Comments

What is wrong in my approach

every state of dp will store a pair

dp(i,0) will store that using arr[i]-k and what will be the (min,max) from index 0 to i

dp(i,1) will store that using arr[i]+k and what will be the (min,max) from index 0 to i

ans=min(dp(n-1)(0).second-dp(n-1)(0).first,dp(n-1)(1).second-dp(n-1)(1).first)

Transition in state

temp1=min(dp[i-1][1].first,arr[i]-k) //dp[i-1][1].first will store the minimum vale till i-1 index considering arr[i]-k

temp2=max(dp[i-1][1].second,arr[i]-k)

temp3=min(dp[i-1][0].first,arr[i]-k)

temp4=max(dp[i-1][0].second,arr[i]-k)

if(abs(temp1-temp2)>abs(temp3-temp4)) dp[i][0]={temp3,temp4}

else dp[i][0]={temp1,temp2}

//same way we can find dp[i][1]