Maximizing $x^4+y^4+z^4 + p(xy+xz+yz)$ on a sphere












3












$begingroup$


What values of $x,y,z$ maximize $f(x,y,z,p) = x^4+y^4+z^4 + p(xy+xz+yz)$ with constant $p geq 0$ with the constraint $x^2+y^2+z^2=1$?



Some preliminary studies in Mathematica showed that the behavior of the solution changes around $p=1$. For $p>1$, the problem seems solved by $x=y=z=frac{1}{sqrt 3}$. For $0 leq p leq 1$ it appears the solution (up to permutations of x,y,z) looks like x > y = z. These solutions appear to favor one or the other of the two terms in $f$, and I find the $p>1$ behavior striking and would appreciate a proof that $x=y=z=frac{1}{sqrt 3}$ indeed maximizes.





UPDATE




For at least $p>4/3$, one can show that $x=y=z=frac{1}{sqrt 3}$ maximizes f.



The proof for $1<p<4/3$ evades me currently (the left bound, as mentioned in comments, may be slightly higher than 1).



I'll assume without loss of generality that $x,y,z$ are all positive. Note that $f(frac{1}{sqrt 3},frac{1}{sqrt 3},frac{1}{sqrt 3}, p) = p+frac{1}{3} $, so if we show that for some $p$ we enjoy $f(x,y,z,p) leq p+frac{1}{3}$, we've shown $x=y=z=frac{1}{sqrt 3}$ maximizes $f$.



Now consider that via the constraint, $$f(x,y,z,p) = 1-2(x^2y^2+x^2z^2+y^2z^2) + p(xy+xz+yz)$$
The root mean square compared to arithmetic mean inequality tells us that for nonnegative $a,b,c$ we enjoy $sqrt{frac{a^2+b^2+c^2}{3}} geq frac{a+b+c}{3},$ so $-(a^2+b^2+c^2) leq -frac{(a+b+c)^2}{3}$. Then $$f(x,y,z,p) leq 1 + p(xy+xz+yz) -frac{2}{3}(xy+xz+yz)^2.$$
Now, let's call $(xy+xz+yz)=q$. Note that it is easy to prove with our constraints that $0leq q leq 1$.
Then we can bound the maximum of $f(x,y,z,p)$ by finding the maximum of $g(q) = 1+pq-frac{2}{3}q^2$ on $0leq q leq 1$.
The derivative $g'(q) = p-frac{4}{3}q$, so $g(q)$ increases up to $q=frac{3}{4}p$. Then on $0leq q leq 1$, we have $q_{max} = min(frac{3}{4}p, 1)$.
Thus for $p geq frac{4}{3}$ we have $q_{max}=1$, so $$ f(x,y,z,p) leq 1+p*1-frac{2}{3}*1^2 = p + 1/3$$
Thus since $f(frac{1}{sqrt 3},frac{1}{sqrt 3},frac{1}{sqrt 3}, p) = p+frac{1}{3}$, we have for $p geq frac{4}{3}$ that $x=y=z=frac{1}{sqrt 3}$ maximizes $f$.






I have attempted the method of Lagrange multipliers, which led me to three cubic equations:




I set out to maximize
$4(x^3+y^3+z^3)+p(xy+xz+yz)+lambda(x^2+y^2+z^2-1)$.



This yielded
$$4x^3+2xlambda + p(y+z)=0$$
$$4y^3+2ylambda + p(x+z)=0$$
$$4z^3+2zlambda + p(y+x)=0$$
But now I feel a little at a loss on how to solve this system of equations.
One can verify that $x=y=z=frac{1}{sqrt{3}}$ solves these equations at any p with the right choice of lamda. I'm unsure about pairing Hessian techniques to verify the local nature of the critical point when using lagrange multipliers.









I have also tried incorporating the constraint by writing $z=cos(theta)$, $x=sin(theta)cos(phi)$ and $y=sin(theta)sin(phi)$.




This yields
$$f(theta, phi, p) = cos(theta)^4+(sin(theta)cos(phi))^4+(sin(theta)sin(phi))^4+p(cos(theta)sin(theta)sin(phi)+cos(theta)sin(theta)cos(phi)+(sin(theta))^2sin(phi)cos(phi) ) $$
which I am unsure how to maximize. However, with motivation from Michael Behrend, I've found that studying $phi=frac{pi}{4}$ shows interesting critical point behavior slightly above p=1 where three critical points undergo some sort of bifurcation leaving just a single critical point at the eternal $x=y=z=frac{1}{3}$ .











share|cite|improve this question











$endgroup$

















    3












    $begingroup$


    What values of $x,y,z$ maximize $f(x,y,z,p) = x^4+y^4+z^4 + p(xy+xz+yz)$ with constant $p geq 0$ with the constraint $x^2+y^2+z^2=1$?



    Some preliminary studies in Mathematica showed that the behavior of the solution changes around $p=1$. For $p>1$, the problem seems solved by $x=y=z=frac{1}{sqrt 3}$. For $0 leq p leq 1$ it appears the solution (up to permutations of x,y,z) looks like x > y = z. These solutions appear to favor one or the other of the two terms in $f$, and I find the $p>1$ behavior striking and would appreciate a proof that $x=y=z=frac{1}{sqrt 3}$ indeed maximizes.





    UPDATE




    For at least $p>4/3$, one can show that $x=y=z=frac{1}{sqrt 3}$ maximizes f.



    The proof for $1<p<4/3$ evades me currently (the left bound, as mentioned in comments, may be slightly higher than 1).



    I'll assume without loss of generality that $x,y,z$ are all positive. Note that $f(frac{1}{sqrt 3},frac{1}{sqrt 3},frac{1}{sqrt 3}, p) = p+frac{1}{3} $, so if we show that for some $p$ we enjoy $f(x,y,z,p) leq p+frac{1}{3}$, we've shown $x=y=z=frac{1}{sqrt 3}$ maximizes $f$.



    Now consider that via the constraint, $$f(x,y,z,p) = 1-2(x^2y^2+x^2z^2+y^2z^2) + p(xy+xz+yz)$$
    The root mean square compared to arithmetic mean inequality tells us that for nonnegative $a,b,c$ we enjoy $sqrt{frac{a^2+b^2+c^2}{3}} geq frac{a+b+c}{3},$ so $-(a^2+b^2+c^2) leq -frac{(a+b+c)^2}{3}$. Then $$f(x,y,z,p) leq 1 + p(xy+xz+yz) -frac{2}{3}(xy+xz+yz)^2.$$
    Now, let's call $(xy+xz+yz)=q$. Note that it is easy to prove with our constraints that $0leq q leq 1$.
    Then we can bound the maximum of $f(x,y,z,p)$ by finding the maximum of $g(q) = 1+pq-frac{2}{3}q^2$ on $0leq q leq 1$.
    The derivative $g'(q) = p-frac{4}{3}q$, so $g(q)$ increases up to $q=frac{3}{4}p$. Then on $0leq q leq 1$, we have $q_{max} = min(frac{3}{4}p, 1)$.
    Thus for $p geq frac{4}{3}$ we have $q_{max}=1$, so $$ f(x,y,z,p) leq 1+p*1-frac{2}{3}*1^2 = p + 1/3$$
    Thus since $f(frac{1}{sqrt 3},frac{1}{sqrt 3},frac{1}{sqrt 3}, p) = p+frac{1}{3}$, we have for $p geq frac{4}{3}$ that $x=y=z=frac{1}{sqrt 3}$ maximizes $f$.






    I have attempted the method of Lagrange multipliers, which led me to three cubic equations:




    I set out to maximize
    $4(x^3+y^3+z^3)+p(xy+xz+yz)+lambda(x^2+y^2+z^2-1)$.



    This yielded
    $$4x^3+2xlambda + p(y+z)=0$$
    $$4y^3+2ylambda + p(x+z)=0$$
    $$4z^3+2zlambda + p(y+x)=0$$
    But now I feel a little at a loss on how to solve this system of equations.
    One can verify that $x=y=z=frac{1}{sqrt{3}}$ solves these equations at any p with the right choice of lamda. I'm unsure about pairing Hessian techniques to verify the local nature of the critical point when using lagrange multipliers.









    I have also tried incorporating the constraint by writing $z=cos(theta)$, $x=sin(theta)cos(phi)$ and $y=sin(theta)sin(phi)$.




    This yields
    $$f(theta, phi, p) = cos(theta)^4+(sin(theta)cos(phi))^4+(sin(theta)sin(phi))^4+p(cos(theta)sin(theta)sin(phi)+cos(theta)sin(theta)cos(phi)+(sin(theta))^2sin(phi)cos(phi) ) $$
    which I am unsure how to maximize. However, with motivation from Michael Behrend, I've found that studying $phi=frac{pi}{4}$ shows interesting critical point behavior slightly above p=1 where three critical points undergo some sort of bifurcation leaving just a single critical point at the eternal $x=y=z=frac{1}{3}$ .











    share|cite|improve this question











    $endgroup$















      3












      3








      3


      1



      $begingroup$


      What values of $x,y,z$ maximize $f(x,y,z,p) = x^4+y^4+z^4 + p(xy+xz+yz)$ with constant $p geq 0$ with the constraint $x^2+y^2+z^2=1$?



      Some preliminary studies in Mathematica showed that the behavior of the solution changes around $p=1$. For $p>1$, the problem seems solved by $x=y=z=frac{1}{sqrt 3}$. For $0 leq p leq 1$ it appears the solution (up to permutations of x,y,z) looks like x > y = z. These solutions appear to favor one or the other of the two terms in $f$, and I find the $p>1$ behavior striking and would appreciate a proof that $x=y=z=frac{1}{sqrt 3}$ indeed maximizes.





      UPDATE




      For at least $p>4/3$, one can show that $x=y=z=frac{1}{sqrt 3}$ maximizes f.



      The proof for $1<p<4/3$ evades me currently (the left bound, as mentioned in comments, may be slightly higher than 1).



      I'll assume without loss of generality that $x,y,z$ are all positive. Note that $f(frac{1}{sqrt 3},frac{1}{sqrt 3},frac{1}{sqrt 3}, p) = p+frac{1}{3} $, so if we show that for some $p$ we enjoy $f(x,y,z,p) leq p+frac{1}{3}$, we've shown $x=y=z=frac{1}{sqrt 3}$ maximizes $f$.



      Now consider that via the constraint, $$f(x,y,z,p) = 1-2(x^2y^2+x^2z^2+y^2z^2) + p(xy+xz+yz)$$
      The root mean square compared to arithmetic mean inequality tells us that for nonnegative $a,b,c$ we enjoy $sqrt{frac{a^2+b^2+c^2}{3}} geq frac{a+b+c}{3},$ so $-(a^2+b^2+c^2) leq -frac{(a+b+c)^2}{3}$. Then $$f(x,y,z,p) leq 1 + p(xy+xz+yz) -frac{2}{3}(xy+xz+yz)^2.$$
      Now, let's call $(xy+xz+yz)=q$. Note that it is easy to prove with our constraints that $0leq q leq 1$.
      Then we can bound the maximum of $f(x,y,z,p)$ by finding the maximum of $g(q) = 1+pq-frac{2}{3}q^2$ on $0leq q leq 1$.
      The derivative $g'(q) = p-frac{4}{3}q$, so $g(q)$ increases up to $q=frac{3}{4}p$. Then on $0leq q leq 1$, we have $q_{max} = min(frac{3}{4}p, 1)$.
      Thus for $p geq frac{4}{3}$ we have $q_{max}=1$, so $$ f(x,y,z,p) leq 1+p*1-frac{2}{3}*1^2 = p + 1/3$$
      Thus since $f(frac{1}{sqrt 3},frac{1}{sqrt 3},frac{1}{sqrt 3}, p) = p+frac{1}{3}$, we have for $p geq frac{4}{3}$ that $x=y=z=frac{1}{sqrt 3}$ maximizes $f$.






      I have attempted the method of Lagrange multipliers, which led me to three cubic equations:




      I set out to maximize
      $4(x^3+y^3+z^3)+p(xy+xz+yz)+lambda(x^2+y^2+z^2-1)$.



      This yielded
      $$4x^3+2xlambda + p(y+z)=0$$
      $$4y^3+2ylambda + p(x+z)=0$$
      $$4z^3+2zlambda + p(y+x)=0$$
      But now I feel a little at a loss on how to solve this system of equations.
      One can verify that $x=y=z=frac{1}{sqrt{3}}$ solves these equations at any p with the right choice of lamda. I'm unsure about pairing Hessian techniques to verify the local nature of the critical point when using lagrange multipliers.









      I have also tried incorporating the constraint by writing $z=cos(theta)$, $x=sin(theta)cos(phi)$ and $y=sin(theta)sin(phi)$.




      This yields
      $$f(theta, phi, p) = cos(theta)^4+(sin(theta)cos(phi))^4+(sin(theta)sin(phi))^4+p(cos(theta)sin(theta)sin(phi)+cos(theta)sin(theta)cos(phi)+(sin(theta))^2sin(phi)cos(phi) ) $$
      which I am unsure how to maximize. However, with motivation from Michael Behrend, I've found that studying $phi=frac{pi}{4}$ shows interesting critical point behavior slightly above p=1 where three critical points undergo some sort of bifurcation leaving just a single critical point at the eternal $x=y=z=frac{1}{3}$ .











      share|cite|improve this question











      $endgroup$




      What values of $x,y,z$ maximize $f(x,y,z,p) = x^4+y^4+z^4 + p(xy+xz+yz)$ with constant $p geq 0$ with the constraint $x^2+y^2+z^2=1$?



      Some preliminary studies in Mathematica showed that the behavior of the solution changes around $p=1$. For $p>1$, the problem seems solved by $x=y=z=frac{1}{sqrt 3}$. For $0 leq p leq 1$ it appears the solution (up to permutations of x,y,z) looks like x > y = z. These solutions appear to favor one or the other of the two terms in $f$, and I find the $p>1$ behavior striking and would appreciate a proof that $x=y=z=frac{1}{sqrt 3}$ indeed maximizes.





      UPDATE




      For at least $p>4/3$, one can show that $x=y=z=frac{1}{sqrt 3}$ maximizes f.



      The proof for $1<p<4/3$ evades me currently (the left bound, as mentioned in comments, may be slightly higher than 1).



      I'll assume without loss of generality that $x,y,z$ are all positive. Note that $f(frac{1}{sqrt 3},frac{1}{sqrt 3},frac{1}{sqrt 3}, p) = p+frac{1}{3} $, so if we show that for some $p$ we enjoy $f(x,y,z,p) leq p+frac{1}{3}$, we've shown $x=y=z=frac{1}{sqrt 3}$ maximizes $f$.



      Now consider that via the constraint, $$f(x,y,z,p) = 1-2(x^2y^2+x^2z^2+y^2z^2) + p(xy+xz+yz)$$
      The root mean square compared to arithmetic mean inequality tells us that for nonnegative $a,b,c$ we enjoy $sqrt{frac{a^2+b^2+c^2}{3}} geq frac{a+b+c}{3},$ so $-(a^2+b^2+c^2) leq -frac{(a+b+c)^2}{3}$. Then $$f(x,y,z,p) leq 1 + p(xy+xz+yz) -frac{2}{3}(xy+xz+yz)^2.$$
      Now, let's call $(xy+xz+yz)=q$. Note that it is easy to prove with our constraints that $0leq q leq 1$.
      Then we can bound the maximum of $f(x,y,z,p)$ by finding the maximum of $g(q) = 1+pq-frac{2}{3}q^2$ on $0leq q leq 1$.
      The derivative $g'(q) = p-frac{4}{3}q$, so $g(q)$ increases up to $q=frac{3}{4}p$. Then on $0leq q leq 1$, we have $q_{max} = min(frac{3}{4}p, 1)$.
      Thus for $p geq frac{4}{3}$ we have $q_{max}=1$, so $$ f(x,y,z,p) leq 1+p*1-frac{2}{3}*1^2 = p + 1/3$$
      Thus since $f(frac{1}{sqrt 3},frac{1}{sqrt 3},frac{1}{sqrt 3}, p) = p+frac{1}{3}$, we have for $p geq frac{4}{3}$ that $x=y=z=frac{1}{sqrt 3}$ maximizes $f$.






      I have attempted the method of Lagrange multipliers, which led me to three cubic equations:




      I set out to maximize
      $4(x^3+y^3+z^3)+p(xy+xz+yz)+lambda(x^2+y^2+z^2-1)$.



      This yielded
      $$4x^3+2xlambda + p(y+z)=0$$
      $$4y^3+2ylambda + p(x+z)=0$$
      $$4z^3+2zlambda + p(y+x)=0$$
      But now I feel a little at a loss on how to solve this system of equations.
      One can verify that $x=y=z=frac{1}{sqrt{3}}$ solves these equations at any p with the right choice of lamda. I'm unsure about pairing Hessian techniques to verify the local nature of the critical point when using lagrange multipliers.









      I have also tried incorporating the constraint by writing $z=cos(theta)$, $x=sin(theta)cos(phi)$ and $y=sin(theta)sin(phi)$.




      This yields
      $$f(theta, phi, p) = cos(theta)^4+(sin(theta)cos(phi))^4+(sin(theta)sin(phi))^4+p(cos(theta)sin(theta)sin(phi)+cos(theta)sin(theta)cos(phi)+(sin(theta))^2sin(phi)cos(phi) ) $$
      which I am unsure how to maximize. However, with motivation from Michael Behrend, I've found that studying $phi=frac{pi}{4}$ shows interesting critical point behavior slightly above p=1 where three critical points undergo some sort of bifurcation leaving just a single critical point at the eternal $x=y=z=frac{1}{3}$ .








      optimization lagrange-multiplier spheres






      share|cite|improve this question















      share|cite|improve this question













      share|cite|improve this question




      share|cite|improve this question








      edited Jan 21 at 21:51







      user196574

















      asked Jan 20 at 6:13









      user196574user196574

      1657




      1657






















          1 Answer
          1






          active

          oldest

          votes


















          1












          $begingroup$

          Lagrange function is
          $$L=x^4+y^4+z^4+p(xy+xz+yz)+lambda(x^2+y^2+z^2-1).$$
          Solving system
          $$L'_x=0,;L'_y=0,;L'_z=0,;L'_x=0,;L'_lambda=0$$
          we get that if $pge1$ global maximum is
          $$f_{max}=p+frac13$$
          Maximum points is:



          if $p>1$ then $[x=frac{1}{sqrt{3}},y=frac{1}{sqrt{3}},z=frac{1}{sqrt{3}}]$ or
          $[x=-frac{1}{sqrt{3}},y=-frac{1}{sqrt{3}},z=-frac{1}{sqrt{3}}]$



          if $p=1$ then $[x=frac{sqrt{6}}{3},y=frac{1}{sqrt{6}},z=frac{1}{sqrt{6}}],[x=-frac{sqrt{6}}{3},y=-frac{1}{sqrt{6}},z=-frac{1}{sqrt{6}}],[x=frac{1}{sqrt{6}},y=frac{sqrt{6}}{3},z=frac{1}{sqrt{6}}],[x=frac{1}{sqrt{6}},y=frac{1}{sqrt{6}},z=frac{sqrt{6}}{3}],[x=-frac{1}{sqrt{6}},y=-frac{sqrt{6}}{3},z=-frac{1}{sqrt{6}}],[x=-frac{1}{sqrt{6}},y=-frac{1}{sqrt{6}},z=-frac{sqrt{6}}{3}],[x=frac{1}{sqrt{3}},y=frac{1}{sqrt{3}},z=frac{1}{sqrt{3}}],[x=-frac{1}{sqrt{3}},y=-frac{1}{sqrt{3}},z=-frac{1}{sqrt{3}}]$



          If $p=0$ then
          $$f_{max}=1,$$
          $[x=pm1,y=0,z=0],[x=0,y=pm1,z=0],[x=0,y=0,z=pm1]$.



          If $0<p<1$ we can't exact solve system.






          share|cite|improve this answer









          $endgroup$













          • $begingroup$
            I appreciate the answer. Could you explain how you solved the system of equations for p>1, and why you think that we cannot exactly solve the system for p between 0 and 1?
            $endgroup$
            – user196574
            Jan 20 at 9:33






          • 1




            $begingroup$
            After trying to prove that the crossover is at p = 1, I'm coming to the conclusion that this isn't true. The crossover seems to be somewhere between 1.01 and 1.02. Hope to post more later.
            $endgroup$
            – Michael Behrend
            Jan 21 at 18:32










          • $begingroup$
            @MichaelBehrend I've been working a bit more with the trigonometric functions under the assumption that phi is pi/4. In that special case, I see interesting behavior with the critical points in theta. No matter the value of p, the equivalent to the x=y=z critical point remains, though whether it's a global maximum or even a local maximum changes with p. In particular, it seems just slightly above 1 that the three critical points become just 1 critical point. If you can find the value of p where that occurs, a 2nd derivative test should show that x=y=z maximizes above that. Thanks for the help!
            $endgroup$
            – user196574
            Jan 21 at 20:55












          • $begingroup$
            The crossover point is the larger real root of $$9x^4 + 10x^3 + 12x^2 - 24x - 8$$ which according to akiti.ca/PolyRootRe.html is 1.0148293664019847 (to 16 d.p.) This was found by a rather lengthy method and checked empirically by a computer program. If I can find a better proof I'll post it as an answer.
            $endgroup$
            – Michael Behrend
            Jan 22 at 15:00










          • $begingroup$
            Sorry, my previous comments were wrong. There is a local maximum at the "pole" $(1/sqrt3, 1/sqrt3, 1/sqrt3)$, and for $p < p_0 = 1.014829ldots$ there are local maxima at three other points (all three with the same function value). For $p < 1$ these three maxima are greater than the maximum at the pole, for p = 1 all four maxima are equal, and for $p > 1$ the maximum at the pole becomes greater. So the crossover is $p = 1$ as originally stated. The change at $p = p_0$ is that the three other local maxima disappear.
            $endgroup$
            – Michael Behrend
            Jan 23 at 9:43











          Your Answer





          StackExchange.ifUsing("editor", function () {
          return StackExchange.using("mathjaxEditing", function () {
          StackExchange.MarkdownEditor.creationCallbacks.add(function (editor, postfix) {
          StackExchange.mathjaxEditing.prepareWmdForMathJax(editor, postfix, [["$", "$"], ["\\(","\\)"]]);
          });
          });
          }, "mathjax-editing");

          StackExchange.ready(function() {
          var channelOptions = {
          tags: "".split(" "),
          id: "69"
          };
          initTagRenderer("".split(" "), "".split(" "), channelOptions);

          StackExchange.using("externalEditor", function() {
          // Have to fire editor after snippets, if snippets enabled
          if (StackExchange.settings.snippets.snippetsEnabled) {
          StackExchange.using("snippets", function() {
          createEditor();
          });
          }
          else {
          createEditor();
          }
          });

          function createEditor() {
          StackExchange.prepareEditor({
          heartbeatType: 'answer',
          autoActivateHeartbeat: false,
          convertImagesToLinks: true,
          noModals: true,
          showLowRepImageUploadWarning: true,
          reputationToPostImages: 10,
          bindNavPrevention: true,
          postfix: "",
          imageUploader: {
          brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
          contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
          allowUrls: true
          },
          noCode: true, onDemand: true,
          discardSelector: ".discard-answer"
          ,immediatelyShowMarkdownHelp:true
          });


          }
          });














          draft saved

          draft discarded


















          StackExchange.ready(
          function () {
          StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f3080248%2fmaximizing-x4y4z4-pxyxzyz-on-a-sphere%23new-answer', 'question_page');
          }
          );

          Post as a guest















          Required, but never shown

























          1 Answer
          1






          active

          oldest

          votes








          1 Answer
          1






          active

          oldest

          votes









          active

          oldest

          votes






          active

          oldest

          votes









          1












          $begingroup$

          Lagrange function is
          $$L=x^4+y^4+z^4+p(xy+xz+yz)+lambda(x^2+y^2+z^2-1).$$
          Solving system
          $$L'_x=0,;L'_y=0,;L'_z=0,;L'_x=0,;L'_lambda=0$$
          we get that if $pge1$ global maximum is
          $$f_{max}=p+frac13$$
          Maximum points is:



          if $p>1$ then $[x=frac{1}{sqrt{3}},y=frac{1}{sqrt{3}},z=frac{1}{sqrt{3}}]$ or
          $[x=-frac{1}{sqrt{3}},y=-frac{1}{sqrt{3}},z=-frac{1}{sqrt{3}}]$



          if $p=1$ then $[x=frac{sqrt{6}}{3},y=frac{1}{sqrt{6}},z=frac{1}{sqrt{6}}],[x=-frac{sqrt{6}}{3},y=-frac{1}{sqrt{6}},z=-frac{1}{sqrt{6}}],[x=frac{1}{sqrt{6}},y=frac{sqrt{6}}{3},z=frac{1}{sqrt{6}}],[x=frac{1}{sqrt{6}},y=frac{1}{sqrt{6}},z=frac{sqrt{6}}{3}],[x=-frac{1}{sqrt{6}},y=-frac{sqrt{6}}{3},z=-frac{1}{sqrt{6}}],[x=-frac{1}{sqrt{6}},y=-frac{1}{sqrt{6}},z=-frac{sqrt{6}}{3}],[x=frac{1}{sqrt{3}},y=frac{1}{sqrt{3}},z=frac{1}{sqrt{3}}],[x=-frac{1}{sqrt{3}},y=-frac{1}{sqrt{3}},z=-frac{1}{sqrt{3}}]$



          If $p=0$ then
          $$f_{max}=1,$$
          $[x=pm1,y=0,z=0],[x=0,y=pm1,z=0],[x=0,y=0,z=pm1]$.



          If $0<p<1$ we can't exact solve system.






          share|cite|improve this answer









          $endgroup$













          • $begingroup$
            I appreciate the answer. Could you explain how you solved the system of equations for p>1, and why you think that we cannot exactly solve the system for p between 0 and 1?
            $endgroup$
            – user196574
            Jan 20 at 9:33






          • 1




            $begingroup$
            After trying to prove that the crossover is at p = 1, I'm coming to the conclusion that this isn't true. The crossover seems to be somewhere between 1.01 and 1.02. Hope to post more later.
            $endgroup$
            – Michael Behrend
            Jan 21 at 18:32










          • $begingroup$
            @MichaelBehrend I've been working a bit more with the trigonometric functions under the assumption that phi is pi/4. In that special case, I see interesting behavior with the critical points in theta. No matter the value of p, the equivalent to the x=y=z critical point remains, though whether it's a global maximum or even a local maximum changes with p. In particular, it seems just slightly above 1 that the three critical points become just 1 critical point. If you can find the value of p where that occurs, a 2nd derivative test should show that x=y=z maximizes above that. Thanks for the help!
            $endgroup$
            – user196574
            Jan 21 at 20:55












          • $begingroup$
            The crossover point is the larger real root of $$9x^4 + 10x^3 + 12x^2 - 24x - 8$$ which according to akiti.ca/PolyRootRe.html is 1.0148293664019847 (to 16 d.p.) This was found by a rather lengthy method and checked empirically by a computer program. If I can find a better proof I'll post it as an answer.
            $endgroup$
            – Michael Behrend
            Jan 22 at 15:00










          • $begingroup$
            Sorry, my previous comments were wrong. There is a local maximum at the "pole" $(1/sqrt3, 1/sqrt3, 1/sqrt3)$, and for $p < p_0 = 1.014829ldots$ there are local maxima at three other points (all three with the same function value). For $p < 1$ these three maxima are greater than the maximum at the pole, for p = 1 all four maxima are equal, and for $p > 1$ the maximum at the pole becomes greater. So the crossover is $p = 1$ as originally stated. The change at $p = p_0$ is that the three other local maxima disappear.
            $endgroup$
            – Michael Behrend
            Jan 23 at 9:43
















          1












          $begingroup$

          Lagrange function is
          $$L=x^4+y^4+z^4+p(xy+xz+yz)+lambda(x^2+y^2+z^2-1).$$
          Solving system
          $$L'_x=0,;L'_y=0,;L'_z=0,;L'_x=0,;L'_lambda=0$$
          we get that if $pge1$ global maximum is
          $$f_{max}=p+frac13$$
          Maximum points is:



          if $p>1$ then $[x=frac{1}{sqrt{3}},y=frac{1}{sqrt{3}},z=frac{1}{sqrt{3}}]$ or
          $[x=-frac{1}{sqrt{3}},y=-frac{1}{sqrt{3}},z=-frac{1}{sqrt{3}}]$



          if $p=1$ then $[x=frac{sqrt{6}}{3},y=frac{1}{sqrt{6}},z=frac{1}{sqrt{6}}],[x=-frac{sqrt{6}}{3},y=-frac{1}{sqrt{6}},z=-frac{1}{sqrt{6}}],[x=frac{1}{sqrt{6}},y=frac{sqrt{6}}{3},z=frac{1}{sqrt{6}}],[x=frac{1}{sqrt{6}},y=frac{1}{sqrt{6}},z=frac{sqrt{6}}{3}],[x=-frac{1}{sqrt{6}},y=-frac{sqrt{6}}{3},z=-frac{1}{sqrt{6}}],[x=-frac{1}{sqrt{6}},y=-frac{1}{sqrt{6}},z=-frac{sqrt{6}}{3}],[x=frac{1}{sqrt{3}},y=frac{1}{sqrt{3}},z=frac{1}{sqrt{3}}],[x=-frac{1}{sqrt{3}},y=-frac{1}{sqrt{3}},z=-frac{1}{sqrt{3}}]$



          If $p=0$ then
          $$f_{max}=1,$$
          $[x=pm1,y=0,z=0],[x=0,y=pm1,z=0],[x=0,y=0,z=pm1]$.



          If $0<p<1$ we can't exact solve system.






          share|cite|improve this answer









          $endgroup$













          • $begingroup$
            I appreciate the answer. Could you explain how you solved the system of equations for p>1, and why you think that we cannot exactly solve the system for p between 0 and 1?
            $endgroup$
            – user196574
            Jan 20 at 9:33






          • 1




            $begingroup$
            After trying to prove that the crossover is at p = 1, I'm coming to the conclusion that this isn't true. The crossover seems to be somewhere between 1.01 and 1.02. Hope to post more later.
            $endgroup$
            – Michael Behrend
            Jan 21 at 18:32










          • $begingroup$
            @MichaelBehrend I've been working a bit more with the trigonometric functions under the assumption that phi is pi/4. In that special case, I see interesting behavior with the critical points in theta. No matter the value of p, the equivalent to the x=y=z critical point remains, though whether it's a global maximum or even a local maximum changes with p. In particular, it seems just slightly above 1 that the three critical points become just 1 critical point. If you can find the value of p where that occurs, a 2nd derivative test should show that x=y=z maximizes above that. Thanks for the help!
            $endgroup$
            – user196574
            Jan 21 at 20:55












          • $begingroup$
            The crossover point is the larger real root of $$9x^4 + 10x^3 + 12x^2 - 24x - 8$$ which according to akiti.ca/PolyRootRe.html is 1.0148293664019847 (to 16 d.p.) This was found by a rather lengthy method and checked empirically by a computer program. If I can find a better proof I'll post it as an answer.
            $endgroup$
            – Michael Behrend
            Jan 22 at 15:00










          • $begingroup$
            Sorry, my previous comments were wrong. There is a local maximum at the "pole" $(1/sqrt3, 1/sqrt3, 1/sqrt3)$, and for $p < p_0 = 1.014829ldots$ there are local maxima at three other points (all three with the same function value). For $p < 1$ these three maxima are greater than the maximum at the pole, for p = 1 all four maxima are equal, and for $p > 1$ the maximum at the pole becomes greater. So the crossover is $p = 1$ as originally stated. The change at $p = p_0$ is that the three other local maxima disappear.
            $endgroup$
            – Michael Behrend
            Jan 23 at 9:43














          1












          1








          1





          $begingroup$

          Lagrange function is
          $$L=x^4+y^4+z^4+p(xy+xz+yz)+lambda(x^2+y^2+z^2-1).$$
          Solving system
          $$L'_x=0,;L'_y=0,;L'_z=0,;L'_x=0,;L'_lambda=0$$
          we get that if $pge1$ global maximum is
          $$f_{max}=p+frac13$$
          Maximum points is:



          if $p>1$ then $[x=frac{1}{sqrt{3}},y=frac{1}{sqrt{3}},z=frac{1}{sqrt{3}}]$ or
          $[x=-frac{1}{sqrt{3}},y=-frac{1}{sqrt{3}},z=-frac{1}{sqrt{3}}]$



          if $p=1$ then $[x=frac{sqrt{6}}{3},y=frac{1}{sqrt{6}},z=frac{1}{sqrt{6}}],[x=-frac{sqrt{6}}{3},y=-frac{1}{sqrt{6}},z=-frac{1}{sqrt{6}}],[x=frac{1}{sqrt{6}},y=frac{sqrt{6}}{3},z=frac{1}{sqrt{6}}],[x=frac{1}{sqrt{6}},y=frac{1}{sqrt{6}},z=frac{sqrt{6}}{3}],[x=-frac{1}{sqrt{6}},y=-frac{sqrt{6}}{3},z=-frac{1}{sqrt{6}}],[x=-frac{1}{sqrt{6}},y=-frac{1}{sqrt{6}},z=-frac{sqrt{6}}{3}],[x=frac{1}{sqrt{3}},y=frac{1}{sqrt{3}},z=frac{1}{sqrt{3}}],[x=-frac{1}{sqrt{3}},y=-frac{1}{sqrt{3}},z=-frac{1}{sqrt{3}}]$



          If $p=0$ then
          $$f_{max}=1,$$
          $[x=pm1,y=0,z=0],[x=0,y=pm1,z=0],[x=0,y=0,z=pm1]$.



          If $0<p<1$ we can't exact solve system.






          share|cite|improve this answer









          $endgroup$



          Lagrange function is
          $$L=x^4+y^4+z^4+p(xy+xz+yz)+lambda(x^2+y^2+z^2-1).$$
          Solving system
          $$L'_x=0,;L'_y=0,;L'_z=0,;L'_x=0,;L'_lambda=0$$
          we get that if $pge1$ global maximum is
          $$f_{max}=p+frac13$$
          Maximum points is:



          if $p>1$ then $[x=frac{1}{sqrt{3}},y=frac{1}{sqrt{3}},z=frac{1}{sqrt{3}}]$ or
          $[x=-frac{1}{sqrt{3}},y=-frac{1}{sqrt{3}},z=-frac{1}{sqrt{3}}]$



          if $p=1$ then $[x=frac{sqrt{6}}{3},y=frac{1}{sqrt{6}},z=frac{1}{sqrt{6}}],[x=-frac{sqrt{6}}{3},y=-frac{1}{sqrt{6}},z=-frac{1}{sqrt{6}}],[x=frac{1}{sqrt{6}},y=frac{sqrt{6}}{3},z=frac{1}{sqrt{6}}],[x=frac{1}{sqrt{6}},y=frac{1}{sqrt{6}},z=frac{sqrt{6}}{3}],[x=-frac{1}{sqrt{6}},y=-frac{sqrt{6}}{3},z=-frac{1}{sqrt{6}}],[x=-frac{1}{sqrt{6}},y=-frac{1}{sqrt{6}},z=-frac{sqrt{6}}{3}],[x=frac{1}{sqrt{3}},y=frac{1}{sqrt{3}},z=frac{1}{sqrt{3}}],[x=-frac{1}{sqrt{3}},y=-frac{1}{sqrt{3}},z=-frac{1}{sqrt{3}}]$



          If $p=0$ then
          $$f_{max}=1,$$
          $[x=pm1,y=0,z=0],[x=0,y=pm1,z=0],[x=0,y=0,z=pm1]$.



          If $0<p<1$ we can't exact solve system.







          share|cite|improve this answer












          share|cite|improve this answer



          share|cite|improve this answer










          answered Jan 20 at 9:08









          Aleksas DomarkasAleksas Domarkas

          1,37216




          1,37216












          • $begingroup$
            I appreciate the answer. Could you explain how you solved the system of equations for p>1, and why you think that we cannot exactly solve the system for p between 0 and 1?
            $endgroup$
            – user196574
            Jan 20 at 9:33






          • 1




            $begingroup$
            After trying to prove that the crossover is at p = 1, I'm coming to the conclusion that this isn't true. The crossover seems to be somewhere between 1.01 and 1.02. Hope to post more later.
            $endgroup$
            – Michael Behrend
            Jan 21 at 18:32










          • $begingroup$
            @MichaelBehrend I've been working a bit more with the trigonometric functions under the assumption that phi is pi/4. In that special case, I see interesting behavior with the critical points in theta. No matter the value of p, the equivalent to the x=y=z critical point remains, though whether it's a global maximum or even a local maximum changes with p. In particular, it seems just slightly above 1 that the three critical points become just 1 critical point. If you can find the value of p where that occurs, a 2nd derivative test should show that x=y=z maximizes above that. Thanks for the help!
            $endgroup$
            – user196574
            Jan 21 at 20:55












          • $begingroup$
            The crossover point is the larger real root of $$9x^4 + 10x^3 + 12x^2 - 24x - 8$$ which according to akiti.ca/PolyRootRe.html is 1.0148293664019847 (to 16 d.p.) This was found by a rather lengthy method and checked empirically by a computer program. If I can find a better proof I'll post it as an answer.
            $endgroup$
            – Michael Behrend
            Jan 22 at 15:00










          • $begingroup$
            Sorry, my previous comments were wrong. There is a local maximum at the "pole" $(1/sqrt3, 1/sqrt3, 1/sqrt3)$, and for $p < p_0 = 1.014829ldots$ there are local maxima at three other points (all three with the same function value). For $p < 1$ these three maxima are greater than the maximum at the pole, for p = 1 all four maxima are equal, and for $p > 1$ the maximum at the pole becomes greater. So the crossover is $p = 1$ as originally stated. The change at $p = p_0$ is that the three other local maxima disappear.
            $endgroup$
            – Michael Behrend
            Jan 23 at 9:43


















          • $begingroup$
            I appreciate the answer. Could you explain how you solved the system of equations for p>1, and why you think that we cannot exactly solve the system for p between 0 and 1?
            $endgroup$
            – user196574
            Jan 20 at 9:33






          • 1




            $begingroup$
            After trying to prove that the crossover is at p = 1, I'm coming to the conclusion that this isn't true. The crossover seems to be somewhere between 1.01 and 1.02. Hope to post more later.
            $endgroup$
            – Michael Behrend
            Jan 21 at 18:32










          • $begingroup$
            @MichaelBehrend I've been working a bit more with the trigonometric functions under the assumption that phi is pi/4. In that special case, I see interesting behavior with the critical points in theta. No matter the value of p, the equivalent to the x=y=z critical point remains, though whether it's a global maximum or even a local maximum changes with p. In particular, it seems just slightly above 1 that the three critical points become just 1 critical point. If you can find the value of p where that occurs, a 2nd derivative test should show that x=y=z maximizes above that. Thanks for the help!
            $endgroup$
            – user196574
            Jan 21 at 20:55












          • $begingroup$
            The crossover point is the larger real root of $$9x^4 + 10x^3 + 12x^2 - 24x - 8$$ which according to akiti.ca/PolyRootRe.html is 1.0148293664019847 (to 16 d.p.) This was found by a rather lengthy method and checked empirically by a computer program. If I can find a better proof I'll post it as an answer.
            $endgroup$
            – Michael Behrend
            Jan 22 at 15:00










          • $begingroup$
            Sorry, my previous comments were wrong. There is a local maximum at the "pole" $(1/sqrt3, 1/sqrt3, 1/sqrt3)$, and for $p < p_0 = 1.014829ldots$ there are local maxima at three other points (all three with the same function value). For $p < 1$ these three maxima are greater than the maximum at the pole, for p = 1 all four maxima are equal, and for $p > 1$ the maximum at the pole becomes greater. So the crossover is $p = 1$ as originally stated. The change at $p = p_0$ is that the three other local maxima disappear.
            $endgroup$
            – Michael Behrend
            Jan 23 at 9:43
















          $begingroup$
          I appreciate the answer. Could you explain how you solved the system of equations for p>1, and why you think that we cannot exactly solve the system for p between 0 and 1?
          $endgroup$
          – user196574
          Jan 20 at 9:33




          $begingroup$
          I appreciate the answer. Could you explain how you solved the system of equations for p>1, and why you think that we cannot exactly solve the system for p between 0 and 1?
          $endgroup$
          – user196574
          Jan 20 at 9:33




          1




          1




          $begingroup$
          After trying to prove that the crossover is at p = 1, I'm coming to the conclusion that this isn't true. The crossover seems to be somewhere between 1.01 and 1.02. Hope to post more later.
          $endgroup$
          – Michael Behrend
          Jan 21 at 18:32




          $begingroup$
          After trying to prove that the crossover is at p = 1, I'm coming to the conclusion that this isn't true. The crossover seems to be somewhere between 1.01 and 1.02. Hope to post more later.
          $endgroup$
          – Michael Behrend
          Jan 21 at 18:32












          $begingroup$
          @MichaelBehrend I've been working a bit more with the trigonometric functions under the assumption that phi is pi/4. In that special case, I see interesting behavior with the critical points in theta. No matter the value of p, the equivalent to the x=y=z critical point remains, though whether it's a global maximum or even a local maximum changes with p. In particular, it seems just slightly above 1 that the three critical points become just 1 critical point. If you can find the value of p where that occurs, a 2nd derivative test should show that x=y=z maximizes above that. Thanks for the help!
          $endgroup$
          – user196574
          Jan 21 at 20:55






          $begingroup$
          @MichaelBehrend I've been working a bit more with the trigonometric functions under the assumption that phi is pi/4. In that special case, I see interesting behavior with the critical points in theta. No matter the value of p, the equivalent to the x=y=z critical point remains, though whether it's a global maximum or even a local maximum changes with p. In particular, it seems just slightly above 1 that the three critical points become just 1 critical point. If you can find the value of p where that occurs, a 2nd derivative test should show that x=y=z maximizes above that. Thanks for the help!
          $endgroup$
          – user196574
          Jan 21 at 20:55














          $begingroup$
          The crossover point is the larger real root of $$9x^4 + 10x^3 + 12x^2 - 24x - 8$$ which according to akiti.ca/PolyRootRe.html is 1.0148293664019847 (to 16 d.p.) This was found by a rather lengthy method and checked empirically by a computer program. If I can find a better proof I'll post it as an answer.
          $endgroup$
          – Michael Behrend
          Jan 22 at 15:00




          $begingroup$
          The crossover point is the larger real root of $$9x^4 + 10x^3 + 12x^2 - 24x - 8$$ which according to akiti.ca/PolyRootRe.html is 1.0148293664019847 (to 16 d.p.) This was found by a rather lengthy method and checked empirically by a computer program. If I can find a better proof I'll post it as an answer.
          $endgroup$
          – Michael Behrend
          Jan 22 at 15:00












          $begingroup$
          Sorry, my previous comments were wrong. There is a local maximum at the "pole" $(1/sqrt3, 1/sqrt3, 1/sqrt3)$, and for $p < p_0 = 1.014829ldots$ there are local maxima at three other points (all three with the same function value). For $p < 1$ these three maxima are greater than the maximum at the pole, for p = 1 all four maxima are equal, and for $p > 1$ the maximum at the pole becomes greater. So the crossover is $p = 1$ as originally stated. The change at $p = p_0$ is that the three other local maxima disappear.
          $endgroup$
          – Michael Behrend
          Jan 23 at 9:43




          $begingroup$
          Sorry, my previous comments were wrong. There is a local maximum at the "pole" $(1/sqrt3, 1/sqrt3, 1/sqrt3)$, and for $p < p_0 = 1.014829ldots$ there are local maxima at three other points (all three with the same function value). For $p < 1$ these three maxima are greater than the maximum at the pole, for p = 1 all four maxima are equal, and for $p > 1$ the maximum at the pole becomes greater. So the crossover is $p = 1$ as originally stated. The change at $p = p_0$ is that the three other local maxima disappear.
          $endgroup$
          – Michael Behrend
          Jan 23 at 9:43


















          draft saved

          draft discarded




















































          Thanks for contributing an answer to Mathematics Stack Exchange!


          • Please be sure to answer the question. Provide details and share your research!

          But avoid



          • Asking for help, clarification, or responding to other answers.

          • Making statements based on opinion; back them up with references or personal experience.


          Use MathJax to format equations. MathJax reference.


          To learn more, see our tips on writing great answers.




          draft saved


          draft discarded














          StackExchange.ready(
          function () {
          StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f3080248%2fmaximizing-x4y4z4-pxyxzyz-on-a-sphere%23new-answer', 'question_page');
          }
          );

          Post as a guest















          Required, but never shown





















































          Required, but never shown














          Required, but never shown












          Required, but never shown







          Required, but never shown

































          Required, but never shown














          Required, but never shown












          Required, but never shown







          Required, but never shown







          Popular posts from this blog

          android studio warns about leanback feature tag usage required on manifest while using Unity exported app?

          SQL update select statement

          'app-layout' is not a known element: how to share Component with different Modules