NIFTy merge requestshttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests2017-11-28T17:31:20Zhttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/205fix critical filter2017-11-28T17:31:20ZJakob Knollmuellerfix critical filterpower operator needs exp(t0), not ps0power operator needs exp(t0), not ps0https://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/204Add PowerProjection operator; rewrite CriticalPowerEnergy2017-10-18T13:08:28ZMartin ReineckeAdd PowerProjection operator; rewrite CriticalPowerEnergyhttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/203Demo tweaks2017-10-06T00:28:42ZMartin ReineckeDemo tweaksThis changes the default distribution strategy in all demos to "not" instead of "fftw", which should make them runnable on most people's computers.
Also some math calls are now explicitly calling numpy instead of the ift-specific versio...This changes the default distribution strategy in all demos to "not" instead of "fftw", which should make them runnable on most people's computers.
Also some math calls are now explicitly calling numpy instead of the ift-specific versions, in preparation for future changes.
log_normal_wiener_filter now works with the new minimizers.https://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/202tweak DomainObject comparisons2017-10-06T00:28:17ZMartin Reinecketweak DomainObject comparisonsThis change should make comparisons between DomainObjects (which happen quite often) much faster.This change should make comparisons between DomainObjects (which happen quite often) much faster.https://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/201Minor changes2017-09-28T01:44:31ZTheo SteiningerMinor changeshttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/200Fixed Field.dim2017-09-28T01:14:54ZTheo SteiningerFixed Field.dimhttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/199Nightly2017-09-26T01:41:07ZTheo SteiningerNightlyhttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/198Gaussian random2017-09-26T00:51:21ZPumpe, Daniel (dpumpe)Gaussian randomAdded a further test for drawning gaussian random fields over multiple domains according to our discussion yesterday. Docstrings of my commits a not correct, I've only added one small test in test_field.pyAdded a further test for drawning gaussian random fields over multiple domains according to our discussion yesterday. Docstrings of my commits a not correct, I've only added one small test in test_field.pyhttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/197Get rid of the "bare" keyword2017-09-26T00:55:10ZMartin ReineckeGet rid of the "bare" keywordThis removes the last remnants of the "bare" key word.
Yes, this means adding an extra line in some of the demos, and it may also result in very slight performance degradation of some vdot() calls, but the gained clarity should be worth...This removes the last remnants of the "bare" key word.
Yes, this means adding an extra line in some of the demos, and it may also result in very slight performance degradation of some vdot() calls, but the gained clarity should be worth it.
(Also I think that the performance regressions can be addressed fairly easily in the future.)https://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/196Nightly2017-09-15T22:49:57ZTheo SteiningerNightlyhttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/195Enforce unique specifications for PowerSpace2017-09-15T22:52:29ZMartin ReineckeEnforce unique specifications for PowerSpaceSee issue #185See issue #185https://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/194Byebye zerocenter2017-09-28T01:59:36ZMartin ReineckeByebye zerocenterhttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/193Tweak limited exp2017-08-23T19:33:28ZMartin ReineckeTweak limited expThis merge request avoids overflows during the computation of the limited exponential function, and adds `limited_exp_deriv()`, which returns a derivative consistent to `limited_exp()`. Performance does not change noticeably.This merge request avoids overflows during the computation of the limited exponential function, and adds `limited_exp_deriv()`, which returns a derivative consistent to `limited_exp()`. Performance does not change noticeably.https://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/192Relative imports2017-08-22T21:54:48ZMartin ReineckeRelative importsThis changes every Nifty-internal import statement (as far as I could find them...) to a relative one.
I'm doing this because I'm not sure whether an absolute import statement would potentially prefer an "official" installed file over a...This changes every Nifty-internal import statement (as far as I could find them...) to a relative one.
I'm doing this because I'm not sure whether an absolute import statement would potentially prefer an "official" installed file over a local checkout, for example.https://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/191Better minimizers2017-09-28T01:18:16ZTheo SteiningerBetter minimizershttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/190Several fixes for SmoothingOperator2017-09-15T23:21:40ZMartin ReineckeSeveral fixes for SmoothingOperatorhttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/189Laplace fixes2017-08-17T23:21:36ZMartin ReineckeLaplace fixesThis should fix the bug mentioned in issue #175 and also improve performance. Some tests have been added as well.This should fix the bug mentioned in issue #175 and also improve performance. Some tests have been added as well.https://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/188normalize descent direction for SteepestDescent and VL_BFGS (but not for Rela...2017-08-21T22:06:07ZMartin Reineckenormalize descent direction for SteepestDescent and VL_BFGS (but not for RelaxedNewton!)Following a discussion with Jakob, it appears that the length of the descent direction vector only has real meaning for RelaxedNewton. The other two minimizers can get into trouble (e.g. by going to extreme coordinates and causing overfl...Following a discussion with Jakob, it appears that the length of the descent direction vector only has real meaning for RelaxedNewton. The other two minimizers can get into trouble (e.g. by going to extreme coordinates and causing overflows) when suplied with very long descent vectors, so I'm normalizing them. A good step length will hopefully be found by the line search algorithm.https://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/187Real power synthesize2017-08-17T23:15:28ZTheo SteiningerReal power synthesizehttps://gitlab.mpcdf.mpg.de/ift/nifty/-/merge_requests/186Master2017-08-15T07:56:25ZTheo SteiningerMaster