solver) — Project Checkpoint
timers module (tic/toc) — 28.1, Project Checkpoint
tiny (intrinsic) — 20.2
token / delimiter — 12.5
tokenizer — 12.5
tolerance (floating-point comparison) — 37.1, 37.2
TOML — 16.3
TOP500 — 1.2
topological sort of dependencies — 8.5, 8.6
trailing-blank trap — 12.3, 12.6
transfer (intrinsic) — 17.3
transfer (intrinsic, replacing EQUIVALENCE) — 19.1
transfer intrinsic — 18.1
translation dictionary — 19.1, 19.5
transpose — 5.4, 21.1
trapezoidal rule — 22.2, 22.5
trim — 12.2, 12.3
truncation error — 22.1, 22.4
truncation toward zero — 3.4
tuned BLAS vs hand-rolled loop — 27.4
two-buffer update (parallel safety) — Project Checkpoint
two-language workflow — 15.1
two-pass variance — case study 2
type extension (inheritance) — 10.1
type is (guard) — 10.3
type promotion — 3.3
type vs class — 10.2
type-bound procedure — 9.2
typed allocation (allocate type ::) — 10.6
typeof — 39.1

U

ULP (unit in the last place) — 20.2
undefined behavior (aliasing violation) — 27.3, Case Study 1
undefined pointer — 11.2
undefined reference (link error) — 21.5
undefined reference (linker error) — 2.3
underflow — 20.4
unformatted I/O — 7.5
uninitialized variable — 13.6
unit — 7.3
unit roundoff — 20.2
unit stride — 29.1, 29.3, 29.4
unit test — 37.1
units (UDUNITS) — 25.5
unlimited dimension (time series) — 25.4 (exercises)
unlimited format repeat — 7.1, 7.4
unlimited-repeat format (*) — 26.4
unstructured grid — 24.6
unstructured grid (note) — 26.1
update self / update host — 35.4
use mpi vs mpi_f08 — 34.5
use statement — 8.1
use, only clause — 8.1, 8.5
utility module — 36.2

V

valgrind — 13.4
validate_config (heat solver) — 13 Project Checkpoint
validation — 38.3
validation (vs verification) — 37.2
value assignment through a pointer — 11.1
value attribute — 14.3, 14.4
value semantics — 9.4
variance / spread (timing) — 28.5
vector triad (STREAM) — Case Study 2
vector width (32-byte vectors, AVX) — 27.5
vectorization (intro) — 27.1, 27.2, 27.4, 27.5
vectorization report (-fopt-info-vec) — 29.3, 29.4
verification — 37.2, 38.3
verification and validation (V&V) — 38.3
verify (intrinsic) — 12.2, 12.5
version control (git, for science) — 37.4
version control (in migration) — 18.2
VisIt — 26.3
von Neumann stability analysis — 24.4
VTI (.vti XML ImageData) — 26.2
VTK (Visualization Toolkit) — 26.1, 26.2
VTK / ParaView (time series) — 38.5
VTK file format (legacy) — 26.1, 26.2
VTK magic string (# vtk DataFile Version 3.0) — 26.2
VTK XML format — 26.1, 26.2

W

wall-clock time — 28.1
warm-up — 28.5
warp — 35.1, 35.5
warp divergence — 35.5
wave equation — 24.1, 24.4
weak scaling — 31.4, 38.4
weather and climate services (NOAA, ECMWF, Met Office) — 40.1
weight (quadrature) — 22.2, 22.3
well-conditioned / ill-conditioned — 20.5
WG5 — 39.2
what reviewers check — 38.6
when GPUs help / when they don't — 35.5
when OOP helps / hurts — 10.5
where — 4.5, 5.7
where state lives — 36.5
where the time goes — 36.5
whole-array operation — 1.3, 5.3
WholeExtent (0-based) — 26.2
with-the-grain vs against-the-grain traversal — 27.2
work sharing — 33.2
workflow (CI, .yml) — 37.3
works-on-my-machine problem — 37.5
workspace query (lwork = -1) — 21.4
WRF (weather model) — 1.2, 36.1
write statement — 7.2, 7.3
write_field — Project Checkpoint
write_field_netcdf (project) — Project Checkpoint
write_vti — 26.2
write_vtk (canonical signature) — 26.2, Project Checkpoint
WSL (Windows) — 2.1

X

X edit descriptor (spaces) — 7.1