|
Hi, Member 4194593,
Thanks for your response !
Yes, the stipulation of a maximum three line segments (four points) is based on the assumption that all the rectangles will be in front of all lines in the z-order, so that a line can pass through the bounds of any other rectangle.
Your idea for a 'silo' computer architecture is interesting, but beyond my "depth" to envision.
Yes, there is a kind of a grid inherent here in that these lines connect nodes in hierarchies together: from node to node within one hierarchy, and/or from node to node 'across' hierarchies. So connecting lines are not limited to 'neighbors' in the same x,y cell of the 'virtual grid.'
Since the visual presentation of these hierarchies will be in TreeView form, then indent order visibly reflects 'inheritance' in the hierarchy, and above-below order reflects 'depth,' and 'ordinal position' in relation to sibling nodes.
However, there may be no fundamental relationship between the #n hierarchies currently displayed. A node at a nesting level of #3 in hierarchy #1 may be completely unrelated to the fact there are nodes in another hierarchy with depth of level #3 ... or, that could be meaningful.
Creating and inspecting these 'connections' between 'nodes' will be facilitated by using some graphic techniques: imagine something like you alt-click on a 'node's display rectangle, and all the displayed rectangles for nodes in all hierarchies that have no 'connection' ... direct or indirect ... to the node you selected become 50% transparent.
Or, imagine a use of one color to indicate 'connections' made by the user directly, and another color used for lines which connect indirectly to the selected node's rectangle representation.
While a 'connection line' is defined as an object of #n:1~3 segments, or #2~4 points, it is, architecturally a data structure inside the 'connection object.'
If you think of the total connection objects per one node of a hierarchy, then you can also think of a set of all possible paths from this node to other nodes: that set, the 'path' object, may be organized based on multiple criteria: for example: sequence in time: the user connected node a to node b then to node c one after the other.
But, no, this is not a scenario where a combinatorial 'explosion' of potential 'routes' will be computed, or, like the infamous 'travelling salesman' problem a case where the shortest route will need to be solved based on passing through a required subset of connected nodes.
best, Bill
"In the River of Delights, Panic has not failed me." Jorge Luis Borges
|
|
|
|
|
Ray trace image[^]
As you can see by the image there are considerable artifacts in the render output. I have no idea where to start with this issue. I can post code on request. This is a standard Raytracing model of shooting rays from the camera.
Does anybody know how to fix artifacts like this? (The lines between the triangles even though they touch)
Made in VS2010 c# using a picture box and one image I made in Photoshop
Thank you all
[edit]algorithm posted[/edit]
Here is the algorithm I am using. A lot of extra features removed for simplicity. But this is its core.
for (int X = 0; X < Width; X++)
{
for (int Y = 0; Y < Height; Y++)
{
Ray TRay = new Ray(V3.Zero(), V3.Normalize(new V3(X - HalfWidth, 20f, Y - HalfHeight)));
foreach (Triangle TempTriangle in SceneTriangles)
{
Intersection TInter = TempTriangle.Intersects(TRay);
if (TInter.Hits) Temp.SetPixel(X, Y, CalculateColour(TInter, ColourMode.Lambert));
}
}
}
A ray is defined as a start point and a direction
V3 is a 3D vector
Intersection stores: V3 of impact, If it hit, colour of pixel on triangle
Thomas Williams - Chimp Twist
modified on Monday, August 1, 2011 5:33 AM
|
|
|
|
|
This is the Algorithms forum, please select a forum more appropriate to your subject.
The best things in life are not things.
|
|
|
|
|
It is an algorithm question. I didn't post code to bog down. See Edit. There is an error somewhere in the process I am using to work out each colour of the pixel on the screen.
Which other forum would it belong in? I figured this would be the best place
|
|
|
|
|
I don't think it's so much the fault of the algorithm as it is of the implementation of the hit test. It appears to be "too narrow".
|
|
|
|
|
I've had a go at moving through the pixels in half steps. This hasn't worked, but:
I am going to have a go at getting the average colour of the pixel between the half steps in an attempt to do some anti-aliasing
|
|
|
|
|
Hi Everyone,
I'm working on a 128 bit unsigned integer class (due to Numerics.BigInt being too slow). I've gotten most of the regular arithmetic functions working, but now I'm on the Modulo one. Does anyone know of shortcuts to finding the remainder without having to go through the full division? (Yes, I have googled this for several days, with little progress)
So far the only one I've found is if the divisor is a power of 2, you can use
Remainder = Dividend & (Divisor - 1);
But that's it. Nothing for any other number. So if anyone knows of anything similar, or heck, just a really fast division/mod algorithm, please let me know.
Thanks,
John
|
|
|
|
|
I'm not sure this will help in your case, but it seems that eons ago I saw an algorithm that basically repeatedly bit shifted and subtracted in binary. Perhaps the Mongomery algorithm[^] would suit your purpose?
CQ de W5ALT
Walt Fair, Jr., P. E.
Comport Computing
Specializing in Technical Engineering Software
|
|
|
|
|
This looks interesting. It seems like it could be it. The only problem I can see with it is that it seems to be just for working with primes and semiprimes? Can I use this with any positive integer?
Thanks,
John
|
|
|
|
|
I really don't know. Sorry about that.
CQ de W5ALT
Walt Fair, Jr., P. E.
Comport Computing
Specializing in Technical Engineering Software
|
|
|
|
|
128 bits isn't "really big", so the asymptotically optimal algorithms aren't a very good choice here.
For 128 bits, I think what would work best is a simple multiword division. There's lots of stuff on google, it's in one of Knuth's books, and hackers delight has some code here[^] that also explains the algorithm.
edit: yes that does a division, and you go to mod from there - I know of no general way to do a mod without a division.
Just wondering, how did you implement multiplication? I hope it's not shifts and adds.. it works, to be fair, but it's very much slower than using long multiplication in base 2^32 (or base 2^64 if you have access to a 64x64->128 multiplication)
|
|
|
|
|
Hi,
i have created one portable application,initially it will take several time to install on to pendrive(installer doing more on dll copy option), for that i want to show time remaining in my UI.
can any one suggest me best algorithm to do this please
|
|
|
|
|
If you know how much data you need to install then you can set a timer to measure your install speed and use that value to estimate how much time the remainder will take. Adjust the value as you go through the install. Something like:
Total to install 1Mb
After installing 100Kb time taken is 2 seconds, so time left is ((1000 - 100) / 100 * 2) = 18 seconds
After installing 200Kb time taken is 3 seconds, so time left is ((1000 - 200 / 200 * 3) = 12
etc.
The best things in life are not things.
|
|
|
|
|
for this always we should know how much size is already installed/copied right...?
|
|
|
|
|
yogish293 wrote: for this always we should know how much size is already installed/copied right...?
Yes, you can only ever make a reasonable estimate by knowing:
- How much has already been copied
- How much time the copying took
- How much data remains
The best things in life are not things.
|
|
|
|
|
I've been implementing a PID control algorithm[^] to control a heater (or cooler) to reach a setpoint. Everything works perfectly so far - PID really is magic.
Now the device I'm using can heat or cool depending on the polarity of the voltage you apply to it, so, with a suitable relay and some hardware, I can automate it to switch from heating to cooling or vice versa. As it currently works (I have a manual switch to reverse polarity), if you have a setpoint of, let's say, 30°C, it'll heat up, overshoot, shut off the voltage and cool naturally, then undershoot and turn the voltage back on. In theory, I suppose, it could be more efficient if, instead of letting it cool naturally, I switch the polarity and drive it back down. It's more aggressive and likely to be less stable perhaps, but my real concern is that it'll abuse the heck out of the relay which will be rapidly switching back and forth around the setpoint. So I was thinking that if I have some kind of "dead zone" area to add a "buffer" around the setpoint it might save the relay. In essence, the relay should only switch if you're a long way off and not when you are right around the setpoint. I was thinking if I made my input range say, -10 to +10 and in software I mapped it such that +10 gives the maximum voltage to the device and it reduces linearly until it reaches 0 at, let's say +2. Then from +2 to -2 the voltage remains at zero, and from -2 to -10 the voltage goes up again, but the polarity is reversed.
One concern here is that in addition to stopping it switching polarity, it'll also delay turning it back on. Also, depending on where the setpoint is, I should be spending most of the time either cooling or heating. For example, if I change the setpoint from 10°C to 5°C, I should spend most of my time cooling, I really shouldn't need to heat at all unless I drastically overshoot, and I'm not sure whether this will automatically pick up on that.
Does this sound like something that might work (I'm waiting on the hardware at the moment, so I'm just thinking it through right now)? Is there a way to handle this kind of setup? Is it just a bad idea? Should I just not try and drive it so aggressively?
Any hardware control experts out there?
|
|
|
|
|
It's normal practice in controls to use hysteresis around a set point to prevent overworking the hardware. Even air conditioning systems use it. If you set a target of 70°F, for instance, the heating unit won't kick on until the temp drops to about 68, and won't shut down until it senses 71 - 72. This prevents constant cycling on and off, which is very important if there's a motor involved. They tend to overheat, and can be damaged by rapid cycling.
Don't try to set it to aggressively, else you might fall into a positive feedback situation, with the controlled value oscillating around, but never actually stabilizing at the setpoint. That will ead to rapid equipment failure.
You're on the right track... carry on!
Will Rogers never met me.
|
|
|
|
|
Thanks for the sanity check. Hopefully I'll get the hardware piece I need tomorrow and can start playing around.
|
|
|
|
|
Do not look so much into code - a good setup of the "hard ware" is more important.
Find a good distance for the thermometer from the heater/cooler proper. Mix the item to be heated/cooled well; if you cannot mix the items proper, mix the surrounding media (air or liquid) and do not forget that the delay will be bigger.
|
|
|
|
|
As already suggested, using histeresys is a good, proven technique.
For your control system, you may have a look at fuzzy logic too.
If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.
This is going on my arrogant assumptions. You may have a superb reason why I'm completely wrong.
-- Iain Clarke
[My articles]
|
|
|
|
|
Hello,
What algorithm does one use to find an empty entry on a 3x3 grid?
Suppose you have a checkers board where 8 of the 9 squares are occupied with a checker piece.
How do you determine the one empty spot that is left?
Thank you
|
|
|
|
|
In the absence of any further logical constraints (i.e. emptiness of each cell is independent of other cells), brute force (checking all the cells until you find an empty one) is the only way. (On a 3×3 grid that is no problem.)
|
|
|
|
|
It depends. What data structure is your grid made of? Is it a bit board?
|
|
|
|
|
My Grid is a Xaml Grid (Silverlight). Its internal structure is like this:
<Grid
xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
Loaded="onLoaded" ShowGridLines="True" Background="CornflowerBlue">
<Grid.ColumnDefinitions>
<ColumnDefinition Width="100"/>
<ColumnDefinition Width="100"/>
<ColumnDefinition Width="100"/>
<ColumnDefinition/>
</Grid.ColumnDefinitions>
<Grid.RowDefinitions>
<RowDefinition Height="100"/>
<RowDefinition Height="100"/>
<RowDefinition Height="100"/>
</Grid.RowDefinitions>
<Border Grid.Row="0" Grid.Column="0" x:Name="b1" MouseLeftButtonUp="switchPlaces" >
<Canvas x:Name="canvas1">
<Ellipse Width="100" Height="100" x:Name="ellipse1" Fill="Red" Visibility="Visible"/>
</Canvas>
</Border>
/* repeat 8 times */
|
|
|
|
|
Brute force it is then.
I hope you're not planning to make a big grid out of this, because it's going to be terribly slow..
|
|
|
|
|