Recent changes to Home

Home modified by Bits of cotton

Bits of cotton — Sun, 31 Dec 2017 08:37:40 -0000

--- v89
+++ v90
@@ -9,7 +9,7 @@
 Freezed.

 * More user-friendly user interface or software needed.
-* For latest information, please visit https//github.com/bitsofcotton/conv_check .
+* For latest information, please visit https://github.com/bitsofcotton/conv_check .
 * Web demo is @ https://services.limpid-intensity.info/
 * Konbu Check is BSD or LGPL licensed.

Home modified by Bits of cotton

Bits of cotton — Sat, 28 Oct 2017 13:29:51 -0000

--- v88
+++ v89
@@ -8,10 +8,9 @@

 Freezed.

-For latest information, please visit https//github.com/bitsofcotton/conv_check .
-
 * More user-friendly user interface or software needed.
-* Web demo is @ konbu.sakura.ne.jp .
+* For latest information, please visit https//github.com/bitsofcotton/conv_check .
+* Web demo is @ https://services.limpid-intensity.info/
 * Konbu Check is BSD or LGPL licensed.

 ## TOC: ##

Home modified by Bits of cotton

Bits of cotton — Mon, 31 Jul 2017 05:47:55 -0000

--- v87
+++ v88
@@ -5,6 +5,8 @@
 You may need this program for collision detecting of linear inequality pairs.
 And this program is for dense and small dimension problems, to solve sparse or large dimension problems, no advantages from proceeding algorithms (so slow or too large accuracy needed.).
 This program gains inner vector in O(3mn^2) arithmetic and O(mn) elementary function (cos, sqrt and so on) operation time for m constraints n variables.
+
+Freezed.

 For latest information, please visit https//github.com/bitsofcotton/conv_check .

Home modified by Bits of cotton

Bits of cotton — Mon, 31 Jul 2017 05:47:38 -0000

--- v86
+++ v87
@@ -4,7 +4,7 @@
 Konbu Check aims to check feasibility and get feasible point from a multiple set of linear constraints.
 You may need this program for collision detecting of linear inequality pairs.
 And this program is for dense and small dimension problems, to solve sparse or large dimension problems, no advantages from proceeding algorithms (so slow or too large accuracy needed.).
-This program gains inner vector in O(3mn^2) arithmetic and elementary function (cos, sqrt and so on) operation time for m constraints n variables.
+This program gains inner vector in O(3mn^2) arithmetic and O(mn) elementary function (cos, sqrt and so on) operation time for m constraints n variables.

 For latest information, please visit https//github.com/bitsofcotton/conv_check .

@@ -20,9 +20,10 @@
 * [Proofs]

 ## Tips ##
-* In steps function, we have a chance that getting huge t in the bdash \* t large enough that accuracy not catches. This depends on ratio of diameters on feasible region of transformed problem (In another word, some type of difficulty of the problem). Unlike another algorithms based on newton's method, this program gains meaning-less vectors with hard problems now. This glitch can be fixed with pre-calculating largest diameters on original problem, and if so, this program also gains nearest vector in which problem is hard enough.
+* It has a chance that we slips feasible point, if then, please extend accuracy or the parameter threshold_loop.

 ## Bug ##
+* **Version 1.01-stable has a bug. please use 1.01-stable2 or 1.00-stable.**
 * *Version 1.00-stable and before have a bug that orthogonality slighty not saves. This is fixed in the Version 1.01.* -> re calculated, it's ok. Because we fix it each loop, so returned result may have fixed calculated vector.
 * **Version 1.00-alpha and before have a bug that O(mn^2L) operations needed. This is fixed in Version 1.00-rc.**
 * **Version 0.11 also have a bug that can't solve large diameters.**

Home modified by Bits of cotton

Bits of cotton — Sat, 29 Jul 2017 13:19:33 -0000

--- v85
+++ v86
@@ -23,7 +23,7 @@
 * In steps function, we have a chance that getting huge t in the bdash \* t large enough that accuracy not catches. This depends on ratio of diameters on feasible region of transformed problem (In another word, some type of difficulty of the problem). Unlike another algorithms based on newton's method, this program gains meaning-less vectors with hard problems now. This glitch can be fixed with pre-calculating largest diameters on original problem, and if so, this program also gains nearest vector in which problem is hard enough.

 ## Bug ##
-* **Version 1.00-stable and before have a bug that orthogonality slighty not saves. This is fixed in the Version 1.01.**
+* *Version 1.00-stable and before have a bug that orthogonality slighty not saves. This is fixed in the Version 1.01.* -> re calculated, it's ok. Because we fix it each loop, so returned result may have fixed calculated vector.
 * **Version 1.00-alpha and before have a bug that O(mn^2L) operations needed. This is fixed in Version 1.00-rc.**
 * **Version 0.11 also have a bug that can't solve large diameters.**
 * **Version 0.10-release3 have a little configured parameter, so this is good for gaining inner points. But in optimizing, edge misses often occures.**

Home modified by Bits of cotton

Bits of cotton — Mon, 24 Jul 2017 02:34:43 -0000

--- v84
+++ v85
@@ -5,6 +5,8 @@
 You may need this program for collision detecting of linear inequality pairs.
 And this program is for dense and small dimension problems, to solve sparse or large dimension problems, no advantages from proceeding algorithms (so slow or too large accuracy needed.).
 This program gains inner vector in O(3mn^2) arithmetic and elementary function (cos, sqrt and so on) operation time for m constraints n variables.
+
+For latest information, please visit https//github.com/bitsofcotton/conv_check .

 * More user-friendly user interface or software needed.
 * Web demo is @ konbu.sakura.ne.jp .
@@ -21,6 +23,7 @@
 * In steps function, we have a chance that getting huge t in the bdash \* t large enough that accuracy not catches. This depends on ratio of diameters on feasible region of transformed problem (In another word, some type of difficulty of the problem). Unlike another algorithms based on newton's method, this program gains meaning-less vectors with hard problems now. This glitch can be fixed with pre-calculating largest diameters on original problem, and if so, this program also gains nearest vector in which problem is hard enough.

 ## Bug ##
+* **Version 1.00-stable and before have a bug that orthogonality slighty not saves. This is fixed in the Version 1.01.**
 * **Version 1.00-alpha and before have a bug that O(mn^2L) operations needed. This is fixed in Version 1.00-rc.**
 * **Version 0.11 also have a bug that can't solve large diameters.**
 * **Version 0.10-release3 have a little configured parameter, so this is good for gaining inner points. But in optimizing, edge misses often occures.**

Home modified by Bits of cotton

Bits of cotton — Sat, 15 Jul 2017 09:38:20 -0000

--- v83
+++ v84
@@ -4,7 +4,7 @@
 Konbu Check aims to check feasibility and get feasible point from a multiple set of linear constraints.
 You may need this program for collision detecting of linear inequality pairs.
 And this program is for dense and small dimension problems, to solve sparse or large dimension problems, no advantages from proceeding algorithms (so slow or too large accuracy needed.).
-This program gains inner vector in O(3mn^2) arithmetic operation time for m constraints n variables, but for low-accuracy stability reason, O(3mn^2\*lg L) now.
+This program gains inner vector in O(3mn^2) arithmetic and elementary function (cos, sqrt and so on) operation time for m constraints n variables.

 * More user-friendly user interface or software needed.
 * Web demo is @ konbu.sakura.ne.jp .
@@ -21,6 +21,7 @@
 * In steps function, we have a chance that getting huge t in the bdash \* t large enough that accuracy not catches. This depends on ratio of diameters on feasible region of transformed problem (In another word, some type of difficulty of the problem). Unlike another algorithms based on newton's method, this program gains meaning-less vectors with hard problems now. This glitch can be fixed with pre-calculating largest diameters on original problem, and if so, this program also gains nearest vector in which problem is hard enough.

 ## Bug ##
+* **Version 1.00-alpha and before have a bug that O(mn^2L) operations needed. This is fixed in Version 1.00-rc.**
 * **Version 0.11 also have a bug that can't solve large diameters.**
 * **Version 0.10-release3 have a little configured parameter, so this is good for gaining inner points. But in optimizing, edge misses often occures.**
 * **Veresion 0.10-release2 gains edges properly but not gains good vector. please wait next release (sorry but it will be late)**

Home modified by Bits of cotton

Bits of cotton — Sat, 07 May 2016 07:41:36 -0000

--- v82
+++ v83
@@ -21,6 +21,7 @@
 * In steps function, we have a chance that getting huge t in the bdash \* t large enough that accuracy not catches. This depends on ratio of diameters on feasible region of transformed problem (In another word, some type of difficulty of the problem). Unlike another algorithms based on newton's method, this program gains meaning-less vectors with hard problems now. This glitch can be fixed with pre-calculating largest diameters on original problem, and if so, this program also gains nearest vector in which problem is hard enough.

 ## Bug ##
+* **Version 0.11 also have a bug that can't solve large diameters.**
 * **Version 0.10-release3 have a little configured parameter, so this is good for gaining inner points. But in optimizing, edge misses often occures.**
 * **Veresion 0.10-release2 gains edges properly but not gains good vector. please wait next release (sorry but it will be late)**
 * **Please use 0.08-stable5 or later to gain proper vectors in proper time.**

Home modified by Bits of cotton

Bits of cotton — Sat, 07 May 2016 06:18:17 -0000

--- v81
+++ v82
@@ -18,7 +18,7 @@
 * [Proofs]

 ## Tips ##
-* In steps function, we have a chance that getting huge t in the bdash \* t large enough that accuracy not catches. This depends on ratio of diameters on feasible region of transformed problem (In another word, some type of difficulty of the problem). Unlike another algorithms based on newton's method, this program gains meaning-less vectors with hard problems now. This glitch can fixed with pre-calculating largest diameters on original problem, and if so, this program also gains nearest vector in which problem is hard enough.
+* In steps function, we have a chance that getting huge t in the bdash \* t large enough that accuracy not catches. This depends on ratio of diameters on feasible region of transformed problem (In another word, some type of difficulty of the problem). Unlike another algorithms based on newton's method, this program gains meaning-less vectors with hard problems now. This glitch can be fixed with pre-calculating largest diameters on original problem, and if so, this program also gains nearest vector in which problem is hard enough.

 ## Bug ##
 * **Version 0.10-release3 have a little configured parameter, so this is good for gaining inner points. But in optimizing, edge misses often occures.**

Home modified by Bits of cotton

Bits of cotton — Sat, 07 May 2016 06:16:07 -0000

--- v80
+++ v81
@@ -18,7 +18,7 @@
 * [Proofs]

 ## Tips ##
-* In steps function, we have a chance that getting huge t in the bdash \* t large enough that accuracy not catches. This depends on ratio of diameters on feasible region of transformed problem.
+* In steps function, we have a chance that getting huge t in the bdash \* t large enough that accuracy not catches. This depends on ratio of diameters on feasible region of transformed problem (In another word, some type of difficulty of the problem). Unlike another algorithms based on newton's method, this program gains meaning-less vectors with hard problems now. This glitch can fixed with pre-calculating largest diameters on original problem, and if so, this program also gains nearest vector in which problem is hard enough.

 ## Bug ##
 * **Version 0.10-release3 have a little configured parameter, so this is good for gaining inner points. But in optimizing, edge misses often occures.**