correction function cumsum(a,s=a[0],s1=[ ],n=1)= n==len(a)-1?s1:cumsum(a,s+a[n],concat(s1,s),n+1); these are always very confusing to me On Tue, 28 Mar 2023, 10:09 Sanjeev Prabhakar, <sprabhakar2006@gmail.com> wrote: > this is definitely quadratic > i did not bother too much about the speed in this case as this was simpler > to write. > > the other way to write the same function which is linear is as following: > function cumsum(a,s=0,s1=[ ],n=0)= > n==len(a)-1?s1:cumsum(a,s+a[n],concat(s1,s),n+1); > > a=[ for(i=[0:1000])i]; > echo(cumsum(a)); > > On Tue, 28 Mar 2023, 02:08 Adrian Mariano, <avm4@cornell.edu> wrote: > >> Summing a vector using the dot product is good---best way to compute sum >> of a list of numbers in OpenSCAD. But doing cumulative sum with a loop of >> sum operations is bad---it turns a linear operation into a quadratic >> operation. Despite the fact that matrix operations are super fast in >> OpenSCAD, the code posted above that does cumulative sums via a loop over >> dot products is slower than the recursive cumulative sum approach once >> N>9. >> >> On Mon, Mar 27, 2023 at 2:01 PM David Phillip Oster < >> davidphilliposter@gmail.com> wrote: >> >>> I was impressed by the idea of summing a vector by doing the dot product >>> of the vector and an all-ones vector. Very clever, and it avoids the >>> recursion. >>> >>> On Mon, Mar 27, 2023 at 7:11 AM Raymond West <raywest@raywest.com> >>> wrote: >>> >>>> Thanks for the info. >>>> >>>> I'm using >>>> >>>> function sumvector(v,i) = i == 0 ? v[i] : v[i] + sumvector(v, i-1); >>>> >>>> function incrementalSum(v,wall) = [for(i = [0:len(v)-1]) >>>> sumvector(v,i)+wall*i]; >>>> >>>> which is as David posted earlier. It works fine for the relatively >>>> short lists I'm using. >>>> >>>> >>>> On 27/03/2023 09:06, Sanjeev Prabhakar wrote: >>>> >>>> cumulative sum calculation for a list >>>> a=[1,2,3,4,5,6,7,8,9]; >>>> cs=[ for(i=[0:len(a)-1]) a*[ for (j=[0:len(a)-1]) j<=i?1:0]]; >>>> echo(cs); >>>> >>>> >>>> On Mon, 27 Mar 2023, 12:54 Curt McDowell, <maker@fishlet.com> wrote: >>>> >>>>> The general answer is to use a recursive function to implement a loop. >>>>> Functional programming is quite painful and is a bane of OpenSCAD. As in, >>>>> experienced developers need help to express what should be a simple "for" >>>>> loop. Forget about nested ones with interrelated loop boundaries and >>>>> variable dependencies. OpenSCAD survives because it has workarounds like >>>>> for() inside modules and list comprehensions. >>>>> >>>>> I get why purists like FP, but as the quote goes: it makes difficult >>>>> things easy, and easy things very difficult. >>>>> >>>>> walls = function(pockets, wall) >>>>> let (loop = function(pockets, wall, ans, i, sum) >>>>> i >= len(pockets) ? ans : >>>>> let (next = sum + pockets[i] + wall) >>>>> loop(pockets, wall, concat(ans, [next]), i + >>>>> 1, next)) >>>>> loop(pockets, wall, [], 0, 0); >>>>> >>>>> echo(walls([20, 20, 30, 30], 2)); >>>>> >>>>> Regards, >>>>> Curt >>>>> On 3/24/2023 8:03 AM, Raymond West wrote: >>>>> >>>>> I guess there is a succinct general answer, but i can't discover it. >>>>> >>>>> Given a list (describing the width of pockets within a drawer, say,) >>>>> and a wall thickness for each pocket, I want to be able to locate >>>>> (translate) the actual position of the walls. >>>>> >>>>> e.g. list of pocket widths list[20,20,30,30] and wall =2, I want >>>>> to derive the values 22, 44,7 6, 108 to be generated. >>>>> _______________________________________________ >>>>> OpenSCAD mailing list >>>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>>> >>>>> _______________________________________________ >>>>> OpenSCAD mailing list >>>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>>> >>>> _______________________________________________ >>>> OpenSCAD mailing list >>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>> >>> _______________________________________________ >>> OpenSCAD mailing list >>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>> >> _______________________________________________ >> OpenSCAD mailing list >> To unsubscribe send an email to discuss-leave@lists.openscad.org >> >

unable to get over it! This is the final one. I worked on openscad after a long time. function cumsum(a,s=a[0],s1=[ ],n=1)= n==len(a)?concat(s1,s):cumsum(a,s+a[n],concat(s1,s),n+1); function sum(a,s=a[0],n=1)= n==len(a)?s:sum(a,s+a[n],n+1); a=[for(i=[0:10])i]; echo(cumsum(a),sum(a)); On Tue, 28 Mar 2023 at 10:20, Sanjeev Prabhakar <sprabhakar2006@gmail.com> wrote: > correction > > function cumsum(a,s=a[0],s1=[ ],n=1)= > n==len(a)-1?s1:cumsum(a,s+a[n],concat(s1,s),n+1); > > these are always very confusing to me > > On Tue, 28 Mar 2023, 10:09 Sanjeev Prabhakar, <sprabhakar2006@gmail.com> > wrote: > >> this is definitely quadratic >> i did not bother too much about the speed in this case as this was >> simpler to write. >> >> the other way to write the same function which is linear is as following: >> function cumsum(a,s=0,s1=[ ],n=0)= >> n==len(a)-1?s1:cumsum(a,s+a[n],concat(s1,s),n+1); >> >> a=[ for(i=[0:1000])i]; >> echo(cumsum(a)); >> >> On Tue, 28 Mar 2023, 02:08 Adrian Mariano, <avm4@cornell.edu> wrote: >> >>> Summing a vector using the dot product is good---best way to compute sum >>> of a list of numbers in OpenSCAD. But doing cumulative sum with a loop of >>> sum operations is bad---it turns a linear operation into a quadratic >>> operation. Despite the fact that matrix operations are super fast in >>> OpenSCAD, the code posted above that does cumulative sums via a loop over >>> dot products is slower than the recursive cumulative sum approach once >>> N>9. >>> >>> On Mon, Mar 27, 2023 at 2:01 PM David Phillip Oster < >>> davidphilliposter@gmail.com> wrote: >>> >>>> I was impressed by the idea of summing a vector by doing the dot >>>> product of the vector and an all-ones vector. Very clever, and it avoids >>>> the recursion. >>>> >>>> On Mon, Mar 27, 2023 at 7:11 AM Raymond West <raywest@raywest.com> >>>> wrote: >>>> >>>>> Thanks for the info. >>>>> >>>>> I'm using >>>>> >>>>> function sumvector(v,i) = i == 0 ? v[i] : v[i] + sumvector(v, i-1); >>>>> >>>>> function incrementalSum(v,wall) = [for(i = [0:len(v)-1]) >>>>> sumvector(v,i)+wall*i]; >>>>> >>>>> which is as David posted earlier. It works fine for the relatively >>>>> short lists I'm using. >>>>> >>>>> >>>>> On 27/03/2023 09:06, Sanjeev Prabhakar wrote: >>>>> >>>>> cumulative sum calculation for a list >>>>> a=[1,2,3,4,5,6,7,8,9]; >>>>> cs=[ for(i=[0:len(a)-1]) a*[ for (j=[0:len(a)-1]) j<=i?1:0]]; >>>>> echo(cs); >>>>> >>>>> >>>>> On Mon, 27 Mar 2023, 12:54 Curt McDowell, <maker@fishlet.com> wrote: >>>>> >>>>>> The general answer is to use a recursive function to implement a >>>>>> loop. Functional programming is quite painful and is a bane of OpenSCAD. As >>>>>> in, experienced developers need help to express what should be a simple >>>>>> "for" loop. Forget about nested ones with interrelated loop boundaries and >>>>>> variable dependencies. OpenSCAD survives because it has workarounds like >>>>>> for() inside modules and list comprehensions. >>>>>> >>>>>> I get why purists like FP, but as the quote goes: it makes difficult >>>>>> things easy, and easy things very difficult. >>>>>> >>>>>> walls = function(pockets, wall) >>>>>> let (loop = function(pockets, wall, ans, i, sum) >>>>>> i >= len(pockets) ? ans : >>>>>> let (next = sum + pockets[i] + wall) >>>>>> loop(pockets, wall, concat(ans, [next]), i + >>>>>> 1, next)) >>>>>> loop(pockets, wall, [], 0, 0); >>>>>> >>>>>> echo(walls([20, 20, 30, 30], 2)); >>>>>> >>>>>> Regards, >>>>>> Curt >>>>>> On 3/24/2023 8:03 AM, Raymond West wrote: >>>>>> >>>>>> I guess there is a succinct general answer, but i can't discover it. >>>>>> >>>>>> Given a list (describing the width of pockets within a drawer, say,) >>>>>> and a wall thickness for each pocket, I want to be able to locate >>>>>> (translate) the actual position of the walls. >>>>>> >>>>>> e.g. list of pocket widths list[20,20,30,30] and wall =2, I want >>>>>> to derive the values 22, 44,7 6, 108 to be generated. >>>>>> _______________________________________________ >>>>>> OpenSCAD mailing list >>>>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>>>> >>>>>> _______________________________________________ >>>>>> OpenSCAD mailing list >>>>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>>>> >>>>> _______________________________________________ >>>>> OpenSCAD mailing list >>>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>>> >>>> _______________________________________________ >>>> OpenSCAD mailing list >>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>> >>> _______________________________________________ >>> OpenSCAD mailing list >>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>> >>

function cumsum(a,s=0,s1=[ ],n=0)= n==len(a)?concat(s1,s):cumsum(a,s+a[n],n>0?concat(s1,s):[],n+1); function sum(a,s=0,n=0)= n==len(a)?s:sum(a,s+a[n],n+1); a=[for(i=[0:10])i]; echo(cumsum(a),sum(a)); b=[1,3,2,5,7]; echo(cumsum(b)); On Tue, 28 Mar 2023 at 20:41, Sanjeev Prabhakar <sprabhakar2006@gmail.com> wrote: > unable to get over it! > This is the final one. I worked on openscad after a long time. > > function cumsum(a,s=a[0],s1=[ ],n=1)= > n==len(a)?concat(s1,s):cumsum(a,s+a[n],concat(s1,s),n+1); > > function sum(a,s=a[0],n=1)= n==len(a)?s:sum(a,s+a[n],n+1); > > a=[for(i=[0:10])i]; > echo(cumsum(a),sum(a)); > > > > On Tue, 28 Mar 2023 at 10:20, Sanjeev Prabhakar <sprabhakar2006@gmail.com> > wrote: > >> correction >> >> function cumsum(a,s=a[0],s1=[ ],n=1)= >> n==len(a)-1?s1:cumsum(a,s+a[n],concat(s1,s),n+1); >> >> these are always very confusing to me >> >> On Tue, 28 Mar 2023, 10:09 Sanjeev Prabhakar, <sprabhakar2006@gmail.com> >> wrote: >> >>> this is definitely quadratic >>> i did not bother too much about the speed in this case as this was >>> simpler to write. >>> >>> the other way to write the same function which is linear is as following: >>> function cumsum(a,s=0,s1=[ ],n=0)= >>> n==len(a)-1?s1:cumsum(a,s+a[n],concat(s1,s),n+1); >>> >>> a=[ for(i=[0:1000])i]; >>> echo(cumsum(a)); >>> >>> On Tue, 28 Mar 2023, 02:08 Adrian Mariano, <avm4@cornell.edu> wrote: >>> >>>> Summing a vector using the dot product is good---best way to compute >>>> sum of a list of numbers in OpenSCAD. But doing cumulative sum with a loop >>>> of sum operations is bad---it turns a linear operation into a quadratic >>>> operation. Despite the fact that matrix operations are super fast in >>>> OpenSCAD, the code posted above that does cumulative sums via a loop over >>>> dot products is slower than the recursive cumulative sum approach once >>>> N>9. >>>> >>>> On Mon, Mar 27, 2023 at 2:01 PM David Phillip Oster < >>>> davidphilliposter@gmail.com> wrote: >>>> >>>>> I was impressed by the idea of summing a vector by doing the dot >>>>> product of the vector and an all-ones vector. Very clever, and it avoids >>>>> the recursion. >>>>> >>>>> On Mon, Mar 27, 2023 at 7:11 AM Raymond West <raywest@raywest.com> >>>>> wrote: >>>>> >>>>>> Thanks for the info. >>>>>> >>>>>> I'm using >>>>>> >>>>>> function sumvector(v,i) = i == 0 ? v[i] : v[i] + sumvector(v, i-1); >>>>>> >>>>>> function incrementalSum(v,wall) = [for(i = [0:len(v)-1]) >>>>>> sumvector(v,i)+wall*i]; >>>>>> >>>>>> which is as David posted earlier. It works fine for the relatively >>>>>> short lists I'm using. >>>>>> >>>>>> >>>>>> On 27/03/2023 09:06, Sanjeev Prabhakar wrote: >>>>>> >>>>>> cumulative sum calculation for a list >>>>>> a=[1,2,3,4,5,6,7,8,9]; >>>>>> cs=[ for(i=[0:len(a)-1]) a*[ for (j=[0:len(a)-1]) j<=i?1:0]]; >>>>>> echo(cs); >>>>>> >>>>>> >>>>>> On Mon, 27 Mar 2023, 12:54 Curt McDowell, <maker@fishlet.com> wrote: >>>>>> >>>>>>> The general answer is to use a recursive function to implement a >>>>>>> loop. Functional programming is quite painful and is a bane of OpenSCAD. As >>>>>>> in, experienced developers need help to express what should be a simple >>>>>>> "for" loop. Forget about nested ones with interrelated loop boundaries and >>>>>>> variable dependencies. OpenSCAD survives because it has workarounds like >>>>>>> for() inside modules and list comprehensions. >>>>>>> >>>>>>> I get why purists like FP, but as the quote goes: it makes difficult >>>>>>> things easy, and easy things very difficult. >>>>>>> >>>>>>> walls = function(pockets, wall) >>>>>>> let (loop = function(pockets, wall, ans, i, sum) >>>>>>> i >= len(pockets) ? ans : >>>>>>> let (next = sum + pockets[i] + wall) >>>>>>> loop(pockets, wall, concat(ans, [next]), i + >>>>>>> 1, next)) >>>>>>> loop(pockets, wall, [], 0, 0); >>>>>>> >>>>>>> echo(walls([20, 20, 30, 30], 2)); >>>>>>> >>>>>>> Regards, >>>>>>> Curt >>>>>>> On 3/24/2023 8:03 AM, Raymond West wrote: >>>>>>> >>>>>>> I guess there is a succinct general answer, but i can't discover it. >>>>>>> >>>>>>> Given a list (describing the width of pockets within a drawer, say,) >>>>>>> and a wall thickness for each pocket, I want to be able to locate >>>>>>> (translate) the actual position of the walls. >>>>>>> >>>>>>> e.g. list of pocket widths list[20,20,30,30] and wall =2, I want >>>>>>> to derive the values 22, 44,7 6, 108 to be generated. >>>>>>> _______________________________________________ >>>>>>> OpenSCAD mailing list >>>>>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>>>>> >>>>>>> _______________________________________________ >>>>>>> OpenSCAD mailing list >>>>>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>>>>> >>>>>> _______________________________________________ >>>>>> OpenSCAD mailing list >>>>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>>>> >>>>> _______________________________________________ >>>>> OpenSCAD mailing list >>>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>>> >>>> _______________________________________________ >>>> OpenSCAD mailing list >>>> To unsubscribe send an email to discuss-leave@lists.openscad.org >>>> >>>