changelog

来自mit的fdtd开放性源代码meep

Thu Jul 24 16:27:52 EDT 2008  [e-mail]
  tagged meep-0.20.3


Thu Jul 24 16:25:03 EDT 2008  [e-mail]
  * removed circular ctl-io.h dependency, bumped version to 0.20.3

    M ./NEWS +5
    M ./configure.ac -2 +2
    M ./libctl/Makefile.am -1 +1

Mon Jul 21 19:40:53 EDT 2008  [e-mail]
  tagged meep-0.20.2


Mon Jul 21 19:40:03 EDT 2008  [e-mail]
  * fix to work with Guile 1.6.x

    M ./NEWS +5
    M ./configure.ac -2 +2
    M ./libctl/meep.scm.in -1 +1

Sun Jul 20 14:35:23 EDT 2008  [e-mail]
  tagged meep-0.20.1


Sun Jul 20 14:35:13 EDT 2008  [e-mail]
  * fix libctl dependencies so that parallel builds work

    M ./NEWS +2
    M ./libctl/Makefile.am -5 +9

Sun Jul 20 13:51:21 EDT 2008  [e-mail]
  * install correct pkgconfig file (meep_mpi.pc) for MPI version; thanks to Majid Sodagar for the bug report; bump version to 0.20.1

     ./meep.pc.in -> ./meep-pkgconfig.in
    M ./Makefile.am -2 +5
    M ./NEWS +7
    M ./configure.ac -3 +3
    M ./meep-pkgconfig.in -1 +1

Sun Jul 20 12:10:59 EDT 2008  [e-mail]
  * more robust handling of nested synchronizations via count; also, allow (expensive) synchronized timestepping if user timesteps inside a synchronized-field state

    M ./src/energy_and_flux.cpp -4 +6
    M ./src/fields.cpp -2 +1
    M ./src/meep.hpp -1 +1
    M ./src/step.cpp -1 +14

Sat Jul 19 20:02:00 EDT 2008  [e-mail]
  * typo

    M ./NEWS -2 +2

Sat Jul 19 19:56:20 EDT 2008  [e-mail]
  tagged meep-0.20


Sat Jul 19 19:53:58 EDT 2008  [e-mail]
  * added perfect-magnetic/electric-conductor predefined materials; epsilon now defaults to 1 for the medium

    M ./NEWS +6
    M ./libctl/meep.scm.in -1 +3

Sat Jul 19 13:28:06 EDT 2008  [e-mail]
  * pml test with length is now a little more stringent (checks that dependence is really 1/L^6, and not better)

    M ./tests/pml.cpp -27 +11

Sat Jul 19 12:39:34 EDT 2008  [e-mail]
  * make sync/restore magnetic fields idempotent, as synchronizations might be nested (especially in libctl UI)

    M ./src/energy_and_flux.cpp +4
    M ./src/fields.cpp +3
    M ./src/meep.hpp +1
    M ./src/step.cpp +2

Fri Jul 18 19:58:06 EDT 2008  [e-mail]
  * clarification

    M ./NEWS -2 +3

Fri Jul 18 19:49:47 EDT 2008  [e-mail]
  * copyright year update to 2008

    M ./COPYRIGHT -1 +1
    M ./NEWS -1 +1
    M ./libctl/meep.scm.in -1 +1
    M ./src/bands.cpp -1 +1
    M ./src/bicgstab.cpp -1 +1
    M ./src/bicgstab.hpp -1 +1
    M ./src/boundaries.cpp -1 +1
    M ./src/control_c.cpp -1 +1
    M ./src/cw_fields.cpp -1 +1
    M ./src/dft.cpp -1 +1
    M ./src/energy_and_flux.cpp -1 +1
    M ./src/fields.cpp -1 +1
    M ./src/grace.cpp -1 +1
    M ./src/h5fields.cpp -1 +1
    M ./src/h5file.cpp -1 +1
    M ./src/initialize.cpp -1 +1
    M ./src/integrate.cpp -1 +1
    M ./src/loop_in_chunks.cpp -1 +1
    M ./src/meep.hpp -1 +1
    M ./src/meep/mympi.hpp -1 +1
    M ./src/meep/vec.hpp -1 +1
    M ./src/meep_internals.hpp -1 +1
    M ./src/monitor.cpp -1 +1
    M ./src/mympi.cpp -1 +1
    M ./src/output_directory.cpp -1 +1
    M ./src/polarization.cpp -1 +1
    M ./src/sources.cpp -1 +1
    M ./src/sphere-quad.cpp -1 +1
    M ./src/step.cpp -1 +1
    M ./src/step_db.cpp -1 +1
    M ./src/structure.cpp -1 +1
    M ./src/time.cpp -1 +1
    M ./src/update_e_from_d.cpp -1 +1
    M ./src/update_from_e.cpp -1 +1
    M ./src/update_h_from_b.cpp -1 +1
    M ./src/vec.cpp -1 +1
    M ./tests/bench.cpp -1 +1
    M ./tests/bragg_transmission.cpp -1 +1
    M ./tests/cylindrical.cpp -1 +1
    M ./tests/flux.cpp -1 +1
    M ./tests/harmonics.cpp -1 +1
    M ./tests/integrate.cpp -1 +1
    M ./tests/known_results.cpp -1 +1
    M ./tests/one_dimensional.cpp -1 +1
    M ./tests/physical.cpp -1 +1
    M ./tests/symmetry.cpp -1 +1
    M ./tests/three_d.cpp -1 +1
    M ./tests/two_dimensional.cpp -1 +1

Fri Jul 18 19:46:30 EDT 2008  [e-mail]
  * missing string.h header (for gcc 4.3; see debian bug #455289)

    M ./src/step_db.cpp +1
    M ./src/update_e_from_d.cpp +2
    M ./src/update_h_from_b.cpp +17

Fri Jul 18 18:58:53 EDT 2008  [e-mail]
  * fix PML test to test whether reflection converges to zero with increasing resolution

    M ./tests/pml.cpp -201 +127

Fri Jul 18 18:18:23 EDT 2008  [e-mail]
  * honor quiet flag in structure.cpp

    M ./src/structure.cpp -3 +4

Fri Jul 18 17:17:40 EDT 2008  [e-mail]
  * updated archflag guesses from fftw

    M ./m4/ax_gcc_archflag.m4 -24 +31

Fri Jul 18 17:11:06 EDT 2008  [e-mail]
  * 0.20 release date

    M ./NEWS -1 +1

Fri Jul 18 17:09:24 EDT 2008  [e-mail]
  * updated TODO

    M ./TODO -8 +2

Fri Jul 18 16:58:59 EDT 2008  [e-mail]
  * libctl 3.0.3 is required, to support boolean object properties

    M ./configure.ac -1 +1

Fri Jul 18 16:34:14 EDT 2008  [e-mail]
  * check for libctl in /usr/share/libctl3

    M ./configure.ac +3

Fri Jul 18 14:00:57 EDT 2008  [e-mail]
  * Hx source seems to couple better to modes than Ez src, gives more reliable results for ring example

    M ./examples/ring.ctl -1 +1

Fri Jul 18 13:46:21 EDT 2008  [e-mail]
  * make is-integrated?=false the default for sources

    M ./NEWS -1 +2
    M ./libctl/meep.scm.in -1 +1

Fri Jul 18 13:12:17 EDT 2008  [e-mail]
  * made 1d symmetry tests a little bigger so that the tests are big enough to run on 8 processors

    M ./tests/symmetry.cpp -2 +2

Fri Jul 18 12:54:01 EDT 2008  [e-mail]
  * tolerance was a little too low in cylindrical.cpp for new code (where the roundoff error depends on the splitting)

    M ./tests/cylindrical.cpp -2 +2

Fri Jul 18 12:52:20 EDT 2008  [e-mail]
  * bug fix in and_to_all/or_to_all - was passing wrong type (bool instead of int); also eliminate a couple of unnecessary memcpy's

    M ./src/mympi.cpp -7 +14

Wed Jul 16 18:51:41 EDT 2008  [e-mail]
  * slight code simplification

    M ./src/structure.cpp -2 +4

Wed Jul 16 18:40:52 EDT 2008  [e-mail]
  * typo in comment

    M ./src/boundaries.cpp -1 +1

Wed Jul 16 18:29:56 EDT 2008  [e-mail]
  * added synchronized-magnetic step func

    M ./NEWS +4
    M ./libctl/meep.scm.in +8

Tue Jul 15 22:51:43 EDT 2008  [e-mail]
  * version bump

    M ./NEWS +32
    M ./configure.ac -2 +2

Tue Jul 15 22:04:35 EDT 2008  [e-mail]
  * use master_printf rather than printf in tests

    M ./tests/cylindrical.cpp -11 +11
    M ./tests/flux.cpp -4 +4
    M ./tests/h5test.cpp -1 +1
    M ./tests/physical.cpp -2 +2

Tue Jul 15 22:03:55 EDT 2008  [e-mail]
  * fix deadlock bug in chunk connection when H is allocated on some processes but not on others

    M ./src/step.cpp -1
    M ./src/update_h_from_b.cpp +4

Mon Jun 30 17:08:02 EDT 2008  [e-mail]
  * added PML test

    M ./tests/Makefile.am -1 +1
    A ./tests/pml.cpp

Mon Jun 30 18:33:33 EDT 2008  [e-mail]
  * allow user-specified PML profile function and asymptotic (round-trip) reflection

    M ./libctl/meep.scm.in -1 +3
    M ./libctl/structure.cpp -23 +59
    M ./src/meep.hpp -8 +19
    M ./src/structure.cpp -13 +17

Mon Jun 23 23:09:49 EDT 2008  [e-mail]
  * added output-bfield etc.

    M ./libctl/meep.scm.in +1

Mon Jun 23 19:19:37 EDT 2008  [e-mail]
  * slight optimization

    M ./src/step_generic.cpp -3 +3

Mon Jun 23 19:02:01 EDT 2008  [e-mail]
  * use some clever/ugly macros to greatly speed up ksig[] index calculation in the PML loops (makes a significant difference in small benchmarks)

    M ./src/step_generic.cpp -50 +34

Mon Jun 23 18:20:19 EDT 2008  [e-mail]
  * a few small optimizations (factoring out siginv[k],  special-casing the most common loop) in update_EDHB

    M ./src/step_generic.cpp -13 +32

Mon Jun 23 17:36:03 EDT 2008  [e-mail]
  * tried using "restrict" keyword to speed things up, but annoying gcc actually slows things down!

    M ./configure.ac +7
    M ./src/step_generic.cpp -22 +27

Mon Jun 23 17:11:40 EDT 2008  [e-mail]
  * fixed a typo in the PML offset calculation; this bug apparently doesn't affect the results, but I'm not sure why

    M ./src/step_generic.cpp -3 +3

Mon Jun 23 16:06:45 EDT 2008  [e-mail]
  * optimization - delete offdiagonal inveps when == 0

    M ./src/anisotropic_averaging.cpp -3 +18

Mon Jun 23 15:22:00 EDT 2008  [e-mail]
  * fix some floating-point thresholds

    M ./tests/cylindrical.cpp -2 +2

Mon Jun 23 14:53:08 EDT 2008  [e-mail]
  * fixed solve_cw, re-enable solve_cw in physical.cpp

    M ./src/cw_fields.cpp -25 +80
    M ./src/meep.hpp -4 +4
    M ./src/step.cpp -24
    M ./src/update_e_from_d.cpp -3 +9
    M ./tests/physical.cpp -22 +18

Sat Jun 21 18:30:29 EDT 2008  [e-mail]
  * fix bug in figuring out curl components in 2d, introduced by cylindrical patch; bypass cw_solve in physical.cpp for now ... make check now passes again!!

    M ./src/fields.cpp -1 +2
    M ./tests/physical.cpp +10

Sat Jun 21 01:56:19 EDT 2008  [e-mail]
  * verified that changes in known_results were due to changes in PML, and that the new results converge to the old ones as the PML thickens; updated numbers so that known_results passes again

    M ./tests/known_results.cpp -8 +9

Sat Jun 21 01:21:55 EDT 2008  [e-mail]
  * new value for polariton_energy known_result - difference is due to new scheme for synchronizing magnetic fields (I verified that switching back to the old synchronization gives the old result again, so it's not the new timestepping)

    M ./tests/known_results.cpp -1 +1

Fri Jun 20 21:29:24 EDT 2008  [e-mail]
  * I'm convinced that the old Meep had a bug in its update code for Dz at r=0 for m=0, where it was off by a factor of 4 (this effectively multipled epsilon by 4 for one pixel at r=0 for m==0, which causesd only small errors in the results, and still converged, so it went unnoticed)

    M ./src/step_db.cpp -1 +1

Fri Jun 20 21:19:49 EDT 2008  [e-mail]
  * lower tolerance (from zero) in cylindrical.cpp -- because of the new way the (1/r) d(r*f)/dr derivative is computed via a cumsum, the roundoff error depends on the splitting in the r direction

    M ./tests/cylindrical.cpp -2 +2

Fri Jun 20 18:50:53 EDT 2008  [e-mail]
  * fixed some bugs in step_db for Dcyl (wrong radius used at one point, and also treat negative m properly...the latter problem being a bug in the previous Meep too)

    M ./src/step_db.cpp -7 +8

Fri Jun 20 18:49:34 EDT 2008  [e-mail]
  * use new LOOP_OVER_VOL_OWNED0 in step_curl, to avoid the r=0 origin which gets special treatment in Dcyl

    M ./src/meep/vec.hpp +10
    M ./src/step_generic.cpp -8 +8
    M ./src/vec.cpp -1 +1

Fri Jun 20 15:42:08 EDT 2008  [e-mail]
  * some fixes; m=0 case now matches old code, and also mmatches if we manually multiply m terms by zero in both programs, so eremaining bug is in im/r terms

    M ./src/step_db.cpp -20 +32

Fri Jun 20 03:27:38 EDT 2008  [e-mail]
  * initial stab at re-implementing cylindrical support; compiles, and flux.cpp passes, but cylindrical.cpp still fails

    M ./src/anisotropic_averaging.cpp -4 +4
    M ./src/fields.cpp -3 +13
    M ./src/meep.hpp +3
    M ./src/meep/vec.hpp +9
    M ./src/step_db.cpp -26 +159
    M ./src/step_generic.cpp -1 +1
    M ./src/structure.cpp -1 +1

Thu Jun 19 20:54:57 EDT 2008  [e-mail]
  * unified step_d and step_b into step_db; also, changed d_minus_p to more-generic f_minus_p (both for step_db and in preparation for more unification of magnetic and electric code)

     ./src/step_d.cpp -> ./src/step_db.cpp
    M ./src/Makefile.am -1 +1
    M ./src/energy_and_flux.cpp -4 +6 r1
    M ./src/fields.cpp -9 +7 r1
    M ./src/meep.hpp -5 +2 r1
    M ./src/meep/vec.hpp -1 +1
    M ./src/step.cpp -2 +2
    R ./src/step_b.cpp
    M ./src/step_db.cpp -13 +22 r1
    M ./src/update_e_from_d.cpp -12 +16 r1

Thu Jun 19 20:54:08 EDT 2008  [e-mail]
  * silence compiler warning

    M ./tests/symmetry.cpp -2 +2

Thu Jun 19 20:34:04 EDT 2008  [e-mail]
  * when copying fields_chunks, be careful not to duplicate H==B fields

    M ./src/fields.cpp -5 +11

Thu Jun 19 20:17:01 EDT 2008  [e-mail]
  * greatly simplified step_d and step_h by allowing step_curl to swap its arguments if g1 is NULL

    M ./src/step_b.cpp -27 +3
    M ./src/step_d.cpp -27 +4
    M ./src/step_generic.cpp -4 +9

Thu Jun 19 20:15:51 EDT 2008  [e-mail]
  * added FOR_FT_COMPONENTS in preparation for more generic stepping functions

    M ./src/meep/vec.hpp +3
    M ./src/vec.cpp +10

Thu Jun 19 19:39:28 EDT 2008  [e-mail]
  * bug fix - B is a pseudovector too (not just H), and use B*H for H_EnergyDensity derived component

    M ./src/vec.cpp -2 +2

Thu Jun 19 18:43:46 EDT 2008  [e-mail]
  * naming consistency (H/E sources vs. B/D sources); also, bug fix in "D" sources, which were supposed to be multiplied by eps but appparently weren't (& implemented similar "B" sources multiplied by mu)

    M ./src/cw_fields.cpp r1