|
code
newsgroups
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Re: Arrays. Beginner's question!"Suzy" <not@valid> wrote in message news:474c68ed$1@dnews.tpgi.com.au... Do I take it from the deafening silence that this method is OK in this case?> > "Steve Gerrard" <mynameh***@comcast.net> wrote in message > news:7PqdnXu93P6JXdbanZ2dnUVZ_hmtnZ2d@comcast.com... >> >> "Suzy" <not@valid> wrote in message news:474bb9bc@dnews.tpgi.com.au... >>> >>> Thanks Mike. I was going to post that when using a variant I am getting >>> slightly different results to when I am using Double and was wondering >>> why. >>> >>> The old code was >>> >>> Dim K1(7) as double >>> ... >>> K1(1)=0.4711 >>> K1(2)=0.462 >>> K1(3)=0.4538 >>> etc... >>> >>> which I changed (on advice) to >>> >>> Dim K1 As Variant >>> ... >>> K1 = Array(0, 0.4711, 0.462, 0.4538, 0.4491, 0.4421, 0.4385, 0.4268) >>> >>> Calculations using this data came out slightly differently using the two >>> methods. Any suggestions? >> >> Can you show an example of a calculation that comes out different? >> >> In your new version, you have a variant, which is an array of variants, >> where each element is variant type double, which is not quite the same as >> being a double. >> >> For a really clean result, I would take Bob's suggestion, and do >> >> Dim KTemp As Variant >> Dim K1(7) As Double >> Dim N As Long >> >> KTemp = Array(0, 0.4711, 0.462, 0.4538, 0.4491, 0.4421, 0.4385, 0.4268) >> >> For N = 0 To 7 >> K1(N) = KTemp(N) >> Next N >> >> Then you have a simple array of doubles, no doubts about data type. >> >> > Well, in fact I did just that and found the anomaly still remained. I went > back to the variant method and the results were exactly the same! So I > had a thirtieth check of all the array elements and... of course there was > a typo! .4358 instead of .4385 or somesuch. I put that right and now the > anomaly has gone. Silly me. > > The code I am using is now: > > Dim K As Variant 'arrays > Dim K1 As Variant > Dim K2 As Variant > Dim K3 As Variant > Dim K4 As Variant > ... > K = Array(0, 0.001, 0.003, 0.005, 0.007, 0.01, 0.015, 0.02) > K1 = Array(0, 0.4711, 0.462, 0.4538, 0.4491, 0.4421, 0.4358, 0.4268) > K2 = Array(0, 0.018, 0.01941, 0.02117, 0.02274, 0.02396, 0.02558, > 0.02614) > K3 = Array(0, 0.08398, 0.08543, 0.0951, 0.08801, 0.1027, 0.1149, > 0.1112) > K4 = Array(0, 0.965, 0.9697, 1.007, 0.9044, 1.038, 1.034, 1.036) > > and it all works fine. Could Mike tell me why (if it is not) the use of a > variant is not advisable in this particular case? >> Suzy wrote:
Show quote > "Suzy" <not@valid> wrote in message news:474c68ed$1@dnews.tpgi.com.au... For some definition of "fine" undoubtedly... :)>> "Steve Gerrard" <mynameh***@comcast.net> wrote in message >> news:7PqdnXu93P6JXdbanZ2dnUVZ_hmtnZ2d@comcast.com... >>> "Suzy" <not@valid> wrote in message news:474bb9bc@dnews.tpgi.com.au... >>>> Thanks Mike. I was going to post that when using a variant I am getting >>>> slightly different results to when I am using Double and was wondering >>>> why. >>>> >>>> The old code was >>>> >>>> Dim K1(7) as double >>>> ... >>>> K1(1)=0.4711 >>>> K1(2)=0.462 >>>> K1(3)=0.4538 >>>> etc... >>>> >>>> which I changed (on advice) to >>>> >>>> Dim K1 As Variant >>>> ... >>>> K1 = Array(0, 0.4711, 0.462, 0.4538, 0.4491, 0.4421, 0.4385, 0.4268) >>>> >>>> Calculations using this data came out slightly differently using the two >>>> methods. Any suggestions? >>> Can you show an example of a calculation that comes out different? >>> >>> In your new version, you have a variant, which is an array of variants, >>> where each element is variant type double, which is not quite the same as >>> being a double. >>> >>> For a really clean result, I would take Bob's suggestion, and do >>> >>> Dim KTemp As Variant >>> Dim K1(7) As Double >>> Dim N As Long >>> >>> KTemp = Array(0, 0.4711, 0.462, 0.4538, 0.4491, 0.4421, 0.4385, 0.4268) >>> >>> For N = 0 To 7 >>> K1(N) = KTemp(N) >>> Next N >>> >>> Then you have a simple array of doubles, no doubts about data type. >>> >>> >> Well, in fact I did just that and found the anomaly still remained. I went >> back to the variant method and the results were exactly the same! So I >> had a thirtieth check of all the array elements and... of course there was >> a typo! .4358 instead of .4385 or somesuch. I put that right and now the >> anomaly has gone. Silly me. >> >> The code I am using is now: >> >> Dim K As Variant 'arrays >> Dim K1 As Variant >> Dim K2 As Variant >> Dim K3 As Variant >> Dim K4 As Variant >> ... >> K = Array(0, 0.001, 0.003, 0.005, 0.007, 0.01, 0.015, 0.02) >> K1 = Array(0, 0.4711, 0.462, 0.4538, 0.4491, 0.4421, 0.4358, 0.4268) >> K2 = Array(0, 0.018, 0.01941, 0.02117, 0.02274, 0.02396, 0.02558, >> 0.02614) >> K3 = Array(0, 0.08398, 0.08543, 0.0951, 0.08801, 0.1027, 0.1149, >> 0.1112) >> K4 = Array(0, 0.965, 0.9697, 1.007, 0.9044, 1.038, 1.034, 1.036) >> >> and it all works fine. Could Mike tell me why (if it is not) the use of a >> variant is not advisable in this particular case? > Do I take it from the deafening silence that this method is OK in this case? What's the real question, again? What's the point of having multiple Kx arrays rather than a 2D single array? There would superficially seem not advantage to Variant arrays for what appear to be "real" numeric values that wouldn't seem likely to need to change their spots in the application, thus negating any reason for Variant other than the convenience of the Array() notation for initializing them. If that's the issue, I personally would be highly likely to use one of the other workarounds to the lack of Data in VB (although I'll never say never :) ). As an unrelated aside--If VBFred reinstituted Data or an equivalent construct, maybe there's one spot I could grant Michael C an advantage to dotnet... <vbg> -- "Suzy" <not@valid> wrote in message news:474db8ed@dnews.tpgi.com.au... <cut>>> and it all works fine. Could Mike tell me why (if it is not) the use of a If it works and the performance is acceptable then go for it. If you have >> variant is not advisable in this particular case? >>> > Do I take it from the deafening silence that this method is OK in this > case? reason to believe that the scope will change such that performance may degrade using this approach then you may want to re-visit it.
Show quote
"Bob Butler" <noway@nospam.ever> wrote in message
news:ewqracfMIHA.1188@TK2MSFTNGP04.phx.gbl... > "Suzy" <not@valid> wrote in message news:474db8ed@dnews.tpgi.com.au... > <cut> >>> and it all works fine. Could Mike tell me why (if it is not) the use of >>> a variant is not advisable in this particular case? >>>> >> Do I take it from the deafening silence that this method is OK in this >> case? > > > If it works and the performance is acceptable then go for it. If you have > reason to believe that the scope will change such that performance may > degrade using this approach then you may want to re-visit it. > >Thanks Bob. It does work now that I have fixed my stupid typo. "Suzy" <not@valid> wrote in message news:474dc878@dnews.tpgi.com.au... Suzy, the reason you did not get a response initially to your earlier post >> If it works and the performance is acceptable then go for it. If >> you have reason to believe that the scope will change such that >> performance may degrade using this approach then you may >> want to re-visit it. >> >>Thanks Bob. It does work now that I have fixed my stupid typo. (where you asked, "Do I take it that this method is okay) is because most people probably missed it. That's because you are making it very difficult for people to see what you have posted. In your responses you quote an extract from the post to which you are responding, thus: > Are you using arrays in a Variant .. . . and you are inserting your own reply beneath that extract, but you are ALSO indenting your own reply, thus: > Yes. That's what I am doing .. . . so you end up with:> Are you using arrays in a Variant This is a very strange way of arranging your post, and we cannot tell where > Yes. That's what I am doing. the "extract from the original" ends and your own response begins. You should NOT do that. Instead of the above you should arrange it as follows: > Are you using arrays in a Variant Yes. That's what I am doing.Note that ONLY the extract from the original is indented. Your own response is NOT indented. In that we we can immediately see what's what :-) Mike
Show quote
"Suzy" <not@valid> wrote Oddly enough, I do not see the post you are replying to. Perhaps others> > The code I am using is now: > > > > Dim K As Variant 'arrays > > Dim K1 As Variant > > Dim K2 As Variant > > Dim K3 As Variant > > Dim K4 As Variant > > ... > > K = Array(0, 0.001, 0.003, 0.005, 0.007, 0.01, 0.015, 0.02) > > K1 = Array(0, 0.4711, 0.462, 0.4538, 0.4491, 0.4421, 0.4358, 0.4268) > > K2 = Array(0, 0.018, 0.01941, 0.02117, 0.02274, 0.02396, 0.02558, > > 0.02614) > > K3 = Array(0, 0.08398, 0.08543, 0.0951, 0.08801, 0.1027, 0.1149, > > 0.1112) > > K4 = Array(0, 0.965, 0.9697, 1.007, 0.9044, 1.038, 1.034, 1.036) > > > > and it all works fine. Could Mike tell me why (if it is not) the use of a > > variant is not advisable in this particular case? > >> > Do I take it from the deafening silence that this method is OK in this case? did not see it as well? About your use of Variants, I would wonder why you would not prefer to use an array of the appropreate data type. There are several ways you could make the assignment to your array(s) and still have arrays that are typed as needed. For example, a sub similar to: Sub ReadData(ByRef Group() As Double, ParamArray List()) Dim index As Long ReDim Group(0 To UBound(List)) For index = 0 To UBound(List) Group(index) = List(index) Next End Sub .... would allow you to build arrays of Doubles in much the same manner as you are using now: Dim K() As Double Dim K1() As Double ' ... Dim K4() As Double ReadData K, 0, 0.001, 0.003, 0.005, 0.007, 0.01, 0.015, 0.02 ReadData K1, 0, 0.4711, 0.462, 0.4538, 0.4491, 0.4421, 0.4358, 0.4268 '... ReadData K4, 0, 0.965, 0.9697, 1.007, 0.9044, 1.038, 1.034, 1.036 If you put your values in a typed array, VB can optimize your efforts knowing you want to work on Double values. If you leave them as Variants, VB really can't make any assumtions about your data. For a contrived example where using a properly typed array makes a difference, consider what the output would be in the following case: Dim A1, A2() As Double ' Assignment of the same data, using the different methods A1 = Array("2", "3") ReadData A2, "2", "3" ' Identical output??? Debug.Print A1(0) + A1(1) Debug.Print A2(0) + A2(1) Try it and see the difference.... LFS "Suzy" <not@valid> wrote in message news:474db8ed@dnews.tpgi.com.au... If that's what you want to use then it's okay. Personally I would much > K1 = Array(0, 0.4711, 0.462, 0.4538, 0.4491, 0.4421, 0.4358, 0.4268) > K2 = Array(0, 0.018, 0.01941, 0.02117, 0.02274, 0.02396, 0.02558, Do I > take it from the deafening silence that this method is OK in this case? rather use a standard array of Doubles and I only ever use Variants if there is simply no other option, partly because they occupy a bit more memory and partly because they are quite a lot slower. But if you're not using thousands of them and if you're not performing tens of thousands of calculations on them then you'll be okay. In your specific case, where it looks as though you are using a limited number of Variants for some fairly simple calculations then you probably won't notice the difference, but for some applications they will slow down your code considerably. As an example, I've just done the following: For p = 1 To 100 For n = 0 To 7 k1(n) = k1(n) + k2(n) Next n Next p If the arrays k1() and k2() are Variants, as in your own code, it takes about 200 microseconds to execute the above loop in a standard native code optimised compiled exe. However, if the arrays k1() and k2() are standard arrays of Doubles the same loop takes only about 3 microseconds to execute (yes, that's 3). I think you'll agree that's quite a considerable difference. Mike
Show quote
"Suzy" <not@valid> wrote in message news:474db8ed@dnews.tpgi.com.au... I don't know about them, but it is certainly OK as far as I am concerned.> > > <snipped> > > > > The code I am using is now: > > > > Dim K As Variant 'arrays > > Dim K1 As Variant > > Dim K2 As Variant > > Dim K3 As Variant > > Dim K4 As Variant > > ... > > K = Array(0, 0.001, 0.003, 0.005, 0.007, 0.01, 0.015, 0.02) > > K1 = Array(0, 0.4711, 0.462, 0.4538, 0.4491, 0.4421, 0.4358, 0.4268) > > K2 = Array(0, 0.018, 0.01941, 0.02117, 0.02274, 0.02396, 0.02558, > > 0.02614) > > K3 = Array(0, 0.08398, 0.08543, 0.0951, 0.08801, 0.1027, 0.1149, > > 0.1112) > > K4 = Array(0, 0.965, 0.9697, 1.007, 0.9044, 1.038, 1.034, 1.036) > > > > and it all works fine. Could Mike tell me why (if it is not) the use of a > > variant is not advisable in this particular case? > >> > Do I take it from the deafening silence that this method is OK in this case? > No reason to develop a rabid fear of Variants. Just don't over-do it. They are part of the culture. Variants show up all over the place so you are likely using them even if you don't think you are. Recordsets for example are nothing but Collections of Collections containing Variants, OLE datatypes are Variants, even properties of common controls, object libraries, and VBA library functions are often thinly veiled Variants. I don't believe in over-thinking pre-optimizing. So the only concern in this case is how often are these arrays used? Are you abusing them to the point where you will get some measurable value out of working them to a native type or not? It looks like you are trying to get around the lack of a Constant Array in VB. But didn't we just have this same discussion a while ago? (Data... Read...) One small thing. In your example above, Kx(0) is a vbInteger, the rest are Doubles. -ralph  |
|||||||||||||||||||||||
Other interesting topics